Quantitative decision-making method for electric vehicle fire safety three-level evaluation system

Abstract

The invention discloses a quantitative decision-making method for an electric vehicle fire safety three-level evaluation system. The quantitative decision-making method comprises the following steps: step 1, constructing a three-level evaluation index system; 2, performing quantitative scoring on the three-level indexes by the plurality of large language models to obtain an expert comprehensive scoring vector of each large language model; 3, calculating a subjective weight vector W (x) of each level of index according to the expert comprehensive scoring vector, and if the consistency ratio CR does not meet the consistency requirement, iteratively correcting W (x) by adopting a power method to obtain an initial weight vector system W; 4, introducing a fire news collection agent to generate a vehicle type subtraction item delta G and an accident cause amplification coefficient (1 + delta), and obtaining a subjective weight vector system W'after dynamic correction; and step 5, calculating a comprehensive score Sfinal, dividing fire risk grades, and generating a visual report. According to the invention, a three-level index system is constructed and a fire news collection agent is introduced, so that the timeliness and adaptability of the system are significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric vehicle safety evaluation, in particular to a quantitative decision-making method for a three-level evaluation system of electric vehicle fire safety. BACKGROUND

[0002] With the rapid popularization of new energy vehicles, especially electric vehicles, their power performance and energy utilization rate have made significant progress, but the risk of fire accidents has also gradually attracted widespread attention. Power batteries are prone to combustion or even explosion under extreme conditions such as thermal runaway, short circuit, and collision, posing a serious threat to vehicles, personnel, and public safety. The existing evaluation methods for electric vehicle fire safety generally have the following shortcomings:

[0003] Static and lagging: Most evaluation systems are based on fixed index weights and historical accident data, lacking the ability to dynamically respond to emerging risk factors, making it difficult to reflect the latest safety situation in a timely manner.

[0004] Single expert opinion: Traditional methods often rely on a small number of field experts for subjective weighting, and the evaluation results are affected by individual experience and cognitive bias, with insufficient stability and objectivity of the weight results.

[0005] Lack of public opinion data: The current system has a low degree of utilization of unstructured data such as real-time accident reports and public opinion, and has not incorporated external data sources such as news and announcements into weight correction and risk determination.

[0006] Therefore, there is an urgent need for a quantitative decision-making method and system that can integrate multi-source expert opinions, dynamically correct index weights, and reflect the level of vehicle fire risk in real time, to improve the scientificity, accuracy, and timeliness of electric vehicle fire safety evaluation, and provide reliable decision-making basis for regulatory agencies, vehicle manufacturers, and users. SUMMARY

[0007] The purpose of the present application is to overcome the defects of static and lagging of existing electric vehicle fire safety evaluation systems, and to propose a quantitative decision-making method for a three-level evaluation system of electric vehicle fire safety, which integrates artificial intelligence multi-expert modeling and dynamic public opinion correction to realize the scientization, precision, and real-time of safety evaluation.

[0008] The technical solution adopted to achieve the purpose of the present application is:

[0009] A quantitative decision-making method for a three-level evaluation system of electric vehicle fire safety, comprising the following steps:

[0010] Step 1: Construct a three-level evaluation index system according to the target layer, criterion layer, and sub-criterion layer. The target layer is the comprehensive evaluation of electric vehicle safety, the criterion layer is the first-level index, and the sub-criterion layer is the second-level and third-level index.

[0011] Step 2: Select multiple large language models to simulate experts and quantify and score each indicator in the first, second and third level indicators in Step 1. Each large language model outputs a comprehensive score vector.

[0012] Step 3: Construct a quantitative decision-making model for the indicator evaluation system: Based on the comprehensive scoring vector of each major language model in Step 2, introduce the authority weight coefficient γ of each major language model. k By using a weighted geometric mean, the importance of any two indicators at each level of the large language model is compared pairwise to obtain the comprehensive judgment matrix element a. ij Based on a ij Construct a comprehensive judgment matrix A, and calculate the subjective weight vector W for each level of indicator based on A. (x) It then determines whether the consistency ratio CR meets the consistency requirements. If not, it uses a power-law iterative correction method to adjust W. (x) Thus, an initial weight vector system W that meets the consistency requirements is obtained;

[0013] Step 4: Introduce the fire news gathering agent to generate a vehicle model deduction term ΔG and an accident cause amplification coefficient (1+δ). Use the accident cause amplification coefficient (1+δ) to evaluate the matrix element a in the comprehensive judgment matrix A from Step 3. ij Dynamic correction is performed to obtain the dynamically corrected subjective weight vector system W′, and to generate the dynamic correction coefficient α=1-δ for the three-level indicator scores;

[0014] Step 5: Manually score each of the three-level indicators from Step 1 based on actual vehicles to obtain the actual vehicle scoring results. The third-level indicator score is calculated by combining the dynamic correction coefficient α obtained in step 4 with the third-level indicator score. The scores of the second-level and first-level indicators are calculated sequentially based on the third-level indicator scores and W′. Finally, the comprehensive score S is calculated by combining the vehicle model deduction item ΔG. final This serves as a comprehensive safety evaluation for electric vehicles.

