Risk assessment method for supercritical carbon dioxide waste heat power generation system based on dynamic aging model

Abstract

The invention relates to a supercritical carbon dioxide waste heat power generation system risk assessment method based on a dynamic aging model, and belongs to the technical field of power generation system risk assessment. The invention relates to dynamic risk identification, evaluation and early warning for a supercritical carbon dioxide Brayton cycle system with high temperature, high pressure and variable working conditions. The method comprises the following steps: (1) risk identification; (2) performing multi-dimensional risk assessment; (3) risk comprehensive evaluation; and (4) risk early warning and emergency response. The method effectively reduces subjectivity and equal-weight hypothesis deviation, has adaptive adjustment ability along with states and environments, gives consideration to multiple objectives of technology, economy and environments, and can perform early warning and support maintenance cost optimization. Experiments show that compared with static TOPSIS, the method provided by the invention has the advantages of higher scoring stability, lower prediction error and obviously reduced false alarm rate, and is suitable for dynamic risk management of S-CO2 waste heat power generation systems with different scales and variable working conditions.

Description

Technical Field

[0001] This invention relates to a risk assessment method for supercritical carbon dioxide waste heat power generation systems based on a dynamic aging model. It is particularly applicable to the dynamic risk assessment and early warning management of supercritical carbon dioxide (S-CO2) Brayton cycle waste heat power generation systems and belongs to the field of power generation system risk assessment technology. Background Technology

[0002] With the increasing industrial application of S-CO2 waste heat power generation technology, system reliability and risk management have become key constraints for the large-scale promotion of this technology. S-CO2 waste heat power generation systems operate under high temperature and high pressure conditions, and key components (such as compressors, turbines, and regenerators) will undergo a dynamic aging process during long-term operation, leading to system performance degradation and increased failure risk.

[0003] Traditional risk assessment methods have significant limitations when applied to S-CO2 systems: their Failure Mode and Effects Analysis (FMEA) process relies on subjective expert scoring and assumes equal weighting of risk factors, failing to reflect the actual importance of each factor; they also use fixed thresholds to classify risk levels, lacking flexibility. Furthermore, the traditional TOPSIS (Topology for Approximating Ideal Solutions) method uses fixed weights and ideal solutions, failing to handle uncertainties caused by aging and the environment, and its assessments are often isolated, lacking time-series analysis capabilities. In addition, traditional methods often focus on a single dimension, making it difficult to comprehensively consider the combined impact of technical, economic, and environmental risks, and thus failing to meet the assessment needs of complex dynamic systems. Summary of the Invention

[0004] The technical problem to be solved by this invention is: how to construct a risk assessment framework that integrates dynamic aging model and multi-criteria decision-making, and solve the defects of static and single-dimensional risk in traditional methods through three-dimensional quantitative analysis of technology, economy and environment, so as to achieve dynamic, accurate and robust assessment and early warning of the risks of S-CO2 waste heat power generation system.

[0005] To address the aforementioned technical problems, this invention provides a risk assessment method for SCO2 waste heat power generation systems based on a dynamic aging model, comprising the following steps:

[0006] Step S1: Risk Identification

[0007] Step S1.1: Construct the time-varying fuzzy judgment matrix

[0008] In response to the special working environment and failure mechanism of S-CO2 waste heat power generation system, a seven-level fuzzy language evaluation system was designed, which maps the qualitative language evaluation of experts into triangular fuzzy numbers, effectively solving the problem of strong subjectivity in traditional FMEA evaluation.

[0009] Suppose n experts evaluate the risk factor j of failure mode i, and the weighted average result of their fuzzy evaluation set is:

[0010]

[0011] in, For expert k, a fuzzy evaluation of failure mode i and risk factor j; w k The weight of expert k depends on the normalization condition met by their professional background, work experience, and professional field.

[0012] Step S1.2: Calculate dynamic weights

[0013] The risk factor weights were determined by combining the Analytic Hierarchy Process (AHP) with the entropy weight method, and a dynamic weight adjustment mechanism was designed.

[0014] First, construct the judgment matrix using the AHP method:

[0015]

[0016] Among them, a ij To represent the importance of risk factor i relative to factor j, Saaty's 1-9 scale is used:

[0017] a ij =1: Factors i and j are equally important;

[0018] a ij =3: Factor i is slightly more important than j;

[0019] a ij =5: Factor i is significantly more important than j;

[0020] a ij =7: Factor i is significantly more important than j;

[0021] a ij =9: Factor i is extremely important than j;

[0022] a ij =2,4,6,8: The intermediate values ​​of adjacent judgments;

[0023] a ji =1 / a ij Inverse relationship.

