A method for quantitatively analyzing indirect economic losses of a fusion multi-element power system safety accident

Abstract

A kind of quantitative analysis method of fusing indirect economic loss of multi-element power system safety accident belongs to the technical field of power system safety risk assessment. It includes: extracting human attribute key indicators for standardization processing;Respectively constructing each quantitative model to obtain each dimension comprehensive quantitative value;Social public opinion-network fermentation-social response force-CGE fusion model is constructed, and the conduction coefficient is calibrated by ridge regression analysis method;The standardized index and quantitative value are substituted into the fusion model, and the indirect economic loss comprehensive quantitative value of the fusion social public opinion-network fermentation-social response force attribute is solved;Finally, through sensitivity analysis, the human element with high impact weight is identified, and the targeted loss prevention and control suggestion is generated. The present application fills the gap of social public opinion-network fermentation-social response force and other elements in the analysis of indirect economic loss of power system, realizes the innovative fusion of economic and sociological model, the quantitative result is accurate, and provides scientific basis for power accident loss assessment and risk prevention and control.

Description

Technical Field

[0001] This invention belongs to the field of power system safety risk assessment technology, and relates to a quantitative analysis method for the indirect economic losses of power system safety accidents that integrates multiple factors. Specifically, it relates to a quantitative analysis method for the indirect economic losses of power system safety accidents that integrates social opinion, online fermentation, and social response capabilities, and is used for accident loss assessment, risk prevention and control, and emergency decision support for power companies and regulatory agencies. Background Technology

[0002] As a critical national infrastructure, the impact of safety accidents on modern power systems extends beyond direct economic losses such as equipment damage and power outages. Indirect economic losses triggered by accidents, such as industrial shutdowns, supply chain disruptions, and social disorder, are becoming increasingly prominent. Current loss assessments of power system safety accidents primarily focus on quantifying direct economic losses. While research on indirect economic losses is gaining attention, it mainly relies on traditional economic tools such as input-output models and computable general equilibrium (CGE) models, limiting itself to purely economic dimensions such as industrial linkage losses and regional economic fluctuations. Existing analytical methods (CN119106231A) define indirect economic losses narrowly, employing a linear extrapolation method of "grid event risk × output value per kilowatt-hour." This method is essentially a static estimation based on average output value, reflecting the direct impact of power outages on output, but failing to dynamically capture the systemic and cascading indirect economic losses caused by power outages through supply chain transmission, declining consumer confidence, and evolving public opinion. Its assessment framework lacks consideration of social psychology and human behavioral factors.

[0003] However, real-world accident cases demonstrate that factors such as the evolution of public opinion, the spread of information online, and social emergency response capabilities have a significant transmission, amplification, or inhibitory effect on indirect economic losses. For example, negative public opinion can trigger public panic, decline in consumer confidence, and damage to corporate brands, thereby exacerbating economic losses; while efficient social response capabilities can shorten the recovery period and reduce the scale of losses. Existing technologies have shortcomings, including the lack of humanistic elements (social opinion, online spread, and social response capabilities are not included in the quantitative models of indirect economic losses); poor model adaptability (qualitative research findings from sociology are not effectively embedded into quantitative economic models); and unclear transmission mechanisms (the lack of a quantitative correlation between humanistic attributes and economic losses leads to significant discrepancies between loss assessment results and actual losses, making it difficult to provide accurate basis for accident handling and risk prevention.

[0004] Therefore, there is an urgent need for a method to quantify the indirect economic losses of power system safety accidents that can integrate social opinion, online fermentation, and social response capabilities to construct an economic-sociological collaborative model. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing a quantitative analysis method for indirect economic losses from power system safety accidents that integrates multiple factors. It solves the problems of neglecting social opinion, online fermentation, and social response capacity factors and models, as well as poor adaptability to real-world scenarios in existing quantitative analysis of indirect economic losses. This invention constructs a quantitative system that integrates social opinion, online fermentation, and social response capacity factors, and combines innovative economic and sociological models to achieve accurate modeling and analysis of indirect economic losses after power system safety accidents, providing a scientific basis for loss assessment, post-accident handling, and risk prevention and control of power system safety accidents.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A quantitative analysis method for indirect economic losses from power system safety accidents that integrates multiple factors, specifically a method that integrates public opinion, online amplification, and social response capabilities, includes the following steps: The first step is to extract humanistic attribute indicators, construct an initial indicator set, and then standardize it to obtain a standardized humanistic attribute indicator set. Specifically: Step 1.1: Select the humanistic attribute indicators of social opinion elements, online fermentation elements, and social response elements to obtain the raw data as the initial indicator set; the social opinion element indicators include: public opinion sentiment tendency, public opinion dissemination breadth, and timeliness of authoritative information release; the online fermentation element indicators include: online topic popularity, influence of dissemination nodes, and fermentation cycle; the social response element indicators include: emergency response efficiency, government-enterprise collaborative handling capability, social psychological resilience, and regional livelihood security capability.

[0007] Step 1.2: Based on the initial index set obtained in Step 1.1, the initial index set is normalized and dimensionless to eliminate dimensional differences and obtain a standardized humanistic attribute index set.

[0008] The second step, based on the standardized humanistic attribute index set obtained in step 1.2, involves comprehensive quantitative modeling of public opinion, online fermentation, and social response capabilities. The constructed quantitative models for each element include a comprehensive quantitative model for public opinion, an improved susceptible-infected-recovered (SIR) online fermentation quantitative model, and a multi-level grey comprehensive evaluation quantitative model for social response capabilities. This yields comprehensive quantitative values ​​for public opinion, online fermentation, and social response capabilities. Specifically: Step 2.1, when establishing a comprehensive quantitative model of public opinion, firstly, based on sentiment analysis, extract the sentiment values ​​of each public opinion sentiment value, then combine the entropy weight method to determine the weight of each public opinion sentiment value, and finally use the weighted summation model to calculate the comprehensive quantitative value O of public opinion. The aforementioned comprehensive quantitative model of public opinion is specifically a weighted summation model based on sentiment analysis and entropy weighting, and its expression is as follows: (1) in, As a comprehensive quantitative value of public opinion; Let be the entropy weight of the a-th public opinion indicator; Let be the standardized value of the a-th public opinion indicator; This refers to the quantity of indicators related to public opinion.