[0015] In the above technical solution, the primary indicators mentioned in step 1 include battery safety, electrical safety, and mechanical safety.

[0016] In the above technical solution, the subjective weight vector W at each level mentioned in step 3 (x) Represented as:

[0017] W (x) =[w1,w2,…,w n ]

[0018] In the formula, w i is the subjective weight of index i, and n is the matrix order;

[0019] The subjective weight w i of the index i is calculated according to the following formula:

[0020]

[0021] In the formula, a ij is the comprehensive judgment matrix element of index i and index j;

[0022] The comprehensive judgment matrix element a ij of index i and index j is calculated according to the following formula:

[0023]

[0024] In the formula, is the importance ratio of index i and index j of the kth large language model, and γ k is the authoritative weight coefficient of the kth large language model, and n is the matrix order;

[0025] In the above technical solution, the calculation formula of the consistency ratio CR in step 3 is:

[0026]

[0027] In the formula, RI is the random consistency index corresponding to the order, and CI is the consistency index.

[0028] The calculation formula of the consistency index CI is:

[0029]

[0030] In the formula, λ max is the maximum eigenvalue of the subjective weight vector W (x) of each level, and n is the matrix order.

[0031] In the above technical solution, the fire news collection agent in step 4 is constructed based on natural language processing technology, including a data acquisition module, a text preprocessing module, an information extraction module and a weight correction module.

[0032] In the above technical solution, according to the comprehensive score S final in step 5, the fire risk level of the electric vehicle is divided, and the preset fire emergency suggestion is generated, and a visual report is generated;

[0033] The electric vehicle fire risk level includes three levels of high risk, medium risk and low risk, wherein:

[0034] High risk: Sfinal <60 points, the system automatically generates fire emergency recommendations, including real-time monitoring, increasing insulation measures, and replacing high-risk components in advance;

[0035] Medium risk: 60≤S final <80 points, provide routine safety guidance and regular detection recommendations;

[0036] Low risk: S final ≥80 points, recommend regular maintenance cycles.

[0037] In the above technical solution, the comprehensive score S final in step 5 is calculated according to the following formula:

[0038] S final = S base -ΔG

[0039] In the formula, S base is the basic score of the electric vehicle to be evaluated, and ΔG is the vehicle model deduction item.

[0040] The calculation formula of the vehicle model deduction item ΔG is:

[0041] ΔG = c × (f-T)

[0042] In the formula, f is the actual frequency, T is the threshold value, and c is the penalty coefficient.

[0043] The calculation formula of the basic score S base of the electric vehicle to be evaluated is:

[0044]

[0045] In the formula, is the score of each primary indicator, is the weight of the primary indicator.

[0046] The calculation formula of the score of each primary indicator is:

[0047]

[0048] In the formula, is the weight of the secondary indicator relative to its primary indicator h, is the score of each secondary indicator.

[0049] The calculation formula of the score of each secondary indicator is:

[0050]

[0051] In the formula, is the score of each tertiary indicator after correction, a weight of the mth tertiary index relative to its secondary index l;

[0052] the modified score of each tertiary index The calculation formula is:

[0053]

[0054] In the formula, is the actual scoring result of the mth tertiary index, and a m is a dynamic correction coefficient of the mth tertiary index score.

[0055] In another aspect of the present application, an electronic device comprises: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the quantitative decision method of the three-level evaluation system for electric vehicle fire safety.

[0056] In another aspect of the present application, computer executable instructions are stored, and the instructions are used to implement the quantitative decision method of the three-level evaluation system for electric vehicle fire safety when executed.

[0057] In another aspect of the present application, a computer program product comprises computer executable instructions, and the instructions are used to implement the quantitative decision method of the three-level evaluation system for electric vehicle fire safety when executed.

