Reservoir dam risk assessment method

Abstract

The invention discloses a reservoir dam risk assessment method, which comprises the following steps of: constructing an index system taking human, technology, organization and environment as criterion layers, performing weighting on indexes by adopting AHP and improved CRITIC methods, performing weighting combination based on a game theory method to obtain a combination weight of each index, and realizing dynamic weighting of top layer indexes by using an incentive variable weight method. Compared with the prior art, the reservoir dam risk assessment method has the advantages that human factor analysis and engineering dam break are combined; index weight combination weighting is carried out; using a variable weight risk assessment method; and the accuracy and the practical property of reservoir dam risk assessment are realized.

Description

Technical Field

[0001] The present invention relates to a method for risk assessment of reservoir dams, belonging to the technical field of dam safety prediction in water conservancy engineering technology. Background Art

[0002] Under the action of multiple disaster-causing factors, the risk of reservoir accidents or even dam breaches still exists, seriously threatening the lives and property safety of the people downstream. The frequent occurrence of extreme weather and the serious aging or even overdue service of reservoirs have increased the probability of dam breaches, and the rapid development of the economy and society and the significant increase in the urbanization rate have exacerbated the losses caused by dam breaches. With the progress of technology and the improvement of emergency management level, the proportion of dam breach accidents under high water levels during the flood season of reservoirs has gradually decreased, while the non-flood season dam breach accidents have increased, indicating that there are still obvious weak links in reservoir management in aspects such as inspection, monitoring and early warning, operation scheduling, and emergency plans.

[0003] The application of quantitative risk analysis for dam safety has received extensive attention. The research trend of dam risk analysis is gradually changing from certainty to probability and possibility. In the past half century, the research on dam engineering has mainly focused on the numerical simulation of dam structures, mainly including the finite difference method, the finite element method, the discrete element method, the meshless method, etc. Usually, software is used to develop the finite element model of the dam and its yielding environment, considering the coupled system mechanics among the dam body, reservoir water, and bedrock. Most of the existing methods only start from the engineering perspective and consider the human-machine system in reservoir management less.

[0004] Although some scholars have introduced human factor reliability analysis into the field of dam risk assessment, analyzed the root causes of human factor failures, and proposed dam breach models and calculation methods, the existing models only use the expert judgment method to complete the conditional probability distribution table of the Bayesian network, without making full use of historical dam breach data and having certain subjectivity. Dam risk management is a problem of multi-index information aggregation, and how to evaluate the management level of reservoirs is a challenge. Summary of the Invention

[0005] Object of the Invention: Aiming at the problems and deficiencies in the prior art, the present invention provides a method for risk assessment of reservoir dams.

[0006] Technical Solution: A method for risk assessment of reservoir dams constructs an index system with human, technology, organization, and environment as the criterion layer, uses the AHP and improved CRITIC methods to assign weights to the indicators, conducts weight combination based on the game theory method to obtain the combined weights of each indicator, and uses the incentive variable weight method to realize the dynamic weighting of the top-level indicators, constructing a new reservoir risk assessment framework considering human factors. The construction method includes the following steps:

[0007] S1. Obtain historical dam - break data information through text mining methods, form a reservoir dam risk assessment factor set, construct a reservoir dam risk assessment index system, and generate a Bayesian network relationship diagram.

[0008] S3. Use the AHP method to calculate the subjective weight matrix.

[0009] S3. Use the improved CRITIC method, combined with historical dam - break data, to calculate the objective weight.

[0010] S4. Use the game theory method to perform weighted combination on the subjective and objective weights obtained in S2 and S3 to obtain the combined weight.

[0011] S5. Based on the obtained combined weight, perform probability calculation based on Bayesian theory, infer the conditional probability distribution of the Bayesian network, introduce it into the Bayesian network relationship constructed in S1, form a Bayesian network diagram of human - factor indicators, and quantify the probability of human - factor management errors in dam - break probability calculation.

[0012] S6. Introduce the variable - weight method into the calculation of failure probability, use the incentive variable - weight method to perform variable - weight on the top - level indicators (dam - break reasons) in dam - break probability calculation, construct a new variable - weight function and combine it with the weight calculation in dam - break calculation (existing methods often use constants), and normalize.

[0013] S7. Introduce the probability of human - factor management errors obtained in S5, combine it with the engineering dam - break probability, and use the variable - weight coefficient obtained in S6 to perform variable - weight on the dam - break probability under different dam - break paths to reflect the dam - break probability considering human - factor errors and variable weights.