[0009] Step 2.2, establishing an improved susceptible-infected-recovered (SIR) network fermentation quantification model. The specific steps are as follows: Step 2.2.1, based on the definitions of cumulative infected nodes, time-varying infection probability between nodes, and transmission speed in Step 1.2, and the traditional SIR transmission model, introduces the influence weight of transmission nodes, and its differential equation system is as follows: (2) (3) (4) in, This represents the probability that node i is in a susceptible state at time t; This represents the probability that node i is in an infected state at time t; Let represent the probability that node i is in the recoverant state at time t; γ is the recovery rate; j represents node i. Represents the set of neighboring nodes of node i; This represents the probability that node i will infect its susceptible neighbor j at time t; This represents the probability that node j is in an infected state at time t; This represents the probability that node i is in the recoverer state at time t.

[0010] By solving this set of differential equations, the curve of the number of infected persons can be obtained. Then calculate the cumulative number of infected nodes. Average propagation speed Average influence weight .

[0011] Step 2.2.2, based on the cumulative number of infected nodes obtained in Step 2.2.1 Average propagation speed Average influence weight The comprehensive quantitative value F of network fermentation is calculated as follows: (5) Where η1, η2, and η3 are weighting coefficients, satisfying η1 + η2 + η3 = 1. For size weighting coefficients, For speed weighting coefficients, The quality weighting coefficient can be determined using the analytic hierarchy process or the entropy weighting method. : Indicates the average influence weight. , wi As the node's influence weight, Let t be the number of infected individuals at time t; this formula reflects the quality of the impact of the main transmission entity. This represents the cumulative number of infected nodes after normalization; This represents the normalized average propagation speed.

[0012] Step 2.3, when establishing a multi-level grey comprehensive evaluation model for social response capacity, a multi-level grey comprehensive evaluation model is adopted, and expert scoring and objective statistical data are combined to calculate the comprehensive quantitative value R of social response capacity. Specifically: Step 2.3.1: Construct an expert scoring matrix. Invite several experts to score four secondary indicators: emergency response efficiency, government-enterprise collaborative handling capability, social psychological resilience, and regional livelihood security capability. The experts cover relevant fields such as power emergency response, public administration, and sociology. Scoring is based on a 1-10 scale, resulting in the original scoring matrix Ap×n, where p is the number of experts, n is the number of indicators, and element a... mk This represents the score given by the m-th expert to the k-th indicator.

[0013] Step 2.3.2: Based on the original expert scoring matrix Ap×n obtained in Step 2.3.1, select the ideal optimal score a from it. 0k =10 is used as the reference sequence, i.e., the reference sequence A0=(10,10,…,10). The discrimination coefficient ρ∈(0,1) is used to adjust the discrimination of the grey relational coefficient.

[0014] Step 2.3.3, based on the highest score 'a' found in Step 2.3.2 0j =10 and using grey relational analysis theory, calculate the grey relational coefficients of each expert for each indicator. : (6) in: Let be the absolute difference between the score given by the m-th expert to the k-th indicator and the ideal value; This represents the minimum difference between two levels; Indicates the maximum difference between two levels; The resolution coefficient is usually set to 0.5, which represents the tolerance for the maximum deviation. This represents the ideal, optimal score; This represents the score given by the m-th expert to the k-th indicator.

[0015] Step 2.3.4, based on the grey relational coefficients obtained in step 2.3.3. The average grey relational degree of social opinion, online fermentation, and social response capacity indicators were calculated respectively. In essence, it involves averaging the grey relational coefficients of each expert to obtain the average grey relational degree of each indicator. : (7) Where p represents the total number of experts.

[0016] Step 2.3.5, based on the average grey relational degree obtained in step 2.3.4 The entropy weight method was used to determine the weights of public opinion, online fermentation, and social response capacity indicators based on objective statistical data. ,satisfy Furthermore, the average grey relational degree of each indicator is weighted and combined with its corresponding weight to obtain a comprehensive quantitative value of social coping capacity. : (8) Among them, the comprehensive quantitative value of social coping ability R∈[0,1], and is positively correlated with social coping ability; The weighting of social response capacity indicators; This represents the average grey relational degree.

[0017] The third step, based on the comprehensive quantitative values ​​(O, F, R) of the elements of public opinion, online fermentation, and social response obtained in the second step, further constructs a fusion model of public opinion-online fermentation-social response-CGE to obtain the comprehensive quantitative value of indirect economic losses integrating the elements of public opinion, online fermentation, and social response. Specifically: Step 3.1: Robustly estimate the impact coefficients of public opinion, online fermentation, and social response on indirect economic losses, setting the ridge regression model as Y=Xβ+ε, X=[O,F,R], β=[k1,k2,-k3]. T The ridge regression estimation formula is used: (9) Where: λ is the ridge parameter, which is selected through cross-validation to minimize the mean squared error; The identity matrix is ​​used; the social opinion transmission coefficient k1, the online fermentation transmission coefficient k2, and the social response transmission coefficient k3 are obtained from this formula, with a value range of 0.1 to 1.0.

[0018] Step 3.2: Based on the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3 obtained in Step 3.1, and using the computable general equilibrium model CGE as a foundation, embedding the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3, a social opinion-online fermentation-social response force-CGE fusion model is constructed. This yields the comprehensive quantitative value of indirect economic losses integrating social opinion, online fermentation, and social response force elements, calculated as follows: (10) Wherein, L represents the comprehensive quantitative value of indirect economic losses, integrating factors such as public opinion, online fermentation, and social response capabilities; L CGE The value represents the indirect economic loss in the pure economic dimension calculated by the traditional CGE model; O, F, and R are the comprehensive quantitative values ​​of public opinion, online fermentation, and social response, respectively; k1 represents the public opinion transmission coefficient, k2 represents the online fermentation transmission coefficient, and k3 represents the social response transmission coefficient.

[0019] Step 3.3: Based on the comprehensive quantitative value calculation expression for indirect economic losses obtained in Step 3.2, which integrates factors such as public opinion, online fermentation, and social response capabilities, data on similar power safety accidents in the region are collected, including the actual indirect economic losses L of each accident obtained from post-accident statistical reports. m CGE model calculated value L CGE And the quantitative value of public opinion O calculated according to the method of the present invention. m Network fermentation quantification value F m Social response force value R m Construct the training dataset: (11) in, This represents the indirect economic loss from the m-th historical accident. =L m / L CGE,m-1 .