[0058] Compared with the prior art, the present application has the following beneficial effects:

[0059] 1. Multi-level index system and strict consistency control, filling the evaluation gap and ensuring stable evaluation. The present application constructs a three-level index system covering key dimensions such as battery safety, electrical safety and mechanical safety, and innovatively includes forward-looking indicators such as early warning system of thermal runaway, emergency response and protection, and redundant safety design into the evaluation framework, effectively filling many gaps in the existing system. The evaluation system of four-level structure (target layer-criterion layer-sub-criterion layer-scheme layer) is established, the weight vectors of primary, secondary and tertiary indicators are calculated using the geometric mean method, and the consistency check standard is tightened to CR<0.08 to ensure the logical consistency and stability of the multi-level index system. Through power method iterative correction of inconsistent matrix, it avoids the distortion of evaluation results caused by weight fluctuation, and provides a stable and scientific benchmark framework for subsequent dynamic correction.

[0060] 2. The application introduces a multi-expert scoring system simulated by a large language model, selects the top ten high-performance models from the authoritative list, combines the general knowledge base and the special knowledge base customized for the characteristics of each large language model, sets the expert identity and task through prompt word engineering, and realizes the quantitative scoring of each three-level indicator. This design not only ensures the professionalism of the scoring and the authority of the data basis, but also effectively reduces the subjective bias and irreproducibility in the traditional expert weighting method by recording the model version of each large language model, the content of the knowledge base, and the allocation of authoritative weights.

[0061] 3. The dynamic correction mechanism driven by public opinion realizes real-time response to emerging risks. The application designs a fire news collection agent based on natural language processing, which can automatically capture relevant news and accident reports every day, extract key information such as vehicle type, accident type, and fire cause, form a multi-dimensional reporting frequency matrix, and modify the AHP weight in real time according to the "double-layer conduction deduction mode". For high-frequency vehicles, generate a vehicle deduction item ΔG, and for high-frequency accident reasons, dynamically adjust the judgment matrix elements and generate a correction coefficient α, so that the comprehensive score can timely reflect the latest risk situation. This mechanism effectively overcomes the defects of long update cycle and strong lag in traditional evaluation system, realizes daily update of safety evaluation and immediate response to emergencies, and significantly improves the timeliness and adaptability of the system.

[0062] 4. Full-process automatic calculation and executable risk output, reducing labor costs and enhancing practicality. The application realizes full-link processing from index scoring, weight correction to comprehensive score and risk level output through automatic calculation, without manual intervention to complete large-scale data collection, analysis and evaluation. The final result not only includes quantitative risk score and grading results, but also generates fire emergency suggestions for different risk levels (such as heat management improvement plan for high-risk vehicles, regular safety inspection guidelines, etc.). Compared with the traditional method relying on manual weighting and offline summary, the application improves efficiency by more than 30%, and can directly provide structured formats (such as JSON, report documents) for regulatory agencies, automakers and consumers, realizing efficient transmission and landing application of safety evaluation results, and promoting the digital upgrade of safety management in the electric vehicle industry. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 The figure shows the quantitative decision-making method of the electric vehicle fire safety three-level evaluation system of the application.

[0064] Figure 2 The figure shows the process of scoring each indicator using a certain large language model. DETAILED DESCRIPTION

[0065] The application will be described in further detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely used to explain the application and are not used to limit the application.

[0066] Example 1

[0067] A quantitative decision-making method for a three-level evaluation system of electric vehicle fire safety, comprising the following steps:

[0068] Step 1, establish a three-level evaluation index system of electric vehicle safety, which is used to support the construction and subsequent calculation of the quantitative decision-making model of electric vehicle fire safety. This method combines the characteristics of fire accidents, safety risk distribution law, national standards and industry specifications to form a structured, measurable and operable multi-level index system.

[0069] 1.1 Overall framework of index system

[0070] Reference Figure 1 A quantitative decision-making method for a three-level evaluation system of electric vehicle fire safety, comprising the following steps:

[0071] Step 1, construct a three-level hierarchical evaluation index system, including:

[0072] Target layer: comprehensive evaluation of electric vehicle safety;

[0073] Criteria layer (first-level index): including battery safety, electrical safety and mechanical safety, to cover the main risk sources of power system, electrical system and vehicle structure in fire situation;

[0074] Sub-criteria layer (second-level and third-level indexes): the first-level index is subdivided into second-level indexes, and each second-level index is further refined into multiple quantifiable third-level indexes, and detection methods and judgment standards are specified for each third-level index.

[0075] 1.2 Setting of first-level index

[0076] Battery safety: covering the structural design, thermal management capability, protection function and abnormal response capability of power battery system. For example: battery pack system safety; battery protection and abnormal response function.