[0014] S8. Introduce the idea of risk ranking, combine the dam - break consequences to rank the group of dams, so as to judge the risk level of the reservoir group and provide guidance for reservoir capital investment and management operation.

[0015] The specific implementation process of S2 is as follows:

[0016] S2.1 Score and compare the factors at the index layer. For the index layer, use the scoring results to construct a judgment matrix A, where

[0017]

[0018] where n is the number of dimensions, that is, the number of indicators in the index layer, and a ij represents the degree of importance between indicator i and indicator j, and is scored according to the importance of indicator i relative to indicator j.

[0019] S2.2. Use the eigenvector method to calculate the weight vector W. First, calculate the maximum eigenvalue λ max of the judgment matrix A, which exists and is unique. According to AW = λ maxCalculate the weight matrix \(W\), the components of which are positive components \(W\). i , representing the weight of index \(a\). i , that is, \(W=(W\). 1 , \(W\). 2 , …, \(W\). n ) T .

[0020] Normalize the obtained weight vector \(W\) to get the required subjective weight matrix \(W\). (1) , where the calculation method of each component \(W\). i (1) is as follows:

[0021]

[0022] S2.3. Conduct a consistency test on the judgment matrix \(A\) according to the following steps:

[0023] CI = (\(\lambda\). max -n) / (n - 1).

[0024] CR = CI / RI

[0025] where CI is the consistency index, RI is the average random consistency index, and CR is the consistency ratio. If CR < 0.1, it can be considered that the consistency of the judgment matrix is acceptable; otherwise, the judgment matrix needs to be corrected.

[0026] The implementation process of the above S3 is as follows:

[0027] S3.1. Score the indicators in the risk assessment factor set according to the historical dam break records to obtain the matrix:

[0028]

[0029] where \(b\). ij represents the score of the \(j\)th indicator of the \(i\)th dam in the historical dam break records.

[0030] S3.2. Data standardization processing:

[0031]

[0032] where \(b\). j ′ is the matrix generated after standardization processing, \(b\). j is the column matrix corresponding to index \(j\), \(maxb\). j , \(minb\). j are the maximum and minimum values in the column matrix \(b\). j respectively.

[0033] S3.3. Calculate the coefficient of variation \(v\). j :

[0034]

[0035] Among them, is the column matrix arithmetic mean of index j, b ij is the score of the i-th reservoir for index j, and σ j is the standard deviation of the index.

[0036] S3.4. Calculate the index independence h j :

[0037]

[0038] Among them, |r jl | is the absolute value of the correlation coefficient between index j and index l.

[0039] S3.5. Calculate the index comprehensive measurement coefficient q j :

[0040] q j = v j ·h j j = 1, 2,..., n

[0041] S3.6. The weight ω j of index j is

[0042]

[0043] Combined by the weight ω j to obtain the objective weight matrix W (2) .

[0044] S4. Use the game theory method to comprehensively combine the obtained subjective weight matrix W (1) and the objective weight matrix W (2) .

[0045] S4.1. Before performing the weighted combination, first use the distance function to perform a consistency test on the obtained weighting method. The distance function is:

[0046]

[0047] In the formula, is the subjective weight of index j, is the objective weight of index j. The smaller d(W (1) , W (2) ), the closer the two weighting results are. When 0 ≤ d(W (1) , W (2) ) ≤ 1, the consistency requirement is met; otherwise, the consistency requirement is not met and the weights need to be recalculated.

[0048] S4.2. Combine the subjective and objective weights obtained in S2 and S3 using game theory. First, generate a weight vector set W = {W (1) , W (2)} from the subjective and objective weights obtained in S2 and S3. These two weight vectors can be linearly combined arbitrarily to form a possible weight set:

[0049] W = α 1 (W (1) ) T + α 2 (W (2) ) T

[0050] where W (1) is the subjective weight calculated in S2, and W (2) is the objective weight calculated in S3; α 1 , α 2 are the linear combination coefficients of the subjective and objective weights. The set of all possible weight vector combinations is the set of all W.

[0051] S4.3. Optimize the linear combination weight coefficient α k with the goal of minimizing the deviation between W and W (1) , W (2) . That is

[0052]

[0053] According to the differential properties of matrices, the first-order derivative condition for the optimization of the above formula is:

[0054]

[0055] S4.4. After obtaining (α 1 , α 2 ) according to the above formula, normalize it:

[0056]

[0057] S4.5. Finally, the optimal combined weight is:

[0058]

[0059] Thus, the weight W (1) composed of the subjective weight matrix W (2) and the objective weight matrix W * is obtained.