[0020] By substituting the training dataset, we obtain a comprehensive quantitative value of indirect economic losses that integrates factors such as public opinion, online fermentation, and social response capabilities.

[0021] The fourth step, based on the comprehensive quantitative values ​​of public opinion, online fermentation, and social response capabilities obtained in the second step, involves conducting sensitivity analysis on these elements to identify high-influence weight indicators. Specifically: The sensitivity analysis method involves adjusting the comprehensive quantitative values ​​of social opinion, online fermentation, and social response capabilities one by one, calculating the rate of change of indirect economic losses ΔL / L and the sensitivity coefficient Sx, where Sx = (ΔL / L) / (ΔX / X), and X = [O, F, R]. The method identifies which of the three factors—social opinion, online fermentation, and social response capabilities—has the highest weight in influencing the loss, thus obtaining the sensitivity analysis results.

[0022] The fifth step, based on the sensitivity analysis results obtained in the fourth step, is to generate targeted loss prevention strategies based on factors such as public opinion, online fermentation, and social response capabilities. Specifically: The loss prevention and control strategy based on the optimization of social opinion, online fermentation, and social response capabilities is generated based on the results of sensitivity analysis. For the high-impact factor indicators, corresponding humanistic loss prevention and control strategies are generated, including one or more of the following: authoritative information release strategy, online public opinion guidance strategy, government-enterprise collaborative emergency response capability enhancement strategy, and social psychological intervention strategy.

[0023] The sixth step involves verifying and comparing the method of this invention with the above-mentioned social opinion-online fermentation-social response-CGE fusion model, and comparing the calculation results of the social opinion-online fermentation-social response-CGE fusion model with the indirect economic losses in actual accident statistics.

[0024] Compared with the prior art, the present invention has the following beneficial effects: (1) The direct effect of this invention is to fill the gap in the quantification of social opinion, online fermentation and social response. For the first time, it systematically incorporates humanistic attributes such as social opinion, online fermentation and social response into the quantification framework of indirect economic losses of power system safety accidents, overcoming the limitations of traditional methods that only focus on the pure economic dimension. Secondly, it realizes the innovative integration of economic and sociological models. By embedding and integrating the improved susceptible-infected-recovered (SIR) network fermentation quantification model, multi-level grey comprehensive evaluation model and CGE model, a loss quantification model with interdisciplinary characteristics is constructed, which improves the scientificity and practicality of the model.

[0025] (2) This invention clarifies the transmission mechanism and influence weight of public opinion, online fermentation, and social response. This invention employs ridge regression and sensitivity analysis to quantitatively reveal the path and degree of influence of various human factors on indirect economic losses, providing data support for precise prevention and control. This method not only provides accurate and highly operable quantitative results, but also reduces subjective bias through standardized processing and multi-model collaboration. The quantitative results can be directly applied to accident assessment, post-accident handling, and risk decision-making, possessing high engineering promotion value. Attached Figure Description

[0026] Figure 1This is an overall flowchart of the method of the present invention; Figure 2 Framework diagram for extracting and standardizing key indicators of humanistic attributes; Figure 3 Spatiotemporal evolution curves for the improved susceptible-infected-recovered (SIR) network fermentation quantification model; Figure 4 A graph showing the sensitivity analysis results of social opinion, online fermentation, and social response capabilities on indirect economic losses; Figure 5 This is a comparison chart of the indirect economic loss quantification results between the traditional CGE model and the social public opinion-online fermentation-social response-CGE fusion model in the embodiments. Detailed Implementation

[0027] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0028] A quantitative analysis method for indirect economic losses from power system safety accidents that integrates multiple factors, specifically a method that integrates public opinion, online amplification, and social response capabilities, includes the following steps: First step, such as Figure 2 As shown, humanistic attribute indicators are extracted, an initial indicator set is constructed, and then standardized to obtain a standardized humanistic attribute indicator set. Specifically: Step 1.1: Select humanistic attribute indicators from the dimensions of public opinion, online fermentation, and social response to obtain raw data as the initial indicator set. The indicators for the public opinion dimension include: sentiment tendency of public opinion, breadth of public opinion dissemination, and timeliness of authoritative information release. The indicators for the online fermentation dimension include: popularity of online topics, influence of dissemination nodes, and fermentation cycle. The indicators for the social response dimension include: emergency response efficiency, government-enterprise collaborative handling capability, social psychological resilience, and regional livelihood security capability.

[0029] Step 1.2: Based on the initial index set obtained in Step 1.1, the initial index set is normalized and dimensionless to eliminate dimensional differences and obtain a standardized humanistic attribute index set.

[0030] Step 1.2.1 involves normalizing and dimensionlessizing the initial set of public opinion indicators. Based on 12,846 crawled comment texts, a fine-grained sentiment analysis model based on BERT is used for annotation. Its core self-attention mechanism is calculated as follows: First, for the input word embedding, calculate the attention score. To measure the relationship between word r and word s: (11) Among them, among them, The unnormalized attention score measures the degree of attention paid to the s-th word by the r-th word. Let r be the word vector of the word at position r; Let be the word vector of the word at position s; This is the query vector, representing the information that the current word wants to find. This is a key vector, representing the information that the word can provide; This is a scaling factor used to prevent the softmax gradient from vanishing due to an excessively large inner product.

[0031] Furthermore, the attention score is calculated using the Softmax function. Normalized to attention weights : (12) in, The sequence length; is the normalized attention weight; k is the summation index, traversing all possible positions.

[0032] Finally, by using these weighted summation vectors, we obtain a new representation of word r. : (13) in, This is a trainable weight matrix used to generate value vectors; This represents the output at the r-th position.

[0033] Statistical analysis shows that if the sentiment value range for the text is defined as [-1, 1], the average score is -0.42, indicating that the overall public opinion is biased towards negativity. Regarding the breadth of public opinion dissemination, the cumulative exposure of related reports and social media topics within 72 hours of the accident was counted, totaling 32.76 million times. To eliminate differences in magnitude, a logarithmic transformation was performed followed by min-max normalization, resulting in a standardized value of 0.81. As for the timeliness of authoritative information release, the time points when the power company and local government first released authoritative information were recorded. The first authoritative information was released 2.3 hours after the accident, based on the timeliness scoring formula defined in this invention. T =max(0,1-t The result of the calculation is 0.97 ( / 72), which reflects the relatively rapid release of information by the authorities.

