CN110059963A - A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network - Google Patents
A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network Download PDFInfo
- Publication number
- CN110059963A CN110059963A CN201910320582.4A CN201910320582A CN110059963A CN 110059963 A CN110059963 A CN 110059963A CN 201910320582 A CN201910320582 A CN 201910320582A CN 110059963 A CN110059963 A CN 110059963A
- Authority
- CN
- China
- Prior art keywords
- probability
- fuzzy
- risk
- tunnel
- bayesian network
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
-
- 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
- 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
- G06Q50/08—Construction
Abstract
The invention discloses a kind of tunnel risk evaluating methods based on fuzzy polymorphism Bayesian network, the present invention proposes the investigation method that confidence index, weighted index and probability interval based on expert judgments combine, this method constructs tunnel risk accidents tree according to existing tunnel accident case, it obtains the probability of occurrence of the elementary event of tunnel risk accidents and each factor under the present art, and polymorphic Bayesian network is constructed by accident tree.Probability obtained by probability obtained by expert investigation and case accident is utilized into objective and subjective synthetic approach, obtains conditional probability, to propose based on polymorphic fuzzy Bayesian network conditional probability construction method and tunnel risk probability calculation method.This method can reduce the subjectivity of risk factors identification, and raising risk profile is accurate, realize information system management and the overall process dynamic evaluation of underground engineering construction, guarantee construction safety.
Description
Technical field
The present invention relates to a kind of tunnel safety methods of risk assessment, specially a kind of mould combined based on subjective and objective data
Polymorphic Bayesian network tunnel safety methods of risk assessment is pasted, Tunnel Engineering technical field is belonged to.
Background technique
The rock soil medium environment of the underground engineerings such as tunnel institute preservation is complicated, and design and construction theory is still incomplete, process of construction
In have very strong uncertainty, be a considerably complicated high risk system engineering, if control is ineffective, engineering risk accident
Generation will will cause great property loss, casualties, construction delay and bad social influence.A large amount of engineering practices show
Current risk management cannot meet information-based, Modern Construction needs again.Therefore, reinforce to underground engineerings such as tunnels
Research on Risk Management, accurate recognition risk factors, in time discovery risk sign, thus improve risk control level, reduce tunnel
The underground engineerings accident probability such as road is current China's underground engineering development & construction urgent problem to be solved.
Currently, the identification and risk assessment of the Research on Risk Management risk factor of underground engineering are more to concentrate on benefit
Overall, static risk class is carried out with expertise to divide, and can play reference function in having the governed engineering of experience,
But engineering this method less for experience is likely to result in error in judgement, it is therefore desirable to propose it is a kind of it is more objective, meet reality
The method on border.
China's underground engineering construction has had accumulated a large amount of engineering construction case, these cases are risks at present and in the future
How the valuable source of management, technological progress, sufficiently excavated these data and commented with more realistic Risk Identification with risk
Valence, up for research.And with the progress of big data theory and technology, data digging method has obtained extensively in all trades and professions
General application, therefore the data accumulated using the huge underground engineering construction in China not only can be the more reasonable duration
The effective method that provides is estimated in prediction, investment cost, more can be Risk Identification, the risk profile, risk assessment of underground engineering
The problems such as new thinking is provided.
For the subjectivity for reducing risk factors identification, raising risk profile is accurate, realizes the informationization of underground engineering construction
Management and overall process dynamic evaluation guarantee construction safety, propose a kind of dynamic evaluation based on measured data variation and correlation
Method has a very important significance the Risk Management of Tunnel Engineering.
Summary of the invention
The object of the invention is that providing a kind of mould combined based on subjective and objective data to solve the above-mentioned problems
Polymorphic Bayesian network tunnel safety methods of risk assessment is pasted, to reduce the subjectivity of tunnel risk factors identification, is mentioned simultaneously
The accuracy and validity of high risk forecast assessment result.
The present invention is achieved through the following technical solutions above-mentioned purpose, a kind of tunnel based on fuzzy polymorphism Bayesian network
Risk evaluating method, this method comprises:
Step 1: establishing the accident tree based on tunnel risk by the analysis to accident histories data；
Step 2: constructing fuzzy polymorphism bayesian network structure based on the accident tree；
Step 3: the Triangular Fuzzy Number of estimation root node probability of occurrence；
Step 4: calculating simultaneously merge node conditional probability；
Step 5: tunnel threat probability values are calculated according to the conditional probability.
Preferably, the accident tree based on risk factors is established in the step 1, is specifically included:
Step 1.1, by analyzing the reason of accident case, determine the elementary event for causing risk to occur and intermediate thing
Part；
Step 1.2 constructs accident tree according to the elementary event and the correlation of intermediate event；
Wherein, the risk include cave-in accident rank, supporting intensity, surrounding rock stability, design rationality, construction because
Element, disadvantageous geologic condition, degree of prospecting, all information degree, parameter choose reasonability, job specfication, construction quality, supporting
Timeliness, by design and construction degree, support material acceptable level, Grades of Surrounding Rock, special geologic condition, Atmospheric precipitation, draining and
Shi Xing, seepage action of ground water amount, geological conditions variability, construction disturbance.
Preferably, fuzzy polymorphism bayesian network structure is constructed based on the accident tree in the step 2, specifically included:
Tunnel risk class is divided into not by step 2.1 according to the different conditions of node each in Tunnel Engineering process of construction
Same stage, different phase are indicated with different fuzzy numbers；
Step 2.2, according to the accident tree determine father node and child node between relationship, by the network knot of accident tree
Structure is converted to the structure of Bayesian network.