[0077] Electrical safety: covering insulation, protection, leakage and electrical protection of high-voltage system of the vehicle. For example: line design and insulation performance (insulation resistance ≥ national standard value, such as 100Ω / V or more as required by GB / T 18384); leakage and short circuit protection (with real-time leakage detection and automatic shutdown function).

[0078] Mechanical safety: covers the impact resistance and vibration resistance of the vehicle body and related structures. For example: mechanical structure design (battery pack structure is reasonable, mechanical connecting parts are reliable, etc.); thermal management and intelligent monitoring (thermal runaway early warning after collision, fire emergency response capability).

[0079] 1.3 Secondary indicators and tertiary indicators

[0080] Under each primary indicator, several secondary indicators are set, and further refined into tertiary indicators to ensure that the indicators are quantifiable, measurable and reproducible. For example: battery pack thermal safety characteristics: tertiary indicators include thermal time alarm timeliness, aging degradation safety margin, etc.; circuit design: tertiary indicators include circuit layout rationality, circuit fixation and protection, etc.; mechanical failure protection: tertiary indicators include vibration and impact protection, foreign matter intrusion protection, mechanical stress detection, etc.

[0081] 1.4 Test methods and judgment criteria

[0082] To ensure data consistency and result comparability, standardize test methods and judgment criteria are defined for each tertiary indicator:

[0083] Test method: specify test equipment model (such as thermal imager resolution, sampling frequency), environmental conditions (environmental temperature, humidity), test steps (heating rate, impact energy level), etc.;

[0084] Judgment criteria: refer to the qualified range specified in GB, ISO and other standards, for example: insulation resistance should not be less than 100Ω / V; thermal runaway peak temperature ≤800℃ is judged as "excellent" grade.

[0085] The electric vehicle safety three-level evaluation index system refers to Table 1-3.

[0086] Table 1 Battery safety evaluation index

[0087]

[0088]

[0089]

[0090] Table 2 Electrical safety evaluation index

[0091]

[0092] Table 3 Mechanical safety evaluation index

[0093]

[0094]

[0095] Step 2: This step aims to simulate senior experts in the field of electric vehicle fire safety evaluation using large language models. Each index in the first, second, and third level indicators in Step 1 will be quantitatively scored. To ensure the scientificity, professionalism, and traceability of the scoring, refer to the following requirements: Figure 2 For example, using a certain large language model, the following steps are included:

[0096] 2.1 Expert role and prompt word construction. Design the core prompt word framework and set each large language model as a specific expert identity, such as "China Electric Vehicle Safety Committee Chief Expert" or "National Laboratory Specialized Safety Evaluation Consultant". Clearly define their task as: under the premise of independence and impartiality, quantitatively score each first, second, and third level indicator according to the 1-10 point system. Embed necessary citation requirements in the prompt words to enable each large language model to evaluate based on authoritative information such as the Ministry of Industry and Information Technology safety standards, laboratory test data, historical accident cases, and technical white papers. Provide scoring reasons and confidence levels in the output to facilitate subsequent tracking and auditing.

[0097] Core prompt word example:

[0098] You are a senior expert with rich experience in the field of new energy vehicle safety, and now you need to score the importance of each level indicator in the "Electric Vehicle Fire Safety Three-Level Evaluation System". The evaluation system includes:

[0099] First-level indicators: battery safety, electrical safety, mechanical safety;

[0100] Second-level indicators: such as battery pack system safety, circuit design and insulation performance, etc.

[0101] Third-level indicators: such as early warning of thermal runaway, intelligent monitoring of mechanical stress, fire emergency response, etc.

[0102] Please score each first, second, and third level indicator according to the 1-10 point system based on general and specialized knowledge bases, combined with real-time network data (latest accident news, technology updates, public opinion information). The score should reflect the importance of the indicator to the overall fire safety, with a higher score representing higher importance. Additional requirements:

[0103] Ensure the explainability of the scoring process, and each score should have corresponding evidence sources;

[0104] Return "N / A" when data is insufficient and explain the reason;

[0105] Give higher attention to recent hot risk events (such as the frequency of battery thermal runaway accidents in the past year);

[0106] Avoid subjective language in the output, and data and reasons should be based on facts.