[0060] S5. Based on the combined weights, probability calculations are carried out using Bayesian theory, and the conditional probability distribution of the Bayesian network is deduced. It is introduced into the Bayesian network relationship constructed in S1 to form a Bayesian network diagram of human factor indicators, so as to quantify the probability of human factor management errors in the calculation of dam break probability.

[0061] S6. The incentive variable weight method is used to vary the weights of the top-level indicators (reasons for dam break) in the calculation of dam break probability to increase the weights of weak indicators. The variable weight function is constructed as follows:

[0062]

[0063] In the formula, S i is the state variable weight value, α is a constant, and P i is the dam break probability value calculated by using the event tree method in combination with the dam break failure path. The above formula is to find the path p with the largest failure probability i,max and set a variable weight value greater than 1 for the dangerous path to amplify its influence.

[0064] The following formula is used to calculate and normalize the variable weight vector μ i :

[0065]

[0066] In the formula, is the initial constant weight value, which is taken as 1 when the priorities of each dam break reason are the same, and μ i is the calculated variable weight value. Since the weights of the incentive variable weight synthesis change with the probability value, after variable weighting, the priority value of the dangerous dam break path is higher, that is, the greater the probability of the dam break mode occurring, the greater the weight. In dam break accidents, the failure risk is mainly controlled by weak indicators. Among many dam break paths, as long as one breaks through the dangerous value, it will collapse. Therefore, the variable weight method that can amplify weak indicators can more scientifically reflect the failure probability.

[0067] Then, considering the variable weight method, the engineering dam break probability P i * is:

[0068]

[0069] S7. In the calculation of dam break probability in traditional risk analysis, only engineering factors are considered. The present invention introduces the relationship between human factor analysis and engineering dam break, and combines the variable weight method in S6 to calculate P f and constructs the formula:

[0070]

[0071] In the formula, P fThe failure probability after combining human error analysis for this project, including the probability of dam failure caused by multiple reasons such as encountering floods exceeding the standard, leakage and piping, earthquakes, etc.; P i The probability of dam failure caused by the i-th dam failure reason, which is the probability considering only engineering factors and is obtained by calculating the dam failure error path using the event tree method; μ i The variable weight vector calculated from the dam failure probabilities of different dam failure paths in S6; H i The total human error probability under the i-th dam failure reason, which is composed of various human factor influencing factors and the variable weight weight matrix W * Combined; h ij The human error probability of the j-th under the i-th dam failure reason; ω ij The weight of the j-th human error factor under the i-th dam failure reason.

[0072] Use β / (1 - H i ) as the amplification factor of the human error probability to reflect the dam failure probability calculated after considering human error, where β is a constant.

[0073] S8, combining the principles of reservoir dam risk management, introducing the idea of risk ranking, and using the comprehensive risk index R to rank the group of dams:

[0074] R = P f L

[0075] In the formula, P f Is the dam failure probability, and L is the comprehensive evaluation function of the dam failure consequence.

[0076] The dam failure consequence is calculated through the existing consequence severity evaluation model, including loss of life, economic loss, and social and environmental impacts.

[0077] A computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the above computer program, it implements the steps of the reservoir dam risk assessment method described above.

[0078] A computer-readable storage medium, which stores a computer program for executing the reservoir dam risk assessment method described above.

[0079] Beneficial effects: The present invention provides a reservoir dam risk assessment framework considering human reliability, constructs an index system considering human factors and its weights, and obtains a variable weight dam safety risk assessment method considering the consequences of failure, which can be used in engineering practice, can more scientifically and intuitively reflect the impact of human factor management on reservoir risk, reflect the risk level of the reservoir, provide a scientific basis for decision-making, and has a wide application prospect. Description of the Drawings

[0080] Figure 1 It is the Bayesian network diagram of the index system constructed in the embodiment of the present invention;

[0081] Figure 2 It is the result comparison diagram of various risk ranking methods in the embodiment of the present invention. Specific implementation manners

[0082] The following further clarifies the present invention in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. After reading the present invention, various equivalent modifications of the present invention by those skilled in the art all fall within the scope defined by the appended claims of this application.