Preferably, the Triangular Fuzzy Number that root node probability of occurrence is estimated in the step 3, specifically includes:
Step 3.1, weight coefficient and confidence index data of the building based on normal distribution；
When determining elementary event probability of occurrence, there are locating probability levels by estimation event and judges event in step 3.2
The probability of appearance, and probability levels are divided according to practical ranges；
Step 3.3 calculates different probability grade according to the weight coefficient, confidence index data and probability levels
Estimated value constructs the sample space of each root node different conditions according to the probabilistic estimated value, calculates all root nodes and exists
Triangular Fuzzy Number under different conditions.
Preferably, calculating and merge node conditional probability in the step 4, specifically include:
Step 4.1, building expert investigation group data, the conditional probability of all nodes is calculated according to investigation group's data；
Step 4.2 carries out dealing of abnormal data, rejecting abnormalities data using Chauvenet method, and carries out state normalization
Processing, so that the probability of all nodes at different conditions is obtained, thus the conditional probability of the whole nodes obtained.
Step 4.3 obtains the conditional probability for merging posterior nodal point according to the conditional probability of whole nodes.
Preferably, tunnel threat probability values are calculated according to the conditional probability in the step 5, specifically included:
Step 5.1, the lower limit value in each node probability of occurrence Triangular Fuzzy Number, mean value and upper limit value calculating can obtain
To the Triangular Fuzzy Number probability value of collapsing risk；
Step 5.2 obscures probability value progress de-fuzzy operation to the triangle, to be carried out according to obtained probability value
Risk class divides, and realizes the risk assessment to tunnel.
Preferably, the structure of the fuzzy polymorphism Bayesian network is identical as common Bayesian network, is joined by description
Several and state distribution probability parameter composition, the node of the fuzzy polymorphism Bayes has multiple states, and its probability is three
Angle fuzzy number.
The present invention is based on the thoughts that subjective and objective data combine to propose a kind of fuzzy polymorphism Bayesian network tunnel safety
Methods of risk assessment, this method can be based on historical data identification risk factors and establish fuzzy polymorphism bayesian network structure, so
Root node probability of occurrence is estimated by the expert investigation method that expert's confidence index, weighted index and probability interval combine again afterwards
Triangular Fuzzy Number, based on the subjective and objective calculate node conditional probability combined, finally combine Bayesian network model reasoning
Algorithm solves tunnel risk size.The method can reduce the subjectivity of risk factors identification, and raising risk profile is accurate, to be
Tunnel construction risk profile provides more reasonable effective foundation.
Detailed description of the invention
Fig. 1 is a kind of process of the tunnel risk evaluating method based on fuzzy polymorphism Bayesian network in the embodiment of the present invention
Schematic diagram；
Fig. 2 is tunnel cave accident tree graph in the embodiment of the present invention；
Fig. 3 is accident tree logic gate and Bayesian network conditional probability corresponding relationship in the embodiment of the present invention；
Fig. 4 is tunnel cave risk bayesian network structure figure in the embodiment of the present invention；
Fig. 5 is the T node condition probability being calculated in the embodiment of the present invention by accident case；
Fig. 6 is that Pro1 estimates the conditional probability of node T in the embodiment of the present invention；
Fig. 7 is the conditional probability of interior joint of embodiment of the present invention T.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, is not used to limit
The fixed present invention.Based on the embodiments of the present invention, those of ordinary skill in the art institute without making creative work
The every other embodiment obtained, shall fall within the protection scope of the present invention
Embodiment 1. is as shown in Figure 1, the present invention provides a kind of tunnel risk assessment based on fuzzy polymorphism Bayesian network
Method, this method comprises the following steps:
Step 1: establishing the accident tree based on tunnel risk by the analysis to accident histories data；
The accident tree based on risk factors is established in the step 1, is specifically included:
Step 1.1, by analyzing the reason of accident case, determine the elementary event for causing risk to occur and intermediate thing
Part；
Step 1.2 constructs accident tree according to the elementary event and the correlation of intermediate event；
Wherein, the risk include cave-in accident rank, supporting intensity, surrounding rock stability, design rationality, construction because
Element, disadvantageous geologic condition, degree of prospecting, all information degree, parameter choose reasonability, job specfication, construction quality, supporting
Timeliness, by design and construction degree, support material acceptable level, Grades of Surrounding Rock, special geologic condition, Atmospheric precipitation, draining and
Shi Xing, seepage action of ground water amount, geological conditions variability, construction disturbance.
Step 2: constructing fuzzy polymorphism bayesian network structure based on the accident tree；
Fuzzy polymorphism bayesian network structure is constructed based on the accident tree in the step 2, is specifically included:
Tunnel risk class is divided into not by step 2.1 according to the different conditions of node each in Tunnel Engineering process of construction
Same stage, different phase are indicated with different fuzzy numbers；Traditional Bayesian network node is simply described as two states, but
It is each node physical presence various states in Tunnel Engineering process of construction.Be not in Risk Analysis Process it is simple occur or
Person does not occur two states, and parameter is really gradual change.Be divided into different phase according to tunnel risk class, and with fuzzy number 1,
2, it 3,4,5 ... indicates.
Step 2.2, according to the accident tree determine father node and child node between relationship, by the network knot of accident tree
Structure is converted to the structure of Bayesian network.Relationship between the father node determined according to step 1) accident tree method and child node.It will
The network structure of accident tree is converted to the structure of Bayesian network.The node and logical relation and thing of fuzzy polymorphism Bayesian network
Gu Shu's is identical, and the probability calculation formula with door isOr the probability calculation of door is public
Formula is
Step 3: the Triangular Fuzzy Number of estimation root node probability of occurrence；
Step 3.1, weight coefficient and confidence index data of the building based on normal distribution；
For the influence for reducing subjectivity and uncertain factor, propose the weight confidence index method based on normal distribution to expert
The judgement provided is handled.The determination of Weight of Expert coefficient need to be carried out according to the practicing time of expert, academic title, education background etc.