[0107] 2.2 Large language model selection and authoritative weight distribution. Referring to the large language model (referred to as large model) performance list (such as Compass Arena), select the top ten large models in terms of comprehensive performance (such as Qwen3-235B-A22B, Spark-X1, GPT-4o-20241120, etc.), and assign the authoritative weight coefficient γ according to the ranking k (The first weight is 0.182, the last weight is 0.018, and the intermediate model is distributed in equal difference), wherein the high-ranking model is given a higher expert qualification role in the prompt word to strengthen its guiding role in the comprehensive scoring.

[0108] 2.3 External knowledge base mechanism. In order to improve the professional judgment ability of the model in the field of electric vehicle fire safety, a double-layer external mechanism of general knowledge base + special knowledge base is adopted:

[0109] General knowledge base: suitable for all ten large language models, including but not limited to national and international electric vehicle safety standards (such as GB / T, ISO standards), annual industry technical white papers, industry annual reports, previous vehicle recall announcements, publicly available fire accident statistics and cause analysis reports, standardized test method explanations, etc., to ensure the consistency and authority of the large model in the general field of information.

[0110] Special knowledge base: according to the context processing ability, information preference and reasoning advantage of different large models, each large model is equipped with differentiated deep industry data. For example:

[0111] Accident case database (subdivided by vehicle type, year, and accident type), suitable for large models with strong long context processing ability, so that they can associate more accident characteristics in one reasoning.

[0112] High-resolution laboratory test raw data set (including temperature curve, gas concentration change, structure stress data, etc.), suitable for large models with outstanding numerical reasoning ability, for fine-grained index analysis.

[0113] Policy and regulation change records and technical standard revision history, suitable for large models with strong legal and regulatory reasoning ability, for evaluating regulatory compliance and compliance risks.

[0114] Overseas market accident and recall event comparison data, suitable for large models with strong multilingual processing ability, for providing international comparison reference.

[0115] The special knowledge base content is obtained through industry association, regulatory data interface, laboratory test project results, academic papers and third-party safety evaluation agency reports, and is formatted and compressed before inputting the large model to match the context length limit and optimal information density of the target large model.

[0116] 2.4 Parallel execution and result aggregation. After loading the respective general and specialized knowledge bases, the selected ten large models independently complete the quantitative scoring of all three-level indicators based on the core prompt words and output the score value, reason, and confidence. To prevent "hallucinations" or unfounded inferences, the system sets an anti-hallucination mechanism in the prompt words. When a large model lacks sufficient evidence, it should output "N / A" and explain the reason for the lack of evidence in the reason.

[0117] 2.5 Score result processing. The system sequentially aggregates the scoring results of the ten large models to form a complete "multi-model expert scoring result set", i.e., ten independent expert model comprehensive scoring vectors. Each group's comprehensive scoring vector will serve as an important input for subsequent AHP judgment matrix construction and weight calculation, ensuring that the final weight and scoring system consider expert experience, objective data, and model diversity in terms of origin.

[0118] Step 3: Based on the three-level evaluation index system constructed in step 1, construct the quantitative decision model of the index evaluation system. The electric vehicle fire safety evaluation problem is divided into a four-level hierarchical structure: the target layer is "electric vehicle safety comprehensive evaluation", the criterion layer is the first-level evaluation index (including battery safety, electrical safety, mechanical safety), the sub-criterion layer is the two-level and three-level refined index under each first-level index, and the scheme layer is the initial weight vector system W constructed based on AHP. Specifically:

[0119] In step 2, the comprehensive scoring vectors of the ten large language models on the evaluation index system established in step 1 have been obtained; in this step, the authoritative weight coefficient γ k of each model is further introduced, and the pairwise comparison results of the ten large models are fused through weighted geometric mean to construct the comprehensive judgment matrix A, and the subjective weight vector W (x) of each level index is calculated using the Analytic Hierarchy Process (AHP) as the basis for subsequent consistency check and dynamic correction.

[0120] First, the judgment matrix between each layer is constructed according to the 1-9 scale method, and the matrix elements reflect the relative importance of two adjacent indicators under the upper-level indicator target. The pairwise comparison results of each large model are used to construct the judgment matrix, and the authoritative weight coefficient of the kth large language model in step 2 is γ k The importance ratio of each level index i and index j is Then the comprehensive judgment matrix element a ij is calculated as:

[0121]

[0122] After obtaining the comprehensive judgment matrix A = (a ij ) n×nAfter that, the geometric mean method is used to calculate the subjective weight vector W (x) = [w1, w2, …, w n ], where x is the index level, the geometric mean method is less sensitive to extreme values than the arithmetic mean method, and can significantly improve the stability of the weight.