[0024] Calculate the eigenvalues ​​and eigenvectors, and calculate the largest eigenvalue λ of the judgment matrix A. max and the corresponding eigenvector w AHP =[w S ,w O ,w D ] T .

[0025] Secondly, collect historical fault dataset X = [x ij [m×3, standardized process:]

[0026]

[0027] Calculate information entropy:

[0028]

[0029] Calculate entropy weights:

[0030]

[0031] Obtain the entropy weight vector:

[0032] w 熵 =[w 熵,S ,w 熵,O ,w 熵,D ] T

[0033] The subjective weights obtained from AHP and the objective weights obtained from the entropy weight method are then weighted and fused together.

[0034] W(0) = β·w AHP +(1-β)·w 熵

[0035] Where β∈[0,1] is the adjustment coefficient.

[0036] To adapt to the dynamic changes in the operating status and environmental conditions of the S-CO2 system, a dynamic weight adjustment mechanism was designed:

[0037] W(t)=Softmax(W(0)+ΔW 老化 (t)+ΔW 环境 (t))

[0038] Among them, the aging adjustment item is:

[0039]

[0040] Where, k j Let be the aging impact coefficient of risk factor j; Let be the rate of change of the aging parameters of component i.

[0041] Environmental adjustment items:

[0042]

[0043] Among them, E l (t) represents the value of the environmental parameter l (such as temperature, pressure, etc.) at time t; E l,normal The normal value for environmental parameter l; El,max The limiting value of environmental parameter l; θ jl Let l be the coefficient of influence of environmental parameter l on risk factor j.

[0044] Step S1.3: Calculate the dynamic risk priority number

[0045] Design a nonlinear combination-based RPN calculation method:

[0046]

[0047] Among them, W j (t) represents the dynamic weight of risk factor j at time t; α j (t) is the nonlinear adjustment exponent; K is the normalization coefficient.

[0048] Step S1.4: Risk Level Clustering

[0049] The risk level adaptive classification is performed using the FCM (Fuzzy C-means) algorithm, based on minimizing the objective function:

[0050]

[0051] Among them, u ij v represents the membership degree of sample i to cluster j. j Let m represent the j-th cluster center, and m be the ambiguity parameter.

[0052] Step S2: Multidimensional Risk Assessment

[0053] Step S2.1: Technology Risk Assessment

[0054] System technical risk is defined as:

[0055] R T,i (t)=S i ×(1-R i (t))

[0056] Among them, R T,i (t) represents the technical risk index of the i-th component at time t; S i R represents the severity of the primary failure mode of the i-th component; i (t) represents the reliability of the main failure mode of the i-th component.

[0057] Step S2.2: Economic Risk Assessment

[0058] Economic risk quantification model:

[0059] R E,i (t)=O i (t)×[C m,i +C d,i ×Tdoen,i ]

[0060] C d,i =P rated ·η loss ·p elec

[0061] Among them, R E,i (t) represents the expected economic risk loss of the i-th component at time t; O i (t) represents the occurrence degree of the main failure mode of this component at time t; C m,i For the direct repair cost of this failure mode; C d,i P represents the power generation loss per unit time caused by system downtime due to the failure of this component; rated Rated power; η loss T is the efficiency loss coefficient caused by shutdown or reduced load operation; doen,i This represents the mean time to repair for this failure mode.

[0062] Step S2.3: Environmental Risk Assessment

[0063] Environmental risk quantification model:

[0064] R V,i (t)=O i (t)×V leak,i ×GWP co2 ×C carbon ×C penalty

[0065] Among them, V leak,i GWP represents the expected working fluid leakage rate when this failure mode occurs. co2 C represents the global warming potential of carbon dioxide. carbon The carbon cost per unit of carbon dioxide equivalent emissions, C penalty This refers to the fixed fines or emergency response costs that may be incurred due to environmental pollution incidents.