[0083] A reservoir dam risk assessment method is proposed, including the following steps:

[0084] Taking 5 small reservoirs in a certain county as objects, they were all identified as second-class dams in the previous dam risk assessment and need to be rectified to meet the standards. Under the condition of limited financial funds, it is decided to study the order of appropriation and reinforcement. The evaluation method is used to evaluate the DS Reservoir, GQ Reservoir, TH Reservoir, TJ Reservoir, and LF Reservoir in this county.

[0085] Step 1: Construct a reservoir dam risk assessment index system

[0086] Construct a risk assessment index system based on reservoir human factors. The human factor failure risk model considering management personnel, technical facilities, management organizations, and environmental factors is:

[0087]

[0088] In the formula, H is the human factor failure risk value, and h ij is the probability of the jth human factor failure under the ith dam-break cause, and ω ij is the weight of the jth human factor failure factor under the ith dam-break cause.

[0089] In the embodiment of the present invention, the index system constituting the human factor failure H of the reservoir dam is derived from historical dam-break data, considering the influence of the operator himself, software and hardware facilities and the human-machine interaction between the operator, management organization, and operating environment on the management personnel. Introduce relevant factors of on-site management of operators, couple engineering factors and non-engineering factors, comprehensively analyze the influence of various factors in dam safety management, and improve the comprehensiveness of the indicators.

[0090] Step 2: Construct a directed acyclic graph of a human factor-driven reservoir dam risk assessment factor Bayesian network.

[0091] Based on historical dam break records and expert opinions, the dependency relationships of various influencing factors in the index system are mined, and a directed acyclic graph of the Bayesian network for risk assessment factors is constructed.

[0092] Step 3: Calculate the weight values of the index system.

[0093] Based on the AHP method and the improved CRITIC method, the subjective and objective weights of the index system are obtained from expert opinions and historical dam break data, and the weight combination is calculated using the game theory method.

[0094] 3.1 Use the Analytic Hierarchy Process (AHP) to analyze expert opinions and calculate the subjective weights of the indicators. Score and compare the factors at the indicator level, and the comparison criteria are shown in Table 1.

[0095] Table 1 Scoring Criteria for AHP

[0096] Scale Meaning 1 Indicates that when two elements are compared, they have the same importance 3 Indicates that when two elements are compared, the former is slightly more important than the latter 5 Indicates that when two elements are compared, the former is significantly more important than the latter 7 Indicates that when two elements are compared, the former is extremely more important than the latter 9 Indicates that when two elements are compared, the former is strongly more important than the latter 2,4,6,8 Represents the intermediate value of the adjacent judgments above The reciprocal of 1 to 9 Indicates the importance of comparing the corresponding two factors by exchanging their orders

[0097] Use the scoring results to construct the judgment matrix A, which is

[0098]

[0099] Use the eigenvector method to calculate the weight vector.

[0100] From the formula:

[0101] AW = λ max W

[0102] Calculate the maximum eigenvalue λ of the judgment matrix A max , which exists and is unique.

[0103] Calculate the weight matrix W, whose components are positive components, that is, W = (W 1 , W 2 , …, W n ) T .

[0104] Normalize the obtained weight vector W to get the desired subjective weight matrix W (1) .

[0105]

[0106] Perform a consistency test on the judgment matrix A according to the following steps, and calculate the consistency index CI and the consistency ratio CR:

[0107] CI = (λ max - n) / (n - 1)

[0108] CR = CI / RI

[0109] RI is the average random consistency index, and its values are shown in the following table. If CR < 0.1, the consistency of the judgment matrix can be considered acceptable; otherwise, the matrix needs to be corrected.

[0110] n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 RI 0 0 0.52 0.89 1.12 1.26 1.36 1.41 1.46 1.49 1.52 1.54 1.56 1.58 1.59

[0111] The calculation results of the subjective weights are shown in Table 2.

[0112] Table 2 Weights of various influencing factors calculated by AHP

[0113]

[0114] 3.2 Use the improved CRITIC method to calculate the objective weights of the indicators.

[0115] Analyze the historical dam-break database and score the indicator factors. For factors with specific values, such as the age of people and working years, they can be selected according to the actual situation and input after normalization. For factors without specific values and for which data cannot be determined, the judgment criteria are divided into several levels for assignment. Since most of the historical dam-break records lack specific data on dam management, including the age and employment time of management personnel, and at the same time, social environment factors are related to local government publicity and are difficult to determine. The main inputs of this study are factors without specific values, and a conversion table of influencing factors for human factor indicators is constructed, and the judgment criteria are selected according to Table 3.