It divides.Determining that Weight of Expert grade is divided into 5, weighting levels coefficient is indicated with ξ, and value is respectively 1,0.9,0.8,0.7,0.6,
The judgement that ξ=1 represents expert is completely credible, as shown in table 1.
1 Weight of Expert of table divides
Confidence indexRefer to expert's oneself degree of recognition to result when making corresponding judgement, i.e., expert to oneself into
The evaluation of capable objectivity and reliability, this will improve the accuracy of data, its value range is determined as [0.5,1], with
Expert corresponding judgement is provided to the accuracy that oneself judges.The confidence index the big, represents expert and thinks the knot oneself provided
Fruit is more reliable.
When determining elementary event probability of occurrence, there are locating probability levels by estimation event and judges event in step 3.2
The probability of appearance, and probability levels are divided according to practical ranges；
When determining elementary event probability of occurrence by expert survey, it is difficult directly to provide accurate probability by expert, one
As using the method for dividing probability levels, there are locating probability levels by estimation event and judges event probability of occurrence in expert,
Rather than the accuracy probability value occurred.The division of probability interval directly influences the reliability of estimation, and small probability interval can mention
The accuracy of height estimation, big probability interval will increase uncertainty.However according to correlative study, 5 to 9 grades of division is conducive to
Accuracy is improved, the probability interval division for crossing multi-grade is unfavorable for practical application.The probability interval that root node occurs the present invention
It is divided into 9 sections, as shown in table 2, section klIt is defined as [al,al+1], use mlIndicate the average value in section.
2 probability levels of table divide (%)
Step 3.3 calculates different probability grade according to the weight coefficient, confidence index data and probability levels
Estimated value constructs the sample space of each root node different conditions according to the probabilistic estimated value, calculates all root nodes and exists
Triangular Fuzzy Number under different conditions.Analyze rejecting abnormalities data.Either field measurement data or expert survey are obtained
Inevitably there is error or exceptional value in the data obtained, the presence of exceptional value will lead to the deviation to result understanding, therefore
It needs to do exceptional value and gives up processing.The present invention selects Chauvenet method to carry out abnormal data analysis, it is believed that error distribution follows
Normal distribution rejects the data when the probability that error occurs is less than 1/ (2n).By sample architecture statisticσ
For standard deviation, significance 1/ (2n) is selected, it can be by formulaAcquire critical value Wn.Work as suspicious data
Residual error meets criterionThen data xdIt should give up.Step is calculated to specifically include that
(1) mean value and standard deviation of data sample are calculated；
(2) maximum or minimum value in data are chosen, maximum allowable offset is calculated, according to formula
WithCalculate judgment criteria WnAnd σ；
(3) formula is utilizedIt is calculated, meets discrimination standard's
Then give up.The mean value and standard deviation for recalculating data space after the completion and again carry out cycle calculations, until no data meets house
Until non-paying bid is quasi-.
Because reducing centered on event probability of occurrence grade selected by the expert to both sides, so present invention assumes that event occurred
Probability levels are distributed Normal Distribution.Consider that confidence index when expert judgments, expert n judge elementary event XiIn xiShape
Expert confidence index of the state in probability interval kUnder normal circumstancesValue will be less than 1, then for the judgement of expert
It hasA possibility that fall within other probability intervals, according to the tracing pattern of normal distribution it is found that expert n judges XiIn xi
The probability that state occurs can gradually become smaller in the two sides of k, and closer to the section of k, its probability will be bigger, falls into each probability interval
A possibility that Level k isIt is determined by following formula:
Expert n estimates event XiStatus xiProbability results consider its weight using formulaIt is handled, expert n can be obtained to event XiState in which xiOccur
Estimated probability value
In practical engineering applications, it will there are several experts to participate in investigation (being assumed to be N), commented what every expert provided
Estimate result to be handled by above-mentioned formula, every expert can be obtained to elementary event XiEach state xiThe estimated value of probability of occurrence
(observation) forms statistical sample by the probabilistic estimated value of all expertsIn 3.3)
Method for processing abnormal data handles sample space data, then in order to integrate the data of all experts to obtain root section
The reasonable probability of point, confidence interval method are used for expert investigation data configuration statistical variable (considering each Weight of Expert).Because of sample
Overall Unknown Variance, therefore select sample varianceTo replace σ2, then the statistical variable that constructsT distribution is obeyed to estimate normal distribution.
Then for given α, enableT can be obtainedα2(N-1) value:
Nodes XiTriangular Fuzzy Number by confidence interval method, by formulaIt is calculated.And this method can be used to other nodes,
To finally obtain all root node fuzzy probability numbers.
Step 4: calculating simultaneously merge node conditional probability.
Step 4.1, the present invention propose to calculate using expert investigation data with the subjective and objective method of weighting that observation data combine,
By utilizing objective weight (representing project data) and subjective weight (representing expertise).The conditional probability of each node, with
Measured data increase, by reducing subjective weight, gradually decrease the influence of expertise, calculation formula A=k1A1+k2A2。
A in formula1And A2It is from that gained conditional probability of project data and expert investigation data respectively；k1And k2It is subjective and objective power respectively
Weight coefficient, is determined by data cases.
Step 5: tunnel threat probability values are calculated according to the conditional probability.