[0066] Step S3: Comprehensive Risk Assessment

[0067] Step S3.1: Construct the dynamic decision matrix

[0068] At each assessment time t, real-time data is collected and integrated with the technology risk index R. T (t), Economic Risk Index R E (t) and environmental risk index R V (t), construct the dynamic decision matrix X(t):

[0069]

[0070] Where rows i = 1, 2, 3, ..., m represent m evaluation schemes, and columns j = 1, 2, 3 represent the three risk dimensions of technology, economy, and environment, respectively. This is calculated from the corresponding sub-model.

[0071] Step S3.2: Matrix Standardization and Entropy Weight Calculation

[0072] The decision matrix is ​​standardized using the extreme value method. For cost-related indicators:

[0073]

[0074] Where, x ij (t) is the element in the i-th row and j-th column of the decision matrix X(t), representing the original evaluation value of the i-th evaluation object at time t in the j-th risk dimension; r ij (t) represents the element in the i-th row and j-th column of the standardized decision matrix R(t), representing x. ij (t) is the value after standardization.

[0075] Calculate the information entropy of each dimension at time t to measure its uncertainty:

[0076]

[0077] Among them, E j (t) represents the information entropy of the j-th evaluation criterion at time t; r ij (t) represents the proportion in this criterion list.

[0078] Calculate the dynamic weights for each dimension:

[0079]

[0080] in, Let be the objective weight of the j-th criterion, which is entirely data-driven at time t, calculated using the information entropy method. Its value is inversely proportional to the information entropy of that criterion.

[0081] To integrate expert prior knowledge, dynamic weighted fusion is performed:

[0082]

[0083] Among them, w j (t) represents the final fusion weight of the j-th criterion at time t; Let be the prior static weights for the j-th criterion. This mechanism allows the weights to reflect both long-term stable risk preferences and adaptively respond to changes in short-term data uncertainty.

[0084] Step S3.3: Constructing the weighted decision matrix and updating the dynamic ideal solution

[0085] Construct the weighted decision matrix V(t):

[0086] v ij (t)=w j (t)·r ij (t)

[0087] Where V(t) is the weighted normalized decision matrix at time t, with dimensions m×3; v ij (t) are elements in the weighted matrix V(t).

[0088] Determination of the dynamic ideal solution:

[0089]

[0090] Where β is the fusion coefficient, used to balance real-time observations with state-based expected values; Let be the conditional expectation function, representing the expected value of the ideal solution under the given system aging state α(t) and control input u(t) at time t; α(t) is the health state vector of the system or component output by the dynamic aging model at time t; u(t) is the control input vector of the system at time t.

[0091] Step S3.4: Distance and Proximity Calculation

[0092] Calculate the Euclidean distance between each evaluation object and the dynamic positive and negative ideal solutions:

[0093]

[0094] Then calculate the relative closeness of each object to the ideal solution:

[0095]

[0096] in, Let be the Euclidean distance from the i-th evaluation object to the dynamic positive ideal solution at time t; Let C be the Euclidean distance from the i-th evaluation object to the dynamic negative ideal solution at time t; i (t) represents the relative proximity of the i-th evaluation object at time t.

[0097] Step S3.5: Risk Prioritization, Forecasting, and Alerts

[0098] According to C i The risk priority at the current moment is determined by ranking all assessed objects based on the (t) value. The proximity within a future time interval Δt is then predicted based on the system state transition equation.

[0099]

[0100] in, Let f(·) be the predicted closeness of the i-th evaluation object at a future time t+Δt; f(·) is the prediction function; and a risk threshold ζ is set. If This will trigger a high-risk alert, prompting operations and maintenance personnel to intervene in the object.

[0101] Step S4: Risk Warning and Emergency Response

[0102] Based on the actual and predicted values ​​of risk proximity, the early warning levels are divided into three levels:

[0103] Table 1. Tiered Early Warning Mechanism Based on Risk Proximity

[0104]

[0105] Step S4.2: Maintenance Cost Optimization

[0106] Based on the risk assessment results, we can achieve precise allocation and optimization of maintenance resources, and follow the principle of "high risk, high investment" to prioritize the allocation of limited maintenance budget to the components with the highest risk level.

[0107] Compared with the prior art, the present invention has the following beneficial effects:

[0108] By introducing a seven-level fuzzy language evaluation system and triangular fuzzy numbers, the qualitative language evaluation of experts is mapped into quantitative fuzzy numbers, effectively solving the problem of strong subjectivity in traditional FMEA evaluation.