[0116] Table 3 Conversion table of influencing factors for human factor indicators

[0117]

[0118] Classify the dam-break reservoirs according to the causes of dam breaks, including overtopping, seepage, and engineering quality problems. Among them, overtopping includes 45 historical dam-break accidents such as Baiguochong Reservoir, Taum Sauk Dam in the United States, and Pantano De Puentes Dam in Spain. There are 35 cases of dam breaks caused by seepage, including Bayi Reservoir, Phoenix Reservoir, and St. Francis Dam in the United States. There are 20 cases of dam breaks caused by engineering quality problems, including Baogu Reservoir, Qixianhu Reservoir, and Patel Dam in the United States.

[0119] According to the historical dam-break records, score the indicators P 1 , P 2 , …, P n in the risk assessment factor set to obtain the matrix:

[0120]

[0121] Standardize the data,

[0122]

[0123] Calculate the coefficient of variation v j :

[0124]

[0125] Calculate the index independence h j :

[0126]

[0127] Calculate the comprehensive measurement coefficient q of the index j :

[0128] q j =v j ·h j j = 1, 2,..., n

[0129] The weight ω of index j j is

[0130]

[0131] From the weight ω j Combine to obtain the objective weight matrix W (2) , as shown in Table 4

[0132] Table 4 Improved CRITIC calculates the weights of each influencing factor

[0133]

[0134] 3.3 Use game theory method to couple the subjective and objective weights

[0135] The weights obtained by the above subjective and objective methods generate a weight vector set W = {W (1) , W (2)}, and any linear combination of the weight vectors forms a possible weight set:

[0136] W = α 1 (W (1) ) T +α 2 (W (2) ) T

[0137] Optimize the linear combination weight coefficients α 1 、α 2 , and the goal is to minimize the deviation between W and each W (i) . That is

[0138]

[0139] According to the differential properties of the matrix, the first-order derivative condition for optimization of the above equation is:

[0140]

[0141] After obtaining (α 1 , α 2 ) from the above equation, perform normalization on it:

[0142]

[0143] Finally, the optimal combined weight is:

[0144]

[0145] Thus, the subjective weight matrix W (1) , the objective weight matrix W (2) are combined to form the weight W * , as shown in Table 5.

[0146] Table 5 Weights of Influencing Factors in the Index System

[0147]

[0148]

[0149] Step 4: Draw a Bayesian network diagram representing the reliability of the reservoir dam driven by human factors.

[0150] Based on the directed acyclic graph constructed in Step 2, the mutual relationships of each node are obtained through Bayesian inference using the index weights calculated in Step 3. Figure 1 It is a diagram of the reliability of human factor management in the scenario of overtopping and dam failure of a certain reservoir. The prior probabilities of each root node are determined by the reservoir management level and the failure scenario. Considering the scenario of overtopping due to floods exceeding the design standard, the physical environment of the reservoir is "not very good", and the reservoir management situation is "relatively standardized" obtained from safety evaluation and on-site inspection. According to the Bayesian network diagram, the evaluation values of relevant root indicators are obtained.

[0151] Step 5: Calculate the reliability of reservoir management

[0152] In this embodiment, combining the dam safety evaluation reports and on-site records of five reservoirs in a certain county, conduct index system scoring, and use the Bayesian network to calculate the management reliability under different dam failure reasons. The calculation results are listed in Table 6.

[0153] Table 6 Scoring of Human Factor Influencing Factors for Five Reservoirs

[0154]

[0155] Step 6: Use the variable weight method to conduct risk assessment on the five reservoirs.

[0156] The incentive variable weight method is used to vary the weights of the top-level indicators (reasons for dam break) in the calculation of dam break probability, so as to increase the weights of weak indicators. The variable weight function is constructed as follows:

[0157]

[0158] The variable weight vector μ is calculated using the following formula i :

[0159]

[0160] Then, the dam break probability of the project considering variable weights is:

[0161]

[0162] Step 7: Calculate the dam break probability driven by human factors according to the proposed risk assessment formula, and construct the formula:

[0163]

[0164] In the embodiment, the reasons for dam break include overtopping, seepage, and engineering quality problems. These are related to project safety and are directly obtained from the event tree analysis under the corresponding dam break modes. For example, for the path of "piping occurrence - continuous large seepage - development - no intervention - collapse / subsidence of the dam crest and breach", the failure probabilities of each node are mainly obtained through the combination of reservoir facilities and quantitative analysis.