Step 5.1, by collect and investigation obtain the fuzzy number of root node probability of occurrence after, as data
Using fuzzy polymorphism Bayesian network risk dynamic evaluation approach to risk probability event carry out probability of happening calculate P (T=t)=
(at,mt,bt)。
Step 5.2, in order to divide risk class, to fuzzy probability carry out de-fuzzy operation, thus general according to what is obtained
Rate value carries out risk class division, realizes the risk assessment to tunnel.According to " urban track traffic underground engineering construction risk pipe
Reason specification " (GB50652-2011), risk class is divided into 5 grades, every grade of probability of happening is as shown in table 3.
3 probability levels of table divide
2. system embodiment of embodiment
Risk factors are identified based on historical data described in the embodiment of the present invention and establish accident tree, are specifically included:
The influence factor of tunnel cave is intricate, and the generation of most accidents is all by many factors while to make
With caused as a result, determining collapsing thing by analyzing and researching using accident tree method to 40 mountain tunnel cave-in accidents
Therefore elementary event it is as shown in the table, accident tree is as shown in Fig. 2, 5 intermediate nodes and 15 root nodes can be divided into.
4 cave-in accident reason of table
Fuzzy polymorphism bayesian network structure is constructed based on accident tree method described in the embodiment of the present invention, is specifically included:
2.1) tunnel cave, tunnel may undergo small deformation, obvious deformation and 3 different phases of collapsing.Therefore, each
Node should be divided into multiple states, such as collapsing risk is divided into 5 stages, be small deformation respectively, large deformation, small-sized collapse
It collapses, medium-sized collapsing and large-scale collapse, and is indicated respectively with fuzzy number 1,2,3,4,5.
2.2) according to the relationship between the accident tree method father node determined and child node.The network structure of accident tree is converted
For the structure of Bayesian network.The node and logical relation of fuzzy polymorphism Bayesian network are identical as accident tree, corresponding relationship
As shown in Figure 3.The bayesian network structure established is as shown in Figure 4.
The Triangular Fuzzy Number that root node probability of occurrence is estimated described in the embodiment of the present invention, specifically includes:
3.1) questionnaire survey is carried out to 22 associated specialists for certain work tunnel, obtains 20 parts of full investigation questionnaires altogether.With
Elementary event X9For, table 5 be 20 parts of questionnaires for country rock grade X9 expert investigation as a result, being each expert in table for hole
Mouth section country rock is in I~V judgement probability levels and its confidence index, and true according to the academic title of expert and working experience
Fixed weight coefficient.Formula is utilized to 20 parts of survey datas first:
The probabilistic estimated value that each expert is I~V grade to Portal Section country rock grade is calculated, then according to probability Estimation
Value constructs 20 sample values that the node is respectively at 5 states, by formula:
The sample space of each root node different conditions is constructed, and X9 root node is calculated and is respectively at 5 states
Triangle obscures probability number.Table 6 show the fuzzy number probability value of country rock grade.
Remaining 14 root node is calculated using identical method, you can get it under 4 kinds of states, 15 root sections
The Probabilistic Fuzzy triangular number that point respectively occurs, the different conditions probability of occurrence for only listing existing X9 because length is limited.
5 X of table9Expert investigation result
6 X of table9Different conditions probability of occurrence
Table 3-9 Probability of different scenarios of node X9
It is calculated and merge node conditional probability, specific packet described in the embodiment of the present invention based on accident case and expertise
It includes:
4.1) conditional probability calculating is carried out by statistical data
It as shown in table 6, is the state grade and its risk class of each node of part cave-in accident case, by each section
Dotted state data substitute into the fuzzy polymorphism Bayesian network computation model constructed, carry out node parameter study, available portion
The hazy condition probability and root node probability of occurrence of partial node, as in Figure 3-5.As can be seen from the table because data are less, lead
Causing part of nodes conditional probability is immediately arrived at by being uniformly distributed, such as M1=1 and M2When=2, node T is in the probability of 5 grades
It is all equal p=0.2, there is also similar situations for other nodes, and it is general to node condition to introduce expert survey in the case
Rate is investigated and analysed, to make up data flaw caused by data deficiencies.
The node T partial condition probability that 7 Pro1 of table is provided
4.2) calculating of expert survey conditional probability
The embodiment of the present invention selects 9 experts to carry out conditional probability investigation.As shown in table 8 it is expert Pro1, is saved in his father
Point M1And M2When taking different conditions respectively, for the conditional probability application form of child node possible state, M in table1And M2Arrange generation respectively
The different conditions of two nodes of table, S1L and S1C respectively represent the possibility probability levels and confidence index when state is 1, using general
The method of rate grade classification obtains the confidence probability of Level1~Level9 respectively, and utilizes formula:
It obtains egress T and is respectively at 5 shape probability of states under each combination of father node, and to the probability under each combination
It is normalized, the reasonability of guarantee probability numerical value.
As shown in table 9 for 9 expert investigation data, after being calculated using above-mentioned formula, obtained node T is in M1=2 and M2
When=2 conditional probability p (T=i | M2=2, M1=2).
The node T partial condition probability that 8 Pro1 of table is provided
9 subjective method node T of table partial condition Probability p (T=i | M2=2, M1=2)
Table 9 is to participate in 9 experts of investigation respectively to node T probability calculation in different states as a result, can from table
To find out that each expert is between 0.15~0.41 the estimated value of conditional probability, carried out using Chauvenet method abnormal
Data processing.