[0034] Step 1.2.2: Normalize and dimensionlessize the initial indicator set of network fermentation. First, define the cumulative number of infected nodes. Let the time interval of the propagation process be [0, T], then the cumulative number of infected nodes is... Defined as the total number of all nodes that were in an infected state up to time T, its calculation expression is: (14) in, This represents the cumulative number of infected nodes up to time T; Let be the number of newly infected nodes at time t; , These represent the number of infected individuals and the number of recovered individuals at time T, respectively.

[0035] In this example, data was collected through a public opinion monitoring platform, and the number of infected individuals at T=48 hours was I(48)=1.2×10⁻⁶. 4 The number of recoveries R(48) = 3.8 × 10 4 The original value of the cumulative number of infected nodes is: (15) To facilitate subsequent calculations, an extreme value normalization method was used to map all indicators to the [0,1] interval. Based on historical public opinion data, the minimum value of the original cumulative number of infected nodes was determined. =0, maximum value =1.0×10⁵, then the normalized cumulative number of infected nodes is: (16) in: and These are the minimum and maximum cumulative number of infected nodes in historical data or simulation scenarios, respectively.

[0036] At the same time, the time-varying infection probability between nodes is quantitatively defined: (17) in: Let be the probability that node i infects its susceptible neighbor j at time t; Basic infection rate; This is an influence adjustment coefficient that controls the enhancing effect of node weights on propagation. θ represents the node's influence weight; θ represents the information timeliness decay rate. To simulate the aging of information over time.

[0037] In this example, the parameters are obtained by fitting historical public opinion data: basic infection rate. =0.03, influence moderating coefficient =0.5, Information timeliness decay rate =0.1.

[0038] To facilitate subsequent calculations, an extreme value normalization method is used to map all indicators to the [0,1] interval. Taking a high-influence node as an example, its number of followers is 2.5 × 10⁻⁶. 6 Then, count the number of followers for all nodes to obtain the minimum value. maximum value =5.0×10⁶, then the normalized node influence weight is: (18) Substituting this value into equation (17), we obtain the time-varying infection probability of this node at the second hour after the accident (t=2): (19) in: The original influence weight of node i (such as number of followers, degree centrality, PageRank value, etc.); and These are the minimum and maximum values ​​of the original influence weights of all nodes, respectively.

[0039] At the same time, quantify the speed of dissemination: Define instantaneous propagation speed The rate of change in the number of infected individuals at time t: (20) At the micro level, It is obtained by summing up all possible new infection events: (twenty one) in: Let i be the set of neighbors of node i; The number of infected persons at time t; Let t be the number of susceptible individuals at time t.

[0040] In this example, the raw values ​​of the instantaneous propagation velocity at each moment are obtained through simulation calculation. The instantaneous propagation velocity reaches its peak at t=8 hours. raw (8) = 2100 nodes / hour.

[0041] Finally, the average propagation speed of the entire propagation process is quantified. The integral average of the instantaneous velocity over [0,T]: (twenty two) To facilitate subsequent calculations, an extreme value normalization method was used to map all indicators to the [0,1] interval. Based on historical public opinion data, the minimum original propagation speed was determined. =0, maximum value =3000 nodes / hour, then the normalized instantaneous propagation speed is: (twenty three) right Integrating and averaging over [0, 48] yields the original value of the average propagation velocity. =850 nodes / hour, normalized to: (twenty four) in: Let be the original propagation speed at time t. and These are the minimum and maximum values ​​of the original propagation speed, respectively.

[0042] Step 1.2.3: The initial set of social response indicators was normalized and dimensionless. First, the emergency response efficiency was analyzed. The average arrival time of emergency teams was 42 minutes. Based on the scoring standard of "<30 minutes = 9-10 points, 30-60 minutes = 6-8 points, >60 minutes = 1-5 points", the corresponding score was 7.2, which was 0.68 after min-max normalization. Next, the government-enterprise collaborative response capability was analyzed. Five emergency management experts were invited to subjectively score the completeness of the linkage mechanism and the efficiency of information sharing from 1 to 10. The average score was 6.8, which was normalized to 0.62. Then, the social psychological resilience was analyzed. Based on the month-on-month change in the number of public complaints and the comprehensive assessment of the online sentiment index, the number of complaints increased by 35% compared with the previous day, and the panic index was 0.57, which was calculated to be 0.55. Finally, the regional livelihood security capability was analyzed. The emergency power supply coverage rate was 73%, and the material security coverage rate was 81%, which was weighted to 0.72.

[0043] The second step, based on the standardized humanistic attribute indicator set obtained in step 1.2, involves quantitatively modeling the elements of public opinion, online fermentation, and social response capacity. The constructed quantitative models for each dimension include a comprehensive quantitative model of public opinion, an improved susceptible-infected-recovered (SIR) online fermentation quantitative model, and a multi-level grey comprehensive evaluation quantitative model of social response capacity. This yields a comprehensive quantitative value for public opinion, online fermentation, and social response capacity. Specifically: Step 2.1, when establishing a comprehensive quantitative model of public opinion, firstly, based on sentiment analysis, extract the sentiment values ​​of each public opinion sentiment value, then combine the entropy weight method to determine the weight of each public opinion sentiment value, and finally use the weighted summation model to calculate the comprehensive quantitative value O of public opinion. Step 2.1.1: Based on the sentiment analysis obtained in Step 1.2.1, extract the sentiment values ​​for each public opinion sentiment value, and use the entropy weight method to determine the weight of each sentiment value based on objective data. 72 time points within 72 hours before and after the accident are selected as samples (one sample point per hour, m=72), and data for the above four indicators are collected at each time point. The calculation steps are as follows: Calculate the weight of the g-th sample value under the a-th indicator: (25) Calculate the entropy value of the a-th index: (26) Calculate the entropy weight of the a-th index: (27) In this example, the entropy weights of each public opinion sentiment value are obtained through the above calculations: sentiment value weight ω1 = 0.35; breadth of public opinion dissemination weight ω2 = 0.28; timeliness of authoritative information release weight ω3 = 0.22; and public attention weight ω4 = 0.15. Step 2.1.2, based on the entropy weight of the a-th index obtained in Step 2.1.1 In addition to the entropy weights of each sentiment tendency value, a weighted summation quantitative model of public opinion is established based on sentiment tendency analysis and the entropy weight method. Its expression is: (1) Substituting the entropy weights of each sentiment index into the formula, we get: =0.1015+0.2128+0.2134+0.0720≈0.60(28) Step 2.1.4: Based on step 2.1.3, the comprehensive quantitative value of public opinion is calculated as O = 0.60. Since this value ranges from [0,1], a larger value indicates a higher intensity of public opinion. Therefore, in this example, O = 0.60 indicates that the public opinion triggered by this power grid accident is at a moderately high level. The sentiment tendency value is 0.29 (corresponding to the original value -0.42), reflecting obvious negative emotions, while the timeliness score of authoritative information release is as high as 0.97, which to some extent alleviates the further deterioration of public opinion.