[0165] In this embodiment, the total human error probability is calculated in Step 5. As shown in Table 6, its main source is the evaluation of the management agency in the risk assessment reports of each reservoir project. Among them, the LF reservoir with an independent management agency has the highest evaluation in terms of the reliability of human factor influencing factors, so the human error probability is the lowest.

[0166] Step 8: Introduce the dam break consequences and conduct risk assessment for five reservoirs.

[0167] In this embodiment, according to the design reports and relevant literature of the five reservoirs, the main social and environmental impacts are obtained, mainly including the social and environmental impacts after the dam break of the reservoirs, the irrigated area, the population protected downstream, cultivated land, transportation facilities, etc., which are listed in Table 7.

[0168] Table 7 Main Social and Environmental Impact Table of Five Reservoirs

[0169]

[0170] The dam break consequences are calculated using the dam break consequence severity evaluation model proposed by Li Lei, including loss of life, economic loss, and social and environmental impacts.

[0171] In this embodiment, based on the design reports of five reservoirs, regional economic development reports, agricultural output value reports and other data, the loss of life is calculated proportionally to the downstream population, and the economic loss takes into account direct and indirect losses. Among them, the average value of fixed assets is 6.6 million yuan / km 2 , the one-time inundation loss is 15%, and the average value of the total industrial and agricultural output is 4.5 million yuan / km 2 , the loss rate is 25%, and the value of the property of urban and rural households is 3.8 million yuan / km 2 , and the loss rate is 20%. The consequences of the failure of each reservoir calculated are listed in Table 8.

[0172] Table 8 Comprehensive evaluation coefficient table of the dam-break consequences of five reservoirs

[0173]

[0174] Step Nine: Use the comprehensive risk index R to rank the group of dams:

[0175] R = P f L

[0176] In the formula, P f is the dam-break probability after amplifying the most dangerous path by the variable weight method and considering the influence of human factor management, and L is the comprehensive evaluation function of the dam-break consequences.

[0177] The calculation results obtained in Steps Seven to Nine are shown in Table 9, including the constant weight dam-break probability without considering amplification, the human factor failure probability, the constant weight calculation failure probability considering the human factor failure amplification coefficient, the variable weight calculation failure probability considering the human factor failure amplification coefficient, and the risk assessment considering the dam-break consequences. The calculation results are as Figure 2 shown.

[0178] Table 9 Dam-break probability table of five reservoirs

[0179]

[0180] Result analysis:

[0181] When only considering engineering factors, the risk ranking of the reservoirs is TH > GQ > TJ > DS > LF. This result is directly obtained from the event tree analysis under three dam-break modes and only reflects engineering safety. For the TH reservoir with poor on-site facility performance and few inspections, the probability of piping occurrence is higher and the probability of timely detection is small, so the dam-break probability is higher.

[0182] When only considering the probability of human error, the reservoir risk ranking is TJ > TH > DS > GQ > LF. This result is obtained from the evaluation of the management agencies in the risk assessment reports of each reservoir project. The LF reservoir has an independent management agency, with the highest human reliability evaluation and the lowest error probability. After superimposing the human factor influencing factors on the engineering factors, the order of the GQ reservoir and the TJ reservoir changes. The human error probabilities of the two reservoirs are 0.3477 and 0.6108 respectively. After introducing the human factor coefficient into the engineering dam-break probability, the dam-break probability of the GQ reservoir is amplified from 1.27E-05 to 0.96, and the dam-break probability of the TJ reservoir is amplified from 1.41E-05 to 1.04. This is because the TJ reservoir lacks a dedicated management department and its management system is not perfect enough. The occurrence of dam-break accidents is not only affected by engineering factors, but also includes the people on the reservoir site management. The human factor coefficient introduces the operator factor into the calculation of the dam-break probability, reflecting the unreliability of safety management.

[0183] Using the variable weight method to conduct risk assessment on the five reservoirs, it can be seen from the calculation results that the dam-break probability results after variable weighting are larger than those of the constant weight synthesis. The introduction of variable weights has the most significant impact on the LF and DS reservoirs, and their dam-break probabilities change from 0.82 and 0.85 to 1.11 and 1.05 respectively. The probability of dam-break caused by overtopping of the LF reservoir is much higher than other situations, while the dam-break probabilities in various situations of the DS reservoir are relatively close. Using the variable weight method to amplify the weight of the overtopping failure probability of the LF reservoir can more scientifically characterize the dangerous situation of the reservoir.