When for T=1,9 estimated datas are shared, by formulaDetermine judgment criteria Wnσ=
1.92 × 0.08325=0.15984, by formulaKnow that most probable value is what expert Pro6 was provided
Estimated value, deviationMinimum probability estimated value is that expert Pro2 is provided
Estimated value, deviationThen need the estimated value for giving up expert Pro2
0.15.Chauvenet method is applied to judge again remaining 8 expert's estimated values, by formula
Determine judgment criteria Wnσ=1.86 × 0.04=0.074, by formulaKnow maximum estimated valueMinimum valuationThen no data needs
It is deleted, the estimated value of other 8 experts is almost the same in addition to the estimated value of Pro2.
The data of other states in table 9 are handled, state 2~5 is no different regular data and carries out delete processing.Utilize the party
Method handles all node datas, finally obtains all estimated values for meeting judgment criterion, and obtain using probabilistic method
Obtain node condition probability tables finally.
After completing the data processing of its application form to each expert respectively, abnormal data is given up method and is examined to data
It tests, rejecting abnormalities data, and state normalization processing is carried out by its weight.
Using the data (deleting estimation of the Pro2 for state 1) and its weight of each expert in table 9 by formulaAfter calculating and being normalized, T can be obtained in M2=2, M1=2
5 states under part conditional probability p (T=i | M2=2, M1It=2) is respectively 0.36,0.44,0.13,0.05 and 0.02.Pass through
The probability of the above available T of relevant calculation process at different conditions obtains whole nodes after calculating all nodes
Conditional probability, as shown in Figure 6.
4.3) subjective and objective to combine conditional probability calculating
The conditional probability of each node obtained by subjective and objective weight combined techniques is assigned to the Bayesian network established, and benefit
With expert survey to scene parameter state probability of occurrence tested and investigated, finally obtain all root nodes it is fuzzy go out
Show probability, and the fuzzy set of risk assessment is calculated jointly with hazy condition probability tables, to obtain its risk class shape
State.Because case data is less in embodiments of the present invention, by objective weight value 0.1, subjective weight value 0.9 utilizes formula A
=k1A1+k2A2The conditional probability for obtaining each node after more merging, is then substituted into fuzzy polymorphism Bayesian network,
Carry out risk assessment, conditional probability distribution such as Fig. 7 of gained node T.
Calculated risk grade described in the embodiment of the present invention and diagnosis accident, specifically include:
It after obtaining the conditional probability of intermediate node, is conducted into the fuzzy polymorphism Bayesian network built up, in institute
On the basis of obtaining root node probability of occurrence, the threat probability values of collapsing can be calculated.It is said by taking Scenario 1 as an example
It is bright, available collapsing risk is calculated with lower limit value, mean value and the upper limit value in each node probability of occurrence Triangular Fuzzy Number
Triangular Fuzzy Number probability value, such as formula
Shown in.It using the α method of weighting and takes α=0.5 and carries out de-fuzzy calculating, acquired results are as shown in table 10.Though
Right risk factors X9 possesses different probability distribution under different situations, but result difference is smaller, illustrates in the Tunnel Engineering
In the two factors for collapsing risk contribute it is smaller.Table 10 illustrates that the risk class of large-scale collapse and the appearance of medium-sized collapsing is
" possible ", and the risk class that small-sized collapsing, obvious deformation occur is " frequent ".
Risk probability and grade under the different situations of table 10
To sum up, the thought combined the present invention is based on subjective and objective data proposes a kind of fuzzy polymorphism Bayesian network tunnel
Safety risk estimating method, this method can be identified risk factors based on historical data and establish fuzzy polymorphism Bayesian network knot
Structure, the expert investigation method estimation root node then combined again by expert's confidence index, weighted index and probability interval go out
The Triangular Fuzzy Number of existing probability finally combines Bayesian network model based on the subjective and objective calculate node conditional probability combined
Reasoning algorithm solve tunnel risk size.The method can reduce the subjectivity of risk factors identification, and raising risk profile is accurate,
To provide more reasonable effective foundation for tunnel construction risk profile.
In addition, it should be understood that although this specification is described in terms of embodiments, but not each embodiment is only wrapped
Containing an independent technical solution, this description of the specification is merely for the sake of clarity, and those skilled in the art should
It considers the specification as a whole, the technical solutions in the various embodiments may also be suitably combined, forms those skilled in the art
The other embodiments being understood that.
Claims (7)
1. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network, it is characterised in that: this method comprises:
Step 1: establishing the accident tree based on tunnel risk by the analysis to accident histories data；
Step 2: constructing fuzzy polymorphism bayesian network structure based on the accident tree；
Step 3: the Triangular Fuzzy Number of estimation root node probability of occurrence；
Step 4: calculating simultaneously merge node conditional probability；
Step 5: tunnel threat probability values are calculated according to the conditional probability.
2. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 1, feature
It is: establishes the accident tree based on risk factors in the step 1, specifically include:
Step 1.1, by analyzing the reason of accident case, determine the elementary event for causing risk to occur and intermediate event；
Step 1.2 constructs accident tree according to the elementary event and the correlation of intermediate event；
Wherein, the risk includes cave-in accident rank, supporting intensity, surrounding rock stability, design rationality, construction factor, no
Sharp geological conditions, degree of prospecting, all information degree, parameter selection reasonability, job specfication, construction quality, supporting are timely
Property, by design and construction degree, support material acceptable level, Grades of Surrounding Rock, special geologic condition, Atmospheric precipitation, draining timeliness,
Seepage action of ground water amount, geological conditions variability, construction disturbance.
3. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 2, feature
It is: fuzzy polymorphism bayesian network structure is constructed based on the accident tree in the step 2, is specifically included:
Tunnel risk class is divided into not same order according to the different conditions of node each in Tunnel Engineering process of construction by step 2.1
Section, different phase are indicated with different fuzzy numbers；
Step 2.2, according to the accident tree determine father node and child node between relationship, by the network structure of accident tree turn
It is changed to the structure of Bayesian network.
4. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 3, feature
It is: estimates the Triangular Fuzzy Number of root node probability of occurrence in the step 3, specifically include:
Step 3.1, weight coefficient and confidence index data of the building based on normal distribution；
When determining elementary event probability of occurrence, there are locating probability levels by estimation event and judge event appearance in step 3.2
Probability, and according to practical ranges divide probability levels；
Step 3.3, the estimation that different probability grade is calculated according to the weight coefficient, confidence index data and probability levels
Value, the sample space of each root node different conditions is constructed according to the probabilistic estimated value, calculates all root nodes in difference
Triangular Fuzzy Number under state.
5. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 4, feature
Be: calculating and merge node conditional probability in the step 4 specifically include:
Step 4.1, building expert investigation group data, the conditional probability of all nodes is calculated according to investigation group's data；
Step 4.2 carries out dealing of abnormal data, rejecting abnormalities data using Chauvenet method, and carries out at state normalization
Reason, so that the probability of all nodes at different conditions is obtained, thus the conditional probability of the whole nodes obtained.
Step 4.3 obtains the conditional probability for merging posterior nodal point according to the conditional probability of whole nodes.
6. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 5, feature
It is: tunnel threat probability values is calculated according to the conditional probability in the step 5, are specifically included:
Step 5.1, the lower limit value in each node probability of occurrence Triangular Fuzzy Number, mean value and upper limit value calculate available collapse
The Triangular Fuzzy Number probability value for risk of collapsing；
Step 5.2 obscures probability value progress de-fuzzy operation to the triangle, to carry out risk according to obtained probability value
Grade classification realizes the risk assessment to tunnel.
7. a kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network according to claim 1, feature
Be: the structure of the fuzzy polymorphism Bayesian network is identical as common Bayesian network, is all by characterising parameter and state point
Cloth probability parameter composition, the node of the fuzzy polymorphism Bayes has multiple states, and its probability is Triangular Fuzzy Number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910320582.4A CN110059963A (en) | 2019-04-20 | 2019-04-20 | A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910320582.4A CN110059963A (en) | 2019-04-20 | 2019-04-20 | A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110059963A true CN110059963A (en) | 2019-07-26 |
Family
ID=67319888
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910320582.4A Pending CN110059963A (en) | 2019-04-20 | 2019-04-20 | A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110059963A (en) |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110427015A (en) * | 2019-08-02 | 2019-11-08 | 广东职业技术学院 | A kind of boiler afterheat explosion accident diagnostic analysis method |
| CN110807580A (en) * | 2019-10-25 | 2020-02-18 | 上海建科工程咨询有限公司 | Method for analyzing key safety risk of super high-rise construction machinery based on complex network |
| CN110968865A (en) * | 2019-11-27 | 2020-04-07 | 桂林电子科技大学 | Android software risk assessment method based on probability ontology |
| CN111105163A (en) * | 2019-12-23 | 2020-05-05 | 交通运输部科学研究院 | Traffic engineering potential safety hazard assessment device and method |
| CN111242504A (en) * | 2020-01-23 | 2020-06-05 | 南京工业大学 | Coal gasification device risk probability calculation method based on domino effect |
| CN111311092A (en) * | 2020-02-13 | 2020-06-19 | 辽宁石油化工大学 | Coal gasification equipment dynamic risk assessment method |
| CN111401653A (en) * | 2020-03-25 | 2020-07-10 | 华中科技大学 | Tunnel water leakage risk spatial dependency prediction method and prediction system |
| CN111598352A (en) * | 2020-05-25 | 2020-08-28 | 哈尔滨工业大学 | Concrete beam type bridge comprehensive evaluation method based on Bayesian network |
| CN111985804A (en) * | 2020-08-18 | 2020-11-24 | 华中科技大学 | Shield approaching existing tunnel safety evaluation method based on data mining and data fusion |
| CN112328961A (en) * | 2020-11-04 | 2021-02-05 | 江苏海拓润达科技发展有限公司 | On-line monitoring device quality evaluation system based on fault tree and Bayesian network |
| CN112465304A (en) * | 2020-11-07 | 2021-03-09 | 西南交通大学 | Railway turnout area train derailment accident assessment method based on Bayesian network |
| CN112668865A (en) * | 2020-12-23 | 2021-04-16 | 贵阳市城市轨道交通集团有限公司 | Urban subway risk dynamic analysis method |
| CN113537695A (en) * | 2021-05-28 | 2021-10-22 | 东莞理工学院 | Quantitative evaluation method for excessive emission risk of flue gas pollutants of waste incineration power plant |
| CN114723993A (en) * | 2022-04-14 | 2022-07-08 | 江苏海洋大学 | Bayesian network-based rural house identification grade classification method |
| CN115392797A (en) * | 2022-10-27 | 2022-11-25 | 北京城建设计发展集团股份有限公司 | Operating tunnel structure disease rapid intelligent diagnosis method based on Bayesian network |