[0044] Step 2.2, establishing an improved susceptible-infected-recovered (SIR) network fermentation quantification model. The specific steps are as follows: Step 2.2.1, based on the definitions of cumulative infected nodes, time-varying infection probability between nodes, and transmission speed in Step 1.2, and the traditional SIR transmission model, introduces the influence weight of transmission nodes, and its differential equation system is as follows: (2) (3) (4) in: This represents the probability that node i is in a susceptible state at time t; This represents the probability that node i is in an infected state at time t; γ represents the probability that node i is in the recoverant state at time t; γ is the recovery rate; j represents node j; Represents the set of neighboring nodes of node i; This represents the probability that node i will infect its susceptible neighbor j at time t; This represents the probability that node j is in an infected state at time t; This represents the probability that node i is in the recoverer state at time t.

[0045] In this example, the recovery rate γ = 0.05 is taken, and the Monte Carlo method is used to numerically solve the above differential equations to obtain the evolution curves of the number of nodes at each time step. ,like Figure 3 As shown. Based on the simulation results, the average influence weight wavg(t) of the infected node at each time point is further calculated: (29) Step 2.2.2, based on the cumulative number of infected nodes obtained in Step 2.2.1 Average propagation speed Average influence weight The overall quantitative value F of the network fermentation is calculated as follows: (5) Where η1, η2, and η3 are weighting coefficients, satisfying η1 + η2 + η3 = 1. For size weighting coefficients, For speed weighting coefficients, The quality weighting coefficient can be determined using the analytic hierarchy process or the entropy weighting method. Indicates the average influence weight. , wi As the node's influence weight, Let t be the number of infected individuals at time t; this formula reflects the quality of the impact of the main transmission entity. This represents the cumulative number of infected nodes after normalization; express; This represents the normalized average propagation speed.

[0046] In this example, the weight coefficients are determined using the analytic hierarchy process (AHP): η1 = 0.4, η2 = 0.3, and η3 = 0.3. Based on the simulation results, the normalized cumulative number of infected nodes N is calculated. infected (48) = 0.50; Normalized average propagation speed =0.283; Average Influence Weighted Integral Value .

[0047] Substitute into formula (5) to calculate the comprehensive quantitative value of network fermentation: =0.7579×0.6974×0.8626=0.456(29) Step 2.2.3, based on the comprehensive quantitative value of network fermentation calculated in step 2.2.2, is F=0.456. Since this value ranges from [0,1], a larger value indicates a higher intensity of network fermentation. Therefore, in this example, F=0.456 indicates that the network public opinion fermentation caused by this power grid accident is at a moderate level. The normalized value of the cumulative number of infected nodes is 0.50, the normalized value of the average propagation speed is 0.283, and the average influence weight integral value is 0.62, reflecting the characteristics of high quality influence of the propagating subject but relatively slow propagation speed.

[0048] Step 2.3, when establishing a multi-level grey comprehensive evaluation model for social response capacity, a multi-level grey comprehensive evaluation model is adopted, and expert scoring and objective statistical data are combined to calculate the comprehensive quantitative value R of social response capacity. Specifically: Step 2.3.1: Construct an expert scoring matrix. Invite several experts to score four secondary indicators: emergency response efficiency, government-enterprise collaborative handling capability, social psychological resilience, and regional livelihood security capability. The experts cover relevant fields such as power emergency response, public administration, and sociology. Scoring is based on a 1-10 scale, resulting in the original scoring matrix Ap×n, where p is the number of experts, n is the number of indicators, and element a... mk This represents the score given by the m-th expert to the k-th indicator.

[0049] Step 2.3.2: Based on the original expert scoring matrix Ap×n obtained in Step 2.3.1, select the ideal optimal score a from it. 0k =10 is used as the reference sequence, i.e., the reference sequence A0=(10,10,…,10). The discrimination coefficient ρ∈(0,1) is used to adjust the discrimination of the grey relational coefficient.

[0050] Step 2.3.3, based on the highest score 'a' found in Step 2.3.2 0j =10 and using grey relational analysis theory, calculate the grey relational coefficients of each expert for each indicator. : (6) in: Let be the absolute difference between the score given by the m-th expert to the k-th indicator and the ideal value; This represents the minimum difference between two levels; Indicates the maximum difference between two levels; The resolution coefficient is usually set to 0.5, which represents the tolerance for the maximum deviation. This represents the ideal, optimal score; This represents the score given by the m-th expert to the k-th indicator.

[0051] Step 2.3.4, based on the grey relational coefficients obtained in step 2.3.3. The average grey relational degree of social opinion, online fermentation, and social response capacity indicators were calculated respectively. In essence, it involves averaging the grey relational coefficients of each expert to obtain the average grey relational degree of each indicator. : (7) Where p is the total number of experts.

[0052] Step 2.3.5, based on the average grey relational degree obtained in step 2.3.4 The entropy weight method was used to determine the weights of public opinion, online fermentation, and social response capacity indicators based on objective statistical data. ,satisfy Furthermore, the average grey relational degree of each indicator is weighted and combined with its corresponding weight to obtain a comprehensive quantitative value of social coping capacity. : (8) Wherein: the quantization value R∈[0,1], and is positively correlated with coping ability.

[0053] In the example, the calculated results are r1=0.68, r2=0.62, r3=0.55, and r4=0.72. Indicator weights. The entropy weight method, based on objective data, was used to determine the values: w1=0.30, w2=0.25, w3=0.20, and w4=0.25. Finally, a weighted summation was performed to obtain the social response force value. (30) This value indicates that the overall level of social response capability in this incident is 0.65, and the value generally ranges from 0 to 1. The larger the value, the stronger the response capability, indicating that there is still considerable room for improvement at this stage.