[0184] After considering the consequences of dam-break, the reservoir risk ranking is TH < GQ < TJ < LF < DS. Compared with the risk assessment without considering the consequences of dam-break, the main changes occur in the GQ and TJ reservoirs. This is because the GQ reservoir protects a large number of people, and the life and economic losses after dam-break are serious, so the consequences of dam-break are serious. Among these five reservoirs, the LF reservoir has the most serious consequences of dam-break, but it is equipped with a dedicated management agency and the management personnel are regularly trained, which greatly improves the reliability of the management personnel and reduces the dam-break risk.

[0185] To sum up, this method combines the reservoir management level with the engineering risk, and enhances the attention to human factors and dangerous dam-break paths. It introduces human factors into the dam risk assessment and constructs a reservoir human factor index system. Based on the combined weighting method, the weights of human factor indicators are assigned, and a Bayesian network is established accordingly. Using the incentive-type variable weight method to conduct variable weighting on the top-level failure modes to amplify the dominant failure modes can more accurately reflect the danger of weak indicators in dam-break.

[0186] A dam-break risk calculation formula driven by human factors is proposed, and based on the risk management concept, the accident consequences are introduced, and a risk assessment method that comprehensively considers the probability of human error, the consequences of dam-break, and the engineering dam-break probability is proposed, providing a method support for the risk assessment of reservoir groups considering human factors.

[0187] Engineering applications show that the proposed risk assessment method can accurately express the impact of management factors on reservoir risks, intuitively reflect the risk levels of different reservoirs, and provide a scientific basis for the management decision-making of reservoir groups.

[0188] Obviously, those skilled in the art should understand that each step of the reservoir dam risk assessment method in the above embodiments of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Optionally, they can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described can be executed in a different order than here, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module to be implemented. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

Claims

1. A reservoir dam risk assessment method, characterized in that: The steps include: S1, based on historical dam break data information, a reservoir dam risk assessment factor set is formed, a reservoir dam risk assessment indicator system is constructed, and a Bayesian network relationship diagram is generated; S2, using the AHP method to calculate the subjective weight matrix; S3, using the improved CRITIC method and combining historical dam-break data to calculate objective weights; S4, using game theory methods, performs a weighted combination of the subjective and objective weights obtained in S2 and S3; S5, construct a Bayesian network based on the obtained indicator weights to form a Bayesian network diagram of human factor indicators to quantify the probability of human management errors in dam failure probability calculation; S6, introduce variable weight method into the calculation of accident probability, use incentive variable weight method to change the weight of top indicators of dam break probability calculation, construct new variable weight function and combine it with weight calculation in dam break calculation, and normalize it; S7, introduces the probability of human management error obtained in S5, combines it with the engineering dam failure probability, and uses the variable weight coefficient obtained in S6 to change the weight of the dam failure probability under different dam failure paths to feedback the dam failure probability after considering human error and variable weight; S8, introduce the idea of ​​risk ranking, and rank the dam groups based on the consequences of dam failure to determine the risk level of the reservoir group and provide guidance for reservoir capital investment and management and operation.

2. The reservoir dam risk assessment method according to claim 1, characterized in that: The specific implementation process of S2 is: S2.1 compares and scores the factors at the indicator layer; for the indicator layer, the judgment matrix A is constructed using the scoring results. Where n is the number of dimensions, that is, the number of indicators in the indicator layer, a ij It is an expression of the importance between indicator i and indicator j, and the score is given according to the importance of indicator i relative to indicator j; S2.2, use the eigenvector method to calculate the weight vector W, first calculate the maximum eigenvalue λ of the judgment matrix A max , according to AW = λ max W, calculate the weight matrix W, whose components are positive components W i , indicating the index a i The weight of W = (W1, W2, ..., W n ) T ; Normalizing the obtained weight vector W gives the required subjective weight matrix W (1) , where each component W i (1) The calculation method is: S2.3, perform consistency check on the judgment matrix A according to the following steps: CI=(λ max -n) / (n-1) CR=CI / RI Among them, CI is the consistency index, RI is the average random consistency index; CR is the consistency ratio. If CR<0.1, the consistency of the judgment matrix is ​​considered acceptable; otherwise, the judgment matrix needs to be modified.