[0123] wherein the calculation formula of the subjective weight w i of the index i is:

[0124]

[0125] To ensure the logical consistency of the judgment matrix, the maximum eigenvalue λ (x) , the consistency index CI and the consistency ratio CR of the subjective weight vector W max of the first-level, second-level and third-level indexes are calculated respectively:

[0126]

[0127] wherein n is the order of the matrix, and RI is the random consistency index corresponding to the order. According to the characteristics of the four-level index system, the standard of traditional CR<0.1 is tightened to CR<0.08 in this embodiment. If CR≥0.08, the power method is used to iteratively correct the subjective weight vector W (x) until the subjective weight vector converges and meets the consistency requirement. The iteration formula of the subjective weight vector W (x) is:

[0128]

[0129] wherein W is the subjective weight vector of the tthiteration, and represents the normalization operation.

[0130] The final stable and consistent initial weight vector system W composed of the weight of each level index is obtained, wherein W = [W (1) , W (2) , W (3) ], wherein W (1) is the first-level index weight vector, W (2) is the second-level index weight vector, and W (3) is the third-level index weight vector, and the storage format is shown in Table 4, which is used for subsequent dynamic public opinion correction and comprehensive score calculation.

[0131] Table 4 Storage form of the initial weight vector system W

[0132]

[0133] Step 4: After obtaining the initial weight vector system W of the electric vehicle fire safety evaluation system in step 3, introduce a fire news collection agent to realize dynamic correction of AHP subjective weight vector based on public opinion information, ensure that the evaluation system can sensitively capture emerging safety risks, and realize dynamic optimization of the online monitoring and evaluation system of electric vehicle fire risk.

[0134] The fire news collection agent is constructed based on natural language processing (NLP) technology and mainly consists of a data acquisition module, a text preprocessing module, an information extraction module, and a weight correction module.

[0135] The data acquisition module periodically accesses news portals, industry association announcement platforms, and accident reports published by government regulatory departments through a pre-set network crawler script and an open data interface (API) to capture unstructured raw text data related to electric vehicle fires. The collection frequency can be set by day or hour, for example, a full update is performed once every day at 0 o'clock in the UTC+8 time zone.

[0136] The text preprocessing module performs denoising, word segmentation, stop word filtering, and format standardization processing on the raw text from the data acquisition module.

[0137] The information extraction module combines text classification and named entity recognition (NER) technology to automatically extract key information such as "vehicle model name", "accident type", "fire cause" from the text processed by the text preprocessing module, and store it in the database according to date, source, and keyword dimensions. Based on the extraction results, the system generates a report frequency matrix F pq :

[0138] F pq = report frequency of vehicle model p under accident cause q

[0139] The weight correction module adopts a double-layer conduction deduction mode of "global vehicle risk - specific cause risk":

[0140] Global vehicle risk correction: According to the report frequency matrix F pq generated by the information extraction module, the report frequency of a certain vehicle model within a pre-set time window is obtained. When the report frequency of a certain vehicle model within a pre-set time window exceeds the threshold value (the threshold value of this embodiment is set to 5 times), a vehicle model deduction item ΔG is generated, and its value can be calculated as ΔG = c × (f - T), where f is the actual frequency, T is the threshold value, and c is the penalty coefficient (such as 1.0 points / time).

[0141] Specific cause risk correction: When the frequency proportion of a certain type of accident cause exceeds the threshold value (such as 20%), the importance of the related third-level indicators in the AHP judgment matrix is increased by the accident cause amplification coefficient (1+δ) (such as δ is the accident cause amplification ratio, δ = 0.1), that is, the corresponding matrix element aij multiply by (1+δ), recalculate weight element w i and perform consistency check (require CR<0.08), get the subjective weight vector W of each level index after dynamic correction (x) ', and further get the subjective weight vector system W' after dynamic correction. For the third level index affected, generate the dynamic correction coefficient α of the third level index score based on the accident cause amplification ratio δ, which is used for subsequent score calculation.

[0142] Replace the initial subjective weight vector system W in step 3 with the subjective weight vector system W' after dynamic correction, ensure that the comprehensive risk assessment result reflects the latest safety situation in real time, and store it to the database together with the dynamic correction coefficient α of the third level index score and the vehicle deduction item ΔG.

[0143] Step 5, on the basis of the weight vector W of each level index after dynamic correction combined with public opinion information (x) ', the dynamic correction coefficient α of each related third level index m and the vehicle deduction item ΔG obtained in step 4, according to the evaluation content of each index in table 1-3, manually score each third level index based on actual vehicle in step 1, get the actual scoring result of vehicle According to the actual scoring result of vehicle Calculate the comprehensive risk score of the vehicle to be evaluated and output the risk level.