[0109] The risk factor weights are determined by combining the analytic hierarchy process (AHP) and the entropy weight method. A dynamic weight adjustment mechanism is designed to adaptively adjust the weights according to changes in system aging and environmental conditions, overcoming the irrationality of the assumption of equal weights for risk factors in traditional FMEA.

[0110] Using the FCM algorithm for adaptive risk level classification avoids the lack of flexibility of traditional fixed threshold classification methods and can adapt to different system characteristics and dynamic changes.

[0111] The S-CO2 waste heat power generation system was comprehensively evaluated from three dimensions: technology, economy, and environment. This evaluation fully reflects the complex risks faced by the system in actual operation and overcomes the limitations of traditional single-dimensional risk assessment.

[0112] We propose an Adaptive Dynamic TOPSIS (AD-TOPSIS) method, which introduces real-time data stream driving, information entropy uncertainty measurement, Bayesian update mechanism and dynamic weight adjustment to realize the temporal evolution and adaptive optimization of risk assessment results, thus solving the limitations of traditional static TOPSIS assessment.

[0113] By introducing predicted values ​​as early warning trigger conditions, advance warnings are achieved, providing the operations and maintenance team with a valuable preparation time window and transforming passive emergency response into proactive intervention.

[0114] Case studies show that the AD-TOPSIS method achieves a scoring stability of 0.87, significantly higher than the traditional TOPSIS's 0.73; the prediction RMSE is 0.089, a 42.9% reduction compared to the traditional TOPSIS's 0.156; and the false alarm rate is 3.2%, compared to the traditional TOPSIS's 5.8%, effectively reducing the false alarm rate while maintaining sensitivity.

[0115] Based on the assessment results, a tiered early warning and emergency response mechanism was designed, forming a closed-loop management system from risk perception to decision-making and execution, providing an effective tool for the safe and economical operation of the system. Attached Figure Description

[0116] Figure 1 This is a flowchart of the risk assessment method of the present invention;

[0117] Figure 2 This is a flowchart comparing the improved FMEA method of this invention with the traditional FMEA method.

[0118] Figure 3 This is the PCA projection clustering diagram of the present invention;

[0119] Figure 4 This is a box plot showing the distribution of membership degrees in this invention;

[0120] Figure 5 This is a convergence curve of the objective function of this invention;

[0121] Figure 6 This is a flowchart of the AD-TOPSIS algorithm of the present invention;

[0122] Figure 7 This is the multi-dimensional risk radar chart of the present invention;

[0123] Figure 8 This is a time-series evolution diagram of the risk proximity of the present invention;

[0124] Figure 9 This is a dynamic change diagram of the three-dimensional risk weights of this invention;

[0125] Figure 10 This is an emergency response flowchart for the present invention. Detailed Implementation

[0126] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0127] The overall technical solution of this invention is a risk assessment method for supercritical carbon dioxide waste heat power generation systems based on a dynamic aging model, comprising the following steps:

[0128] Step S1: Risk identification, including constructing a time-varying fuzzy judgment matrix, calculating dynamic weights, calculating dynamic risk priority numbers, and clustering risk levels;

[0129] Step S2: Multidimensional risk assessment, including technological risk assessment, economic risk assessment and environmental risk assessment;

[0130] Step S3: Comprehensive risk assessment, using the adaptive dynamic TOPSIS method for comprehensive evaluation;

[0131] Step S4: Risk warning and emergency response, including a tiered warning mechanism and maintenance cost optimization.

[0132] Example 1

[0133] A risk assessment was conducted using R Company's 30MWS-CO2 waste heat power generation system as the specific analysis object.

[0134] 1. Parameter settings

[0135] This step provides the initial conditions for the entire evaluation process. The research objects are the compressor unit, turbine unit, and heat exchanger unit. Risk dimensions include technical risk, economic risk, and environmental risk. The time frame is 180 months (from the 37th month to the 216th month after commissioning). The learning rate η = 0.7, the fusion coefficient β = 0.6, and the risk threshold ζ = 0.35. Prior weights W... static =[0.5,0.3,0.2] T .

[0136] 2. Risk Identification

[0137] This step corresponds to step S1 and... Figure 2 First, five experts were organized to conduct fuzzy language evaluations of the severity (S), occurrence (O), and detectability (D) of the nine failure modes. For example, the severity (S) of "compressor seal leakage" was mapped from the expert evaluations to triangular fuzzy numbers, and then calculated using the formula... The weighted average yields a clarity number S = 7.2, and the remaining S, O, and D values ​​are obtained through the same process.