[0054] The third step, based on the comprehensive quantitative values ​​(O, F, R) of the elements of public opinion, online fermentation, and social response obtained in the second step, further constructs a fusion model of public opinion-online fermentation-social response-CGE, and builds a training dataset. Substituting the training dataset into the fusion model yields the comprehensive quantitative value of indirect economic losses integrating the elements of public opinion, online fermentation, and social response. Specifically: Step 3.1: Robustly estimate the impact coefficients of public opinion, online fermentation, and social response on indirect economic losses, setting the ridge regression model as Y=Xβ+ε, X=[O,F,R], β=[k1,k2,-k3]. T The ridge regression estimation formula is used: (9) Where: λ is the ridge parameter, which is selected by cross-validation to minimize the mean square error, and the transmission coefficients k1, k2, and k3 range from 0.1 to 1.0.

[0055] Step 3.2: Based on the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3 obtained in Step 3.1, and using the computable general equilibrium model CGE as a foundation, embedding the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3, a social opinion-online fermentation-social response force-CGE fusion model is constructed. This yields the comprehensive quantitative value of indirect economic losses integrating social opinion, online fermentation, and social response force elements, calculated as follows: (10) Wherein, L represents the comprehensive quantitative value of indirect economic losses, integrating factors such as public opinion, online fermentation, and social response capabilities; L CGE The value represents the indirect economic loss in the pure economic dimension calculated by the traditional CGE model; O, F, and R are the comprehensive quantitative values ​​of public opinion, online fermentation, and social response, respectively; k1 represents the public opinion transmission coefficient, k2 represents the online fermentation transmission coefficient, and k3 represents the social response transmission coefficient.

[0056] Step 3.3: Based on the comprehensive quantitative value calculation expression for indirect economic losses obtained in Step 3.2, which integrates factors such as public opinion, online fermentation, and social response capabilities, data on 10 similar power safety accidents that occurred in the region in the past 5 years were collected, including the actual indirect economic losses L of each accident obtained from post-accident statistical reports. m CGE model calculated value L CGE And the quantitative value of public opinion O calculated according to the method of the present invention. m Network fermentation quantification value F m Social response force value R mConstruct the training dataset: (11) in: This represents the indirect economic loss from the m-th historical accident. =L m / L CGE,m-1 .

[0057] By substituting the training dataset, we obtain a comprehensive quantitative value of indirect economic losses that integrates factors such as public opinion, online fermentation, and social response capabilities.

[0058] Furthermore, based on the comprehensive quantitative value calculation expression for indirect economic losses that integrates social opinion, online fermentation, and social response capabilities, the calibrated transmission coefficient is substituted to obtain: (31) Substituting O=0.78, F=0.82, R=0.65 and LCGE=230 million yuan for this accident, we calculate: L = 2.3 × (1 + 0.32 × 0.78 + 0.45 × 0.82 - 0.50 × 0.65) = 2.97528 billion yuan (32) The fourth step, based on the comprehensive quantitative values ​​of the various humanistic attributes obtained in the second step, involves sensitivity analysis of public opinion, online fermentation, and social response capabilities to identify high-influence weight indicators. Specifically: To identify the contribution of each human attribute element to the overall loss, a sensitivity analysis was conducted using the single-factor perturbation method established in this invention. O, F, and R were increased or decreased by 5%, 10%, and 20% respectively from the baseline values, and the corresponding indirect economic loss change rate ΔL / L and sensitivity coefficient Sx=(ΔL / L) / (Δx / x) were calculated. Figure 4 As shown, the sensitivity analysis results indicate that a ±10% disturbance in public opinion (O) leads to a ±3.2% change in losses, with a sensitivity coefficient SO = 0.32; a ±10% disturbance in online fermentation (F) leads to a ±4.5% change in losses, with a sensitivity coefficient SF = 0.45; and a ±10% disturbance in social coping mechanisms (R) leads to a change in losses. The impact was 5.0%, with a sensitivity coefficient SR of -0.50. Analysis shows that the positive amplification effect of online escalation on indirect economic losses was most significant, followed by the negative inhibitory effect on social response capacity, while the impact of public opinion was relatively small. Therefore, the focus of humanistic loss prevention and control in this incident should be on guiding online public opinion and improving social emergency response capabilities.

[0059] The fifth step, based on the sensitivity analysis results obtained in the fourth step, is to generate specific prevention and control strategies for humanistic attribute indicators with high influence weights. Regarding the prevention and control of online escalation, it is recommended to establish a rapid response mechanism for core dissemination nodes. Within one hour of an incident, identify the top 20 dissemination nodes with high influence weights on social media platforms, and reduce the probability of spreading false information by targeted push of authoritative information and invitations to participate in official releases. Simultaneously, implement dynamic monitoring and intervention of online topics, using natural language processing technology to monitor topic popularity and sentiment in real time. When the popularity exceeds a threshold or signs of malicious escalation appear, initiate emergency intervention, such as pinning debunking information and restricting the dissemination permissions of malicious accounts to shorten the escalation cycle. Regarding improving social response capabilities, it is recommended to improve the government-enterprise collaborative emergency mechanism, promote the establishment of a regular linkage mechanism between power companies and local governments, clarify information sharing channels and emergency command procedures, and conduct regular joint drills to improve collaborative response efficiency. At the same time, enhance regional livelihood security capabilities, increase investment in emergency power supply equipment and material reserves, increase emergency power supply coverage to over 90%, and establish a public psychological intervention hotline to promptly alleviate panic and reduce the number of public complaints. Regarding public opinion guidance, it is recommended to optimize the timeliness of the first release of authoritative information, compressing the release time of authoritative information to within 1 hour. The information content should include the cause of the accident, the scope of impact, the expected recovery time, and temporary support measures to enhance public trust. Information should also be released simultaneously on mainstream platforms such as Weibo, WeChat, and Douyin, with dedicated personnel responding to public concerns to reduce negative sentiment.