[0144] 5.1 Index score calculation:

[0145] First, calculate the score of each third level index after correction

[0146]

[0147] Where, is the actual scoring result of vehicle for the mth third level index (0-100 points), α m is the dynamic correction coefficient of the mth third level index.

[0148] Second, calculate the score of each second level index

[0149]

[0150] Where, is the weight of the mth third level index relative to its second level index l.

[0151] Finally, calculate the score of each first level index

[0152]

[0153] wherein, is the weight of the secondary indicator relative to its primary indicator h.

[0154] 5.2 Base score and final score:

[0155] The base score S of the electric vehicle to be evaluated is calculated as follows: base :

[0156]

[0157] wherein, is the weight of the primary indicator.

[0158] The calculation formula of the comprehensive score S is as follows: final

[0159] S final = S base - ΔG

[0160] wherein, ΔG is the vehicle model deduction item, output by step 4.

[0161] 5.3 Risk level division and result output:

[0162] The system divides the vehicle into three levels of high risk, medium risk and low risk according to the comprehensive score S final . Among them:

[0163] High risk: S final <60 points, the system automatically generates fire emergency suggestions, including real-time monitoring, increasing heat insulation measures, replacing high-risk components in advance, etc.

[0164] Medium risk: 60≤S final <80 points, provide routine safety guidance and regular detection suggestions;

[0165] Low risk: S final ≥80 points, it is recommended to follow the regular maintenance cycle.

[0166] 5.4 Data storage and visualization:

[0167] The final score S final , each indicator score, weight coefficient, risk level and suggestion are stored in the database in the form of structured records, and the example table structure is shown in Table 5:

[0168] Table 5 Storage form of final score S final , each indicator score, weight coefficient, risk level and suggestion

[0169]

[0170] ​

[0171] The system generates a visual report (column chart, radar chart, etc.) according to the final score, the score of each index, the weight coefficient, the risk level and the suggestion, and provides a PDF / HTML export function, so as to facilitate the risk analysis and decision-making reference of the regulatory authorities, manufacturing enterprises and users.

[0172] Embodiment 2

[0173] An electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement a quantitative decision method of a three-level evaluation system for fire safety of electric vehicles.

[0174] Embodiment 3

[0175] A computer readable storage medium, storing computer executable instructions, wherein the instructions, when executed, implement a quantitative decision method of a three-level evaluation system for fire safety of electric vehicles.

[0176] Embodiment 4

[0177] A computer program product, comprising computer executable instructions, wherein the instructions, when executed, implement a quantitative decision method of a three-level evaluation system for fire safety of electric vehicles.

[0178] The above only describes the preferred embodiments of the present application, and it should be pointed out that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should also be considered as the protection scope of the present application.

Claims

1. A quantitative decision-making method for a three-level fire safety evaluation system for electric vehicles, characterized in that, Includes the following steps: Step 1: Construct a three-level evaluation index system according to the target layer, criterion layer, and sub-criterion layer. The target layer is the comprehensive evaluation of electric vehicle safety, the criterion layer is the first-level index, and the sub-criterion layer is the second-level and third-level index. Step 2: Select multiple large language models to simulate experts and quantify and score each indicator in the first, second and third level indicators in Step 1. Each large language model outputs a comprehensive score vector. Step 3: Construct a quantitative decision-making model for the indicator evaluation system: Based on the comprehensive scoring vector of each major language model in Step 2, introduce the authority weight coefficient γ of each major language model. k By using a weighted geometric mean, the importance of any two indicators at each level of the large language model is compared pairwise to obtain the comprehensive judgment matrix element a. ij Based on a ij Construct a comprehensive judgment matrix A, and calculate the subjective weight vector W for each level of indicator based on A. (x) It then determines whether the consistency ratio CR meets the consistency requirements. If not, it uses a power-law iterative correction method to adjust W. (x) Thus, an initial weight vector system W that meets the consistency requirements is obtained; Step 4: Introduce the fire news gathering agent to generate a vehicle model deduction term ΔG and an accident cause amplification coefficient (1+δ). Use the accident cause amplification coefficient (1+δ) to evaluate the matrix element a in the comprehensive judgment matrix A from Step 3. ij Dynamic correction is performed to obtain the dynamically corrected subjective weight vector system W′, and to generate the dynamic correction coefficient α=1-δ for the three-level indicator scores; Step 5: Manually score each of the three-level indicators from Step 1 based on actual vehicles to obtain the actual vehicle scoring results. The third-level indicator score is calculated by combining the dynamic correction coefficient α obtained in step 4 with the third-level indicator score. The scores of the second-level and first-level indicators are calculated sequentially based on the third-level indicator scores and W′. Finally, the comprehensive score S is calculated by combining the vehicle model deduction item ΔG. final This serves as a comprehensive safety evaluation for electric vehicles.