| CN115438867A (en) * | 2022-09-14 | 2022-12-06 | 中国矿业大学 | Coal mine roof accident risk prediction method |
| CN115907565A (en) * | 2023-02-14 | 2023-04-04 | 清华四川能源互联网研究院 | Diversion tunnel structure safety evaluation method and device, electronic equipment and storage medium |
| CN116629614A (en) * | 2023-06-05 | 2023-08-22 | 北京城建设计发展集团股份有限公司 | Dynamic evaluation method for urban deep karst collapse risk based on Bayesian network |
| CN110738399B (en) * | 2019-12-16 | 2023-10-13 | 中山大学 | Judicial trial flow deviation early warning method based on fuzzy set theory |
| CN116882548A (en) * | 2023-06-15 | 2023-10-13 | 中国矿业大学 | Tunneling roadway coal and gas outburst prediction method based on dynamic probability reasoning |
| CN116629614B (en) * | 2023-06-05 | 2024-05-10 | 北京城建设计发展集团股份有限公司 | Dynamic evaluation method for urban deep karst collapse risk based on Bayesian network |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5830135A (en) * | 1994-03-31 | 1998-11-03 | Bosque; Elena M. | Fuzzy logic alarm system for pulse oximeters |
| US7774293B2 (en) * | 2005-03-17 | 2010-08-10 | University Of Maryland | System and methods for assessing risk using hybrid causal logic |
| CN105717912A (en) * | 2016-01-27 | 2016-06-29 | 西北工业大学 | Reliability analysis method for electromechanical actuator based on fuzzy dynamic fault tree |
| CN106846155A (en) * | 2017-03-29 | 2017-06-13 | 哈尔滨理工大学 | Submarine pipeline leakage accident methods of risk assessment based on fuzzy Bayesian network |
| CN107179765A (en) * | 2017-06-08 | 2017-09-19 | 电子科技大学 | A kind of heavy digital control machine tool electrical control and drive system reliability analysis method |
| CN109191005A (en) * | 2018-09-25 | 2019-01-11 | 武汉理工大学 | Retired auto parts and components workability evaluation method for classified and graded management |
| CN109657907A (en) * | 2018-11-13 | 2019-04-19 | 香港理工大学深圳研究院 | Method of quality control, device and the terminal device of geographical national conditions monitoring data |
-
2019
- 2019-04-20 CN CN201910320582.4A patent/CN110059963A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5830135A (en) * | 1994-03-31 | 1998-11-03 | Bosque; Elena M. | Fuzzy logic alarm system for pulse oximeters |
| US7774293B2 (en) * | 2005-03-17 | 2010-08-10 | University Of Maryland | System and methods for assessing risk using hybrid causal logic |
| CN105717912A (en) * | 2016-01-27 | 2016-06-29 | 西北工业大学 | Reliability analysis method for electromechanical actuator based on fuzzy dynamic fault tree |
| CN106846155A (en) * | 2017-03-29 | 2017-06-13 | 哈尔滨理工大学 | Submarine pipeline leakage accident methods of risk assessment based on fuzzy Bayesian network |
| CN107179765A (en) * | 2017-06-08 | 2017-09-19 | 电子科技大学 | A kind of heavy digital control machine tool electrical control and drive system reliability analysis method |
| CN109191005A (en) * | 2018-09-25 | 2019-01-11 | 武汉理工大学 | Retired auto parts and components workability evaluation method for classified and graded management |
| CN109657907A (en) * | 2018-11-13 | 2019-04-19 | 香港理工大学深圳研究院 | Method of quality control, device and the terminal device of geographical national conditions monitoring data |
Non-Patent Citations (4)
| Title |
|---|
| JINGLAI SUN等: "Tunnel collapse risk assessment based on multistate fuzzy Bayesian networks", 《QUALITY AND RELIABILITY ENGINEERING INTERNATIONAL》 * |
| 周宗青等: "浅埋隧道塌方地质灾害成因及风险控制", 《岩土力学》 * |
| 郑俊杰等: "基于模糊故障树的盾构隧道施工成本风险评估", 《岩土工程学报》 * |
| 马德仲: "基于贝叶斯网络和多源信息构建可靠性分析模型方法研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110427015A (en) * | 2019-08-02 | 2019-11-08 | 广东职业技术学院 | A kind of boiler afterheat explosion accident diagnostic analysis method |
| CN110807580B (en) * | 2019-10-25 | 2022-03-04 | 上海建科工程咨询有限公司 | Method for analyzing key safety risk of super high-rise construction machinery based on complex network |
| CN110807580A (en) * | 2019-10-25 | 2020-02-18 | 上海建科工程咨询有限公司 | Method for analyzing key safety risk of super high-rise construction machinery based on complex network |
| CN110968865A (en) * | 2019-11-27 | 2020-04-07 | 桂林电子科技大学 | Android software risk assessment method based on probability ontology |
| CN110968865B (en) * | 2019-11-27 | 2022-03-11 | 桂林电子科技大学 | Android software risk assessment method based on probability ontology |
| CN110738399B (en) * | 2019-12-16 | 2023-10-13 | 中山大学 | Judicial trial flow deviation early warning method based on fuzzy set theory |
| CN111105163A (en) * | 2019-12-23 | 2020-05-05 | 交通运输部科学研究院 | Traffic engineering potential safety hazard assessment device and method |
| CN111242504A (en) * | 2020-01-23 | 2020-06-05 | 南京工业大学 | Coal gasification device risk probability calculation method based on domino effect |
| CN111242504B (en) * | 2020-01-23 | 2023-08-22 | 南京工业大学 | Method for calculating risk probability of coal gasification device based on domino effect |
| CN111311092A (en) * | 2020-02-13 | 2020-06-19 | 辽宁石油化工大学 | Coal gasification equipment dynamic risk assessment method |
| CN111311092B (en) * | 2020-02-13 | 2024-04-19 | 辽宁石油化工大学 | Evaluation method based on dynamic risk of coal gasification equipment |
| CN111401653A (en) * | 2020-03-25 | 2020-07-10 | 华中科技大学 | Tunnel water leakage risk spatial dependency prediction method and prediction system |
| CN111598352A (en) * | 2020-05-25 | 2020-08-28 | 哈尔滨工业大学 | Concrete beam type bridge comprehensive evaluation method based on Bayesian network |