[0138] Secondly, the S, O, and D values ​​in Table 2, as well as the W calculated by the dynamic weight formula in S1.2, are used. S (t), W O (t) and W D Substituting (t) into the dynamic RPN calculation formula in S1.3, we obtain the value of RPN(t) in Table 2.

[0139] Finally, the RPN(t) values ​​of all failure modes at each time point are used as samples and input into the FCM clustering algorithm formula in S1.4. After iteration, the algorithm automatically classifies the risk into three levels: "high", "medium", and "low". Figure 3 This demonstrates the clustering of samples in the reduced-dimensional space; Figure 4 This shows the degree of ambiguity in which each sample belongs to different risk levels; Figure 5 The convergence of the FC M algorithm was proven. The RPN range corresponding to the final cluster centers determined the "risk level" in Table 2.

[0140] Table 2. Results of Key Component Failure Mode Risk Identification Based on Improved FMEA

[0141]

[0142] 3. Multidimensional risk assessment

[0143] This step corresponds to step S2 and provides sub-quantitative indicators for the comprehensive evaluation. From the three dimensions of technology (C1), economy (C2), and environment (C3), the components belonging to the four selected key failure modes are quantified as alternative solutions for the comprehensive risk evaluation.

[0144] First, based on the mechanisms of the main failure modes of each component, technical, economic, and environmental risk models are applied for calculation.

[0145] Taking the technical risk of compressor (A1) as an example: the severity of its main failure mode "sealing leakage" is S = 7.2. Using the reliability model, its reliability at time t is calculated to be R(t) = 0.86. Substituting this into the formula in S2.1, R... T (t) = 7.2 × (1 - 0.86) = 1.008. Normalizing this value to the [0,1] interval, we obtain the technical score C1 = 0.97 in Table 3. The other scores are calculated according to this process, and the results are shown in Table 3.

[0146] Finally, the data in Table 3 are plotted as follows: Figure 7 This intuitively reveals the differences in the dimensions of risk composition for each component, such as the prominent technical risks of the compressor and the significant environmental risks of the low-temperature regenerator.

[0147] Table 3 Original Risk Indicator Data

[0148]

[0149] 4. Comprehensive Risk Assessment

[0150] This step is the core of the method, corresponding to step S3 and... Figure 6First, at each time step t, the scores of the four components across three dimensions are used to form a matrix X(t). This matrix is ​​then standardized using the extremum method formula in S3.2 to obtain R(t). Subsequently, the information entropy E of each criterion column is calculated. j (t) and entropy weight

[0151] Secondly, at time t=1, assume the calculated entropy weights are [0.45, 0.32, 0.23]. Using the prior weights W_static = [0.5, 0.3, 0.2] and the fusion coefficient η = 0.7, the final weights are calculated according to the fusion formula in S3.2: w j (1) = 0.7*[0.5,0.3,0.2] + 0.3*[0.45,0.32,0.23] = [0.485,0.306,0.209]. This weight changes over time, and its evolution trend is as follows: Figure 9 middle.

[0152] Then, use w j Multiplying (t) by the normalized matrix R(t) yields the weighted matrix V(t). The dynamic positive ideal solution V + (t) is generated by fusing the real-time maximum value with the expected value predicted based on the aging state α(t).

[0153] Finally, the Euclidean distance from each scheme to the dynamic positive / negative ideal solution is calculated. and Then follow the formula in S3.3 Calculate the closeness. (C) i The smaller (t) is, the higher the risk.

[0154] Table 4 shows the average closeness and ranking of AD-TOPSIS and traditional TOPSIS calculations over the entire time window. AD-TOPSIS, due to its dynamic adjustment of weights and ideal solutions, has a lower closeness estimate for high-risk signals and is more sensitive. Figure 8 The proximity C of the four components was plotted over 180 time steps. i The (t) curve visually illustrates the dynamic fluctuation pattern of its risk.

[0155] Table 4 Comparison of evaluation results between AD-TOPSIS and static TOPSIS

[0156]

[0157] The final risk level ranking is: turbine > compressor > low-temperature regenerator > high-temperature regenerator.