2. The quantitative decision-making method for the three-level evaluation system of electric vehicle fire safety as described in claim 1, characterized in that, The primary indicators mentioned in step 1 include battery safety, electrical safety, and mechanical safety.

3. The quantitative decision-making method for the three-level fire safety evaluation system for electric vehicles as described in claim 1, characterized in that, The subjective weight vector W at each level mentioned in step 3 (x) Represented as: IN (x) =[w1,w2,…,w n ] In the formula, w i Let be the subjective weight of index i, and n be the matrix order; The subjective weight w of index i i The calculation formula is: In the formula, a ij These are the elements of the comprehensive judgment matrix for indicator i and indicator j; The comprehensive judgment matrix element a of indicator i and indicator j ij The calculation formula is: In the formula, γ is the importance ratio of indicator i to indicator j for the k-th large language model. k denoted as the authority weight coefficient of the k-th large language model, where n is the matrix order.

4. The quantitative decision-making method for the three-level evaluation system of electric vehicle fire safety as described in claim 1, characterized in that, The formula for calculating the consistency ratio CR in step 3 is as follows: In the formula, RI is the random consistency index of the corresponding order, and CI is the consistency index. The formula for calculating the consistency index CI is as follows: In the formula, λ max The subjective weight vector W at each level (x) The largest eigenvalue, where n is the matrix order.

5. The quantitative decision-making method for the three-level fire safety evaluation system for electric vehicles as described in claim 1, characterized in that, The fire news gathering agent described in step 4 is built based on natural language processing technology and includes a data acquisition module, a text preprocessing module, an information extraction module, and a weight correction module.

6. The quantitative decision-making method for the three-level evaluation system of electric vehicle fire safety as described in claim 1, characterized in that, In step 5, based on the comprehensive score S final Classify electric vehicle fire risk levels and generate preset fire emergency suggestions, while also generating visual reports; The electric vehicle fire risk level is divided into three levels: high risk, medium risk, and low risk. High risk: S final If the score is less than 60, the system will automatically generate fire emergency suggestions, including real-time monitoring, increasing insulation measures, and replacing high-risk components in advance. Medium risk: 60≤S final For scores below 80, we provide general safety guidance and recommendations for regular inspections. Low risk: S final If the score is ≥80, it is recommended to perform maintenance according to the regular maintenance cycle.

7. The quantitative decision-making method for the three-level fire safety evaluation system for electric vehicles as described in claim 1, characterized in that, The comprehensive score S mentioned in step 5 final The calculation formula is: S final =S base -ΔG In the formula, S base This represents the base score for the electric vehicle to be evaluated, with ΔG being the deduction for the vehicle model. The formula for calculating the deduction ΔG for the vehicle model is as follows: ΔG=c×(fT) In the formula, f is the actual frequency, T is the threshold, and c is the penalty coefficient; The base score of the electric vehicle to be evaluated base The calculation formula is: In the formula, Scoring for each primary indicator, Weights for primary indicators; The scores of each primary indicator The calculation formula is: In the formula, This represents the weight of the secondary indicator relative to its primary indicator h. Scoring for each secondary indicator; The scores of each secondary indicator The calculation formula is: In the formula, The revised scores for each of the three levels of indicators. Let be the weight of the m-th tertiary indicator relative to its secondary indicator l; The revised scores of each of the three-level indicators The calculation formula is: In the formula, Let α be the actual score of the vehicle for the m-th tertiary indicator. m is the dynamic correction coefficient for the score of the m-th tertiary indicator.

8. An electronic device, characterized in that, include: One or more processors; A memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the quantitative decision-making method for the three-level evaluation system of electric vehicle fire safety as described in claim 1.

9. A computer-readable storage medium, characterized in that, It stores computer-executable instructions, which, when executed, are used to implement the quantitative decision-making method of the three-level evaluation system for electric vehicle fire safety as described in claim 1.

10. A computer program product, characterized in that, The aforementioned computer program product includes computer-executable instructions, which, when executed, are used to implement the quantitative decision-making method for the three-level evaluation system of electric vehicle fire safety as described in claim 1.

Images