| CN111985804B (en) * | 2020-08-18 | 2021-09-10 | 华中科技大学 | Shield approaching existing tunnel safety evaluation method based on data mining and data fusion |
| CN111985804A (en) * | 2020-08-18 | 2020-11-24 | 华中科技大学 | Shield approaching existing tunnel safety evaluation method based on data mining and data fusion |
| CN112328961A (en) * | 2020-11-04 | 2021-02-05 | 江苏海拓润达科技发展有限公司 | On-line monitoring device quality evaluation system based on fault tree and Bayesian network |
| CN112465304A (en) * | 2020-11-07 | 2021-03-09 | 西南交通大学 | Railway turnout area train derailment accident assessment method based on Bayesian network |
| CN112668865A (en) * | 2020-12-23 | 2021-04-16 | 贵阳市城市轨道交通集团有限公司 | Urban subway risk dynamic analysis method |
| CN113537695A (en) * | 2021-05-28 | 2021-10-22 | 东莞理工学院 | Quantitative evaluation method for excessive emission risk of flue gas pollutants of waste incineration power plant |
| CN113537695B (en) * | 2021-05-28 | 2023-11-21 | 东莞理工学院 | Quantitative evaluation method for risk of excessive emission of flue gas pollutants in garbage incineration power plant |
| CN114723993A (en) * | 2022-04-14 | 2022-07-08 | 江苏海洋大学 | Bayesian network-based rural house identification grade classification method |
| CN114723993B (en) * | 2022-04-14 | 2023-07-04 | 江苏海洋大学 | Rural house identification grade classification method based on Bayesian network |
| CN115438867A (en) * | 2022-09-14 | 2022-12-06 | 中国矿业大学 | Coal mine roof accident risk prediction method |
| CN115392797A (en) * | 2022-10-27 | 2022-11-25 | 北京城建设计发展集团股份有限公司 | Operating tunnel structure disease rapid intelligent diagnosis method based on Bayesian network |
| CN115907565A (en) * | 2023-02-14 | 2023-04-04 | 清华四川能源互联网研究院 | Diversion tunnel structure safety evaluation method and device, electronic equipment and storage medium |
| CN116629614A (en) * | 2023-06-05 | 2023-08-22 | 北京城建设计发展集团股份有限公司 | Dynamic evaluation method for urban deep karst collapse risk based on Bayesian network |
| CN116629614B (en) * | 2023-06-05 | 2024-05-10 | 北京城建设计发展集团股份有限公司 | Dynamic evaluation method for urban deep karst collapse risk based on Bayesian network |
| CN116882548A (en) * | 2023-06-15 | 2023-10-13 | 中国矿业大学 | Tunneling roadway coal and gas outburst prediction method based on dynamic probability reasoning |
| CN116882548B (en) * | 2023-06-15 | 2024-05-17 | 中国矿业大学 | Tunneling roadway coal and gas outburst prediction method based on dynamic probability reasoning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059963A (en) | A kind of tunnel risk evaluating method based on fuzzy polymorphism Bayesian network | |
| CN107943880B (en) | Geological disaster susceptibility improvement and evaluation method based on analytic hierarchy process | |
| CN112506990B (en) | Hydrological data anomaly detection method based on spatiotemporal information | |
| CN104899437A (en) | Early-warning method for heavy-rainfall type landslide hazard | |
| CN103345566B (en) | Based on the geochemical anomaly discrimination and evaluation method of Geological Connotation | |
| CN110610285A (en) | Underground metal mine goaf risk grading evaluation method | |
| CN112182234B (en) | Basin flood control planning data knowledge graph construction method | |
| CN107610021A (en) | The comprehensive analysis method of environmental variance spatial and temporal distributions | |
| CN101673369A (en) | Projection pursuit-based method for evaluating flooding risk of drainage pipe network | |
| CN113642849A (en) | Geological disaster risk comprehensive evaluation method and device considering spatial distribution characteristics | |
| US20230385366A1 (en) | Method for ecological disturbance risk identification and assessment based on automatic parameter adjusting optimization model | |
| CN112396320A (en) | Tunnel collapse risk assessment method based on ISM and fuzzy Bayesian network | |
| CN112884014A (en) | Traffic speed short-time prediction method based on road section topological structure classification | |
| CN111144637A (en) | Regional power grid geological disaster forecasting model construction method based on machine learning | |
| CN111199298A (en) | Flood forecasting method and system based on neural network | |
| CN115049124A (en) | Deep and long tunnel water inrush prediction method based on Bayesian network | |
| CN114881396A (en) | Tunnel collapse risk assessment method based on AHP and TOPSIS | |
| CN117113038B (en) | Urban water and soil loss Huang Nishui event tracing method and system | |
| CN107977727B (en) | Method for predicting blocking probability of optical cable network based on social development and climate factors | |
| JP3674707B1 (en) | Disaster prevention business plan support system and method | |
| CN109190783A (en) | City river network leaks spatial aggregation detection and key influence factor recognition methods | |
| Schismenos et al. | Risk Flood Assessment for the Arachthos River using Analytic Hierarchy Process | |
| Teh Noranis et al. | Fuzzy AHP in a knowledge-based framework for early flood warning | |
| Wang et al. | Hazard assessment of debris flow based on infinite irrelevance method and probabilistic neural network coupling Model | |
| Li et al. | Risk Assessment of Debris Flow in Huyugou River Basin Based on Machine Learning and Mass Flow |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190726 |
|
| RJ01 | Rejection of invention patent application after publication |