[0158] 5. Risk Warning

[0159] This step corresponds to step S4, translating the evaluation results into action. First, based on Table 1, combined with... Figure 8 Real-time proximity C i (t) and its predicted value trigger an early warning. For example, if the high-temperature regenerator at t=70°C... i (t) = 0.38 and the predicted value C i If (t+5) = 0.33, a "Level II Warning" is triggered, requiring enhanced monitoring to be deployed within one hour. Secondly, once the warning is triggered, [the system] will be activated. Figure 10 The closed-loop management process shown ensures rapid coordination from risk perception to response and feedback.

[0160] 6. Maintenance cost optimization

[0161] Based on the risk assessment results, the following maintenance cost allocation ratio is recommended:

[0162] Compressor unit: 40% (the core of energy conversion; failure results in the greatest economic loss);

[0163] Heat exchanger unit: 35% (withstanding high temperature and pressure, the risk of working fluid leakage is serious);

[0164] Turbine unit: 15% (high-speed rotating components, reliability is critical);

[0165] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0166] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0167] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the scope of the present invention should be included within the protection scope of the present invention.

Claims

1. A risk assessment method for supercritical carbon dioxide waste heat power generation systems based on a dynamic aging model, characterized in that, Includes the following steps: Step S1: Risk identification, including constructing a time-varying fuzzy judgment matrix, calculating dynamic weights, calculating dynamic risk priority numbers, and clustering risk levels; Step S2: Multidimensional risk assessment, including technological risk assessment, economic risk assessment and environmental risk assessment; Step S3: Comprehensive risk assessment, using the adaptive dynamic TOPSIS method for comprehensive evaluation; Step S4: Risk warning and emergency response, including a tiered warning mechanism and maintenance cost optimization.

2. The method according to claim 1, characterized in that, The specific method for constructing the time-varying fuzzy judgment matrix in step S1 is as follows: The qualitative language evaluation of experts is mapped to triangular fuzzy numbers to construct a seven-level fuzzy language evaluation system; Suppose n experts evaluate the risk factor j of failure mode i, and the weighted average result of their fuzzy evaluation set is: in, For expert k, a fuzzy evaluation of failure mode i and risk factor j; w k The weight of expert k depends on the normalization condition met by their professional background, work experience, and professional field. To reflect the changes in risk assessment during the dynamic aging process of the system, a time-varying judgment matrix update model is constructed: Among them, A ij (t) represents the fuzzy evaluation of risk factor j by failure mode i at time t, α i (t) represents the aging parameter value of component i at time t, γ j denoted as the aging sensitivity coefficient of risk factor j.

3. The method according to claim 1 or 2, characterized in that, The specific method for calculating the dynamic weights in step S1 is as follows: The risk factor weights are determined by combining the Analytic Hierarchy Process (AHP) and the entropy weight method, and the subjective weights obtained from the AHP are weighted and integrated with the objective weights obtained from the entropy weight method. W(0)=β·w AHP +(1-β)·w 熵 Where β∈[0,1] is the adjustment coefficient. To adapt to the dynamic changes in the operating status and environmental conditions of the supercritical carbon dioxide system, a dynamic weight adjustment mechanism was designed. W(t)=Softmax(W(0)+ΔW 老化 (t)+ΔW 环境 (t)) Among them, the aging adjustment item is: k j Let be the aging impact coefficient of risk factor j; The rate of change of aging parameters for component i; Environmental adjustment items: E l (t) represents the value of the environmental parameter l (such as temperature, pressure, etc.) at time t; E l,normal The normal value for environmental parameter l; E l,max The limiting value of environmental parameter l; θ jl Let l be the coefficient of influence of environmental parameter l on risk factor j.

4. The method according to claim 3, characterized in that, The specific method for calculating the dynamic risk priority number in step S1 is as follows: A nonlinear combination-based RPN calculation method was designed: in, It is a non-linear adjustment index. This is the normalization coefficient.

5. The method according to claim 4, characterized in that, The specific method for technical risk assessment in step S2 is as follows: System technical risk is defined as: R T,i (t)=S i ×(1-R i (t)) Among them, R T,i (t) represents the technical risk index of the i-th component at time t; S i R represents the severity of the primary failure mode of the i-th component; i (t) represents the reliability of the main failure mode of the i-th component.