[0060] The sixth step involves verifying and comparing the method used in this embodiment based on the aforementioned social opinion-online fermentation-social response-CGE fusion model. Figure 5As shown, the calculation results of the fusion model are compared with the indirect economic losses from the actual accident statistics. According to post-accident statistics, the indirect economic losses caused by this accident, including industrial losses, supply chain disruptions, and social order impacts, amounted to approximately 300 million yuan. The calculation value of this invention is 298 million yuan, with an absolute error of 2 million yuan and a relative error of 0.67%; while the calculation value of the traditional CGE model is 230 million yuan, with a relative error of 30.43%. The comparison results show that the method of this invention significantly improves the quantitative accuracy of indirect economic losses, effectively overcomes the limitation of traditional models that ignore human factors, and verifies the scientific nature and engineering applicability of the model. Meanwhile, to test the robustness of the model to input data disturbances, a multi-dimensional robustness analysis was conducted. First, the original data of human factor indicators were subjected to a ±5% random disturbance to simulate data collection errors, and the calculation was repeated 100 times. The results show that the coefficient of variation (CV) of the predicted indirect economic losses is 0.032, indicating that the model is not sensitive to data errors. Secondly, by varying the ridge parameter λ of the ridge regression within the range of 0.05 to 0.30, the changes in the social opinion transmission coefficient k1, the online fermentation transmission coefficient k2, and the social response transmission coefficient k3 were all less than 0.05, and the changes in the predicted loss were less than 0.1 billion yuan, verifying the robustness of the model to the hyperparameter selection. Furthermore, using leave-one-out cross-validation, after successively removing one historical accident, the parameters were recalibrated and the loss of this accident was predicted. The mean of the 10 predicted values ​​was 297 million yuan, and the standard deviation was 0.06 billion yuan, which highly matched the actual values, indicating that the model has good generalization ability.

[0061] In summary, this embodiment fully realizes the application of the method proposed in this invention in a real accident scenario. It systematically demonstrates the entire process from data preparation, indicator quantification, model fusion, parameter calibration to prevention and control strategy generation. The quantification results are accurate and the analysis conclusions are reasonable, which can provide strong decision support for loss assessment and risk prevention and control of power system safety accidents.

[0062] The above specific embodiments further illustrate the purpose, technical solution and beneficial effects of this application. It should be understood that the above are only specific embodiments of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of this application should be included within the scope of protection of this application.

Claims

1. A quantitative analysis method for indirect economic losses from power system safety accidents that integrates multiple factors, characterized in that, The quantitative analysis method described is a method for quantitatively analyzing indirect economic losses to power system safety that integrates factors such as public opinion, online fermentation, and social response capabilities. It includes the following steps: The first step is to extract humanistic attribute indicators, construct an initial indicator set, and standardize it to obtain a standardized humanistic attribute indicator set. The second step, based on the standardized humanistic attribute indicator set obtained in the first step, involves comprehensive quantitative modeling of social opinion, online fermentation, and social response capabilities. The constructed quantitative models for each element include a comprehensive quantitative model of social opinion, an improved quantitative model of susceptible-infected-recovered online fermentation, and a multi-level grey comprehensive evaluation quantitative model of social response capabilities. This yields comprehensive quantitative values ​​for social opinion, online fermentation, and social response capabilities. Specifically: Step 2.1, when establishing a comprehensive quantitative model of public opinion, firstly, based on sentiment analysis, extract the sentiment values ​​of each public opinion sentiment value, then combine the entropy weight method to determine the weight of each public opinion sentiment value, and finally use the weighted summation model to calculate the comprehensive quantitative value O of public opinion. Step 2.2, establishing an improved susceptible-infected-recovered network fermentation quantification model. The specific steps are as follows: Introducing the influence weights of propagation nodes yields a differential equation; solving it produces an infected person count curve; and the cumulative number of infected nodes is then calculated. Average propagation speed Average influence weight The comprehensive quantitative value F of network fermentation is calculated. Step 2.3: When establishing a multi-level grey comprehensive evaluation model for social response capacity, a multi-level grey comprehensive evaluation model is adopted, and the comprehensive quantitative value R of social response capacity is calculated by combining scoring and statistical data. The third step, based on the comprehensive quantitative values ​​O, F, and R of the elements of public opinion, online fermentation, and social response obtained in the second step, constructs a fusion model of public opinion-online fermentation-social response-CGE to obtain the comprehensive quantitative value of indirect economic losses integrating the elements of public opinion, online fermentation, and social response; specifically: Step 3.1: Estimate the impact coefficients of public opinion, online fermentation, and social response on indirect economic losses; Step 3.2: Based on the influence coefficients obtained in Step 3.1, and using the computable general equilibrium model CGE as a foundation, embed the social opinion transmission coefficient, the network fermentation transmission coefficient, and the social response force transmission coefficient to construct a social opinion-network fermentation-social response force-CGE fusion model. Step 3.3: Collect data on similar power safety accidents and construct a training dataset; substitute the training dataset into the social opinion-online fermentation-social response-CGE fusion model to obtain the comprehensive quantitative value of indirect economic losses that integrates social opinion, online fermentation, and social response factors; The fourth step involves conducting a sensitivity analysis of the social public opinion, online fermentation, and social response factors obtained in the second step, based on the comprehensive quantitative values ​​of these factors, to identify high-impact weight indicators. The fifth step involves generating targeted loss prevention strategies based on the sensitivity analysis results obtained in the fourth step, taking into account factors such as public opinion, online fermentation, and social response capabilities.

2. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 1, is characterized in that... The first step specifically includes: Step 1.1: Select the humanistic attribute indicators of social opinion elements, online fermentation elements, and social response elements to obtain the raw data as the initial indicator set; The social public opinion elements include: public sentiment, breadth of public opinion dissemination, and timeliness of authoritative information release; the online fermentation elements include: online topic popularity, influence of dissemination nodes, and fermentation cycle; the social response capacity elements include: emergency response efficiency, government-enterprise collaborative handling capacity, social psychological resilience, and regional livelihood security capacity. Step 1.2: Based on the initial index set obtained in Step 1.1, the initial index set is normalized and dimensionless to eliminate dimensional differences and obtain a standardized humanistic attribute index set.

3. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 2, is characterized in that... Step 2.1 specifically involves: The comprehensive quantitative model of public opinion is a weighted summation model based on sentiment analysis and entropy weighting, and its expression is as follows: （1） in, As a comprehensive quantitative value of public opinion; Let be the entropy weight of the a-th public opinion indicator; Let be the standardized value of the a-th public opinion indicator; This refers to the quantity of indicators related to public opinion.