3. The reservoir dam risk assessment method according to claim 1, characterized in that: The implementation process of S3 is as follows: S3.1, based on the historical dam failure records, the indicators of risk assessment factors are scored to obtain the matrix: where b ij It represents the jth indicator score of the i-th dam in the historical dam failure record; S3.2, data standardization processing: Among them, b j ′ is the matrix generated after standardization, b j is the column matrix corresponding to index j, maxb j , minb j The column matrices b are j The maximum and minimum values ​​in ; S3.3, calculate the coefficient of variation v j : in, is the arithmetic mean of the column matrix of index j, b ij is the score of the i-th reservoir of index j, σ j is the standard deviation of the indicator, S3.4, calculation of index independence h j : Among them, |r jl | is the absolute value of the correlation coefficient between indicator j and indicator l; S3.5, calculate the comprehensive measurement coefficient q of the indicator j : q j =v j ·h j j=1,2,…,n S3.6, weight ω of indicator j j for By weight ω j Combined to get the objective weight matrix W (2) .

4. The reservoir dam risk assessment method according to claim 1, characterized in that: S4, using game theory methods, the subjective weight matrix W is obtained (1) , objective weight matrix W (2) Perform weighted combination; S4.1, before performing the weighted combination, the obtained weighting method is firstly checked for consistency using a distance function; the distance function is: In the formula, is the subjective weight of indicator j, is the objective weight of indicator j; d(W (1) ,W (2) ) is smaller, the two weighted results are closer. (1) ,W (2) )≤1, the consistency requirement is met; otherwise, the consistency requirement is not met and the weight needs to be recalculated; S4.2, using game theory to combine the subjective and objective weights obtained by S2 and S3; first, the subjective and objective weights obtained by S2 and S3 generate a weight vector set W = {W (1) ,W (2) }, these two weight vectors can be arbitrarily linearly combined into a possible weight set: W=α1(W (1) ) T +α2(W (2) ) T Where W (1) is the subjective weight calculated by S2, W (2) is the objective weight calculated by S3; α1, α2 are the linear combination coefficients of the main and objective weights; the whole of W is the set of all possible weight vector combinations; S4.3, for the linear combination weight coefficient α k The goal of the optimization is to make W and W (1) , W (2) The deviation of is minimized; that is According to the differential properties of the matrix, the optimal first-order derivative condition of the above formula is: S4.4, after obtaining (α1, α2) according to the above formula, normalize it: S4.5, the optimal combination weight is finally: Thus, the subjective weight matrix W is obtained (1) , objective weight matrix W (2) The combined weight W * .

5. The reservoir dam risk assessment method according to claim 1, characterized in that: S6, using the incentive variable weight method to change the weight of the top indicators of dam break probability calculation to increase the weight of weak indicators, the variable weight function is constructed as follows: In the formula, S i is the state variable weight, α is a constant, P i is the dam failure probability value calculated by using the event tree method combined with the dam failure path; the above formula is to find the path p with the highest failure probability i,max And set a variable weight greater than 1 for the dangerous path to amplify its impact; The following formula is used to calculate and normalize the variable weight vector μ i : In the formula, is the initial constant weight, which is 1 when the priorities of the dam failure causes are the same, μ i is the calculated variable weight value. Since the weight of the incentive-based variable weight synthesis changes with the probability value, after the weight change, the priority value of the dangerous dam break path is higher, that is, the greater the probability of the dam break mode, the greater the weight; The probability of dam failure P considering the variable weight method i * for:

6. The reservoir dam risk assessment method according to claim 1, characterized in that: S7, introduces the relationship between human factor analysis and engineering dam failure, combines the variable weight method of S6, and analyzes P f Perform calculations and construct formulas: Where P f P is the failure probability of the project after combining human error analysis; i is the probability of dam failure caused by the i-th dam failure cause, is the probability under only engineering factors, and is calculated by the event tree method for the dam failure path; μ i is the variable weight vector calculated from the dam failure probabilities of different dam failure paths in S6; H i is the total human error probability under the i-th dam failure cause, which is composed of each human factor influencing factor and the variable weight matrix W * Combination composition; h ij is the probability of the jth person making a mistake under the i-th dam failure cause; ω ij is the weight of the jth individual’s error factor under the i-th dam failure cause; Use β / (1-H i ) is used as the amplification factor of the probability of human error to feedback the dam failure probability calculated after considering human error, and β is a constant.

7. The reservoir dam risk assessment method according to claim 1, characterized in that: S8, combined with the risk management principle of reservoir dams, introduces the idea of ​​risk ranking and uses the comprehensive risk index R to rank the dam groups: R=P f L Where P f is the probability of dam failure, and L is the comprehensive evaluation function of dam failure consequences.

8. A computer device, characterized in that: The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the reservoir dam risk assessment method as described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the reservoir dam risk assessment method according to any one of claims 1 to 7.

Images