6. The method according to claim 5, characterized in that, The specific method for economic risk assessment in step S2 is as follows: Economic risk quantification model: R E,i (t)=O i (t)×[C m,i +C d,i ×T doen,i ] C d,i JP rated ·η loss ·p elec Among them, O i (t) represents the occurrence degree of the failure mode of the i-th component at time t, C m,i To address the direct repair costs for this failure mode, C d,i T represents the power generation loss per unit time caused by system downtime due to the failure of this component. doen,i This represents the mean time to repair for this failure mode.

7. The method according to claim 6, characterized in that, The specific method for environmental risk assessment in step S2 is as follows: Environmental risk quantification model: R V,i (t)=O i (t)×V leak,i ×GWP co2 ×C carbon ×C penalty Among them, V leak,i GWP represents the expected working fluid leakage rate when this failure mode occurs. co2 C represents the global warming potential of carbon dioxide. carbon The carbon cost per unit of carbon dioxide equivalent emissions, C penalty This refers to the fixed fines or emergency response costs that may be incurred due to environmental pollution incidents.

8. The method according to claim 6, characterized in that, The specific steps of the adaptive dynamic TOPSIS method in step S3 are as follows: (1) Matrix construction At each assessment time t, real-time data is collected and integrated with the technology risk index R. T (t), Economic Risk Index R E (t) and environmental risk index R V (t), construct the dynamic decision matrix X(t): Where rows i = 1, 2, 3, ..., m represent m evaluation schemes, and columns j = 1, 2, 3 represent the three risk dimensions of technology, economy, and environment, respectively. This is calculated from the corresponding sub-model. (2) Matrix standardization and entropy weight calculation The decision matrix is ​​standardized using the extreme value method, and the information entropy and dynamic weights of each dimension at time t are calculated: Where, x ij (t) is the element in the i-th row and j-th column of the decision matrix X(t), representing the original evaluation value of the i-th evaluation object at time t in the j-th risk dimension; r ij (t) represents the element in the i-th row and j-th column of the standardized decision matrix R(t), representing x. ij (t) is the value after standardization. (3) Matrix and Dynamic Ideal Solution Update Construct a weighted decision matrix V(t) to determine the dynamic ideal solution: Where β is the fusion coefficient, used to balance real-time observations with state-based expected values; Let be the conditional expectation function, representing the expected value of the ideal solution under the given system aging state α(t) and control input u(t) at time t; α(t) is the health state vector of the system or component output by the dynamic aging model at time t; u(t) is the control input vector of the system at time t. (4) Distance and application progress calculation First, calculate the Euclidean distance between each evaluation object and the dynamic positive and negative ideal solutions: Next, calculate the relative closeness of each object to the ideal solution: in, Let be the Euclidean distance from the i-th evaluation object to the dynamic positive ideal solution at time t; Let C be the Euclidean distance from the i-th evaluation object to the dynamic negative ideal solution at time t; i (t) represents the relative proximity of the i-th evaluation object at time t. (5) Risk ranking, prediction and alert According to C i The risk ranking of all assessed objects is based on the (t) value, and the proximity is predicted in the future time interval Δt based on the system state transition equation: in, Let f(·) be the predicted closeness of the i-th evaluation object at a future time t+Δt; f(·) is the prediction function; and a risk threshold ζ is set. If This will trigger a high-risk alert, prompting operations and maintenance personnel to intervene in the object.

9. The method according to claim 8, characterized in that, The specific method of the graded early warning mechanism in step S4 is as follows: Level 1 Warning (Red): When C i (t)<0.3 or If triggered, immediately perform a shutdown check, activate the emergency plan, respond within 15 minutes, and complete the preliminary assessment within 1 hour. Level II Warning (Orange): When 0.3 ≤ C i (t)≤0.4 or When triggered, the component monitoring frequency is increased to 2-4 times the original plan, the warning is confirmed and monitoring is deployed within 1 hour, and all maintenance preparations are completed within 24 hours. Level 3 Warning (Yellow): When 0.4 ≤ C i (t)≤0.5 or When triggered, strengthen daily monitoring, record abnormal situations, prepare for preventive maintenance, respond within 4 hours, and complete maintenance preparation within 72 hours.

10. The method according to claim 9, characterized in that, The specific method for optimizing maintenance costs in step S4 is as follows: Based on the risk assessment results, and following the principle of "high risk, high investment", the limited maintenance budget is prioritized for allocation to the components with the highest risk level, so as to achieve precise allocation and optimization of maintenance resources.

Images