4. The quantitative analysis method for indirect economic losses from multi-factor power system safety accidents according to claim 3, characterized in that, Step 2.2 specifically involves: Step 2.2.1 introduces the influence weight of the propagation nodes, and its differential equation system is as follows: （2） （3） （4） in, This represents the probability that node i is in a susceptible state at time t; This represents the probability that node i is in an infected state at time t; Let represent the probability that node i is in the recoverant state at time t; γ is the recovery rate; j represents node i. Represents the set of neighboring nodes of node i; This represents the probability that node i will infect its susceptible neighbor j at time t; This represents the probability that node j is in an infected state at time t; This represents the probability that node i is in the recoverer state at time t; The curve of the number of infected persons was obtained by solving the differential equation. Then calculate the cumulative number of infected nodes. Average propagation speed Average influence weight ; Step 2.2.2, based on the cumulative number of infected nodes obtained in Step 2.2.1 Average propagation speed Average influence weight The comprehensive quantitative value F of network fermentation is calculated as follows: （5） Where η1, η2, and η3 are weighting coefficients, satisfying η1 + η2 + η3 = 1. For size weighting coefficients, For speed weighting coefficients, The quality weighting coefficients are determined using the analytic hierarchy process and the entropy weighting method. Indicates the average influence weight. , wi As the node's influence weight, The number of infected persons at time t; This represents the cumulative number of infected nodes after normalization; This represents the normalized average propagation speed.

5. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 4, is characterized in that... Step 2.3 specifically involves: Step 2.3.1: Construct the expert scoring matrix using a 1-10 scoring system to obtain the original scoring matrix Ap×n, where p is the number of experts, n is the number of indicators, and element a mk This represents the score given by the m-th expert to the k-th indicator; Step 2.3.2: Based on the original expert scoring matrix Ap×n obtained in Step 2.3.1, select the ideal optimal score a from it. 0k =10 is used as the reference sequence, that is, the reference sequence A0=(10,10,…,10); the discrimination coefficient ρ∈(0,1) is used to adjust the discrimination of the gray relational coefficient; Step 2.3.3, based on the highest score 'a' found in Step 2.3.2 0j =10 and using grey relational analysis theory, calculate the grey relational coefficients of each expert for each indicator. : （6） in: Let be the absolute difference between the score given by the m-th expert to the k-th indicator and the ideal value; This represents the minimum difference between two levels; Indicates the maximum difference between two levels; The resolution coefficient is set to 0.

5. This represents the ideal, optimal score; This represents the score given by the m-th expert to the k-th indicator; Step 2.3.4, based on the grey relational coefficients obtained in step 2.3.

3. The average grey relational degree of social opinion, online fermentation, and social response capacity indicators were calculated respectively. The average grey relational degree of each indicator was obtained. : （7） Where p is the total number of experts; Step 2.3.5, based on the average grey relational degree obtained in step 2.3.4 The entropy weight method was used to determine the weights of public opinion, online fermentation, and social response capacity indicators based on objective statistical data. ,satisfy The average grey relational degree of each indicator is weighted and combined with its corresponding weight to obtain a comprehensive quantitative value of social coping capacity. : （8） Among them, the comprehensive quantitative value of social coping ability R∈[0,1], and is positively correlated with social coping ability; The weighting of social response capacity indicators; This represents the average grey relational degree.

6. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 5, is characterized in that... The third step is specifically as follows: Step 3.1: Estimate the impact coefficients of public opinion, online fermentation, and social response on indirect economic losses. Set the ridge regression model as Y=Xβ+ε, X=[O,F,R], β=[k1,k2,-k3]. T Ridge regression estimation formula is used: （9） Where: λ is the ridge parameter, which is selected through cross-validation to minimize the mean squared error; It is the identity matrix; Step 3.2: Based on the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3 obtained in Step 3.1, and using the equilibrium model CGE as a foundation, embedding the social opinion transmission coefficient k1, online fermentation transmission coefficient k2, and social response force transmission coefficient k3, a social opinion-online fermentation-social response force-CGE fusion model is constructed. This yields the comprehensive quantitative value of indirect economic losses integrating social opinion, online fermentation, and social response force elements, calculated as follows: （10） Wherein, L represents the comprehensive quantitative value of indirect economic losses, integrating factors such as public opinion, online fermentation, and social response capabilities; L CGE The indirect economic loss value is calculated in the pure economic dimension of the traditional CGE model; k1 represents the social opinion transmission coefficient, k2 represents the online fermentation transmission coefficient, and k3 represents the social response transmission coefficient. Step 3.3: Based on the comprehensive quantitative value calculation expression for indirect economic losses obtained in Step 3.2, which integrates factors such as public opinion, online fermentation, and social response capabilities, collect data on similar power safety accidents in the region, including the actual indirect economic losses L of each accident obtained from post-accident statistical reports. m CGE model calculated value L CGE And the quantitative value of public opinion O m Network fermentation quantification value F m Social response force value R m ; Construct the training dataset: （11） in, This represents the indirect economic loss from the m-th historical accident. =L m / L CGE,m-1 ; Based on the training dataset, the social opinion-online fermentation-social response-CGE fusion model is substituted to obtain the comprehensive quantitative value of indirect economic loss that integrates the elements of social opinion, online fermentation, and social response.

7. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 6, is characterized in that... In the third step, the social opinion transmission coefficient k1 ranges from 0.1 to 1.0; the online fermentation transmission coefficient k2 ranges from 0.1 to 1.0; and the social response transmission coefficient k3 ranges from 0.1 to 1.

0.

8. The quantitative analysis method for indirect economic losses from power system safety accidents integrating multiple factors, as described in claim 7, is characterized in that... In the fourth step, the sensitivity analysis method involves adjusting the comprehensive quantitative values ​​of social opinion, online fermentation, and social response capabilities one by one, calculating the rate of change of indirect economic losses ΔL / L and the sensitivity coefficient Sx, where Sx=(ΔL / L) / (ΔX / X), and X=[O,F,R]; identifying which of the four factors (social opinion, online fermentation, and social response capabilities) has the highest weight in terms of impact on losses, and obtaining the sensitivity analysis results.

9. The quantitative analysis method for indirect economic losses from multi-factor power system safety accidents according to claim 8, characterized in that, In the fifth step, the loss prevention and control strategy for optimizing the elements of public opinion, online fermentation, and social response is generated based on the results of sensitivity analysis, targeting the element indicators with high impact weights.

Images