An aviation dynamic financial risk assessment method, device and computer equipment

Abstract

This invention discloses a method, apparatus, and computer equipment for dynamic financial risk assessment in aviation. The method includes: acquiring multi-source heterogeneous financial data throughout the entire aviation operation process; preprocessing the multi-source heterogeneous financial data to obtain standardized financial data; constructing an aviation financial risk assessment model based on a robust optimization BERT pre-training method (RoBERTa), introducing aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determining the model's core parameters; inputting the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, capturing the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk; and outputting the financial risk assessment result, which includes the financial risk level and the corresponding financial risk impact factors. This method can improve the accuracy of dynamic financial risk assessment in aviation and reduce the false alarm rate.

Description

Technical Field

[0001] This invention relates to the field of aviation management technology, and in particular to a method, apparatus and computer equipment for assessing dynamic financial risks in aviation. Background Technology

[0002] Existing aviation financial risk assessment technologies have significant limitations, making it difficult to meet the dynamic financial risk prevention and control needs of modern aviation operations. Traditional assessment methods rely on limited data sources, often depending on test data from the design phase or limited flight records, failing to fully integrate multi-source heterogeneous financial data such as real-time financial market data, policy documents, and profit fluctuation data from the operational phase. This leads to one-sided financial risk identification. Financial risk assessment models are mostly built on fixed mathematical formulas or empirical rules, constituting a static assessment model. They cannot adapt to the evolving financial risks brought about by new routes and new financial policies, and have a weak ability to capture nonlinear financial risks involving multiple coupled factors. Furthermore, traditional models are insufficient in processing unstructured data such as financial news texts and market analysis reports, making it difficult to uncover hidden early warning signs of financial risks. They also lack dynamic update mechanisms, resulting in poor timeliness of assessment results and warnings lagging behind the evolution of financial risks, failing to provide effective intervention in the early stages of financial losses. These problems lead to low accuracy and high false alarm rates in existing financial risk assessment technologies, making it difficult to support the refined and forward-looking management needs of aviation financial security. Summary of the Invention

[0003] In view of this, the purpose of this invention is to propose a method, apparatus and computer equipment for assessing dynamic financial risks in aviation, which can improve the accuracy of dynamic financial risk assessment in aviation and reduce the false alarm rate, so as to better support the refined and forward-looking management and control needs of aviation financial security.

[0004] According to one aspect of the present invention, a method for assessing dynamic financial risks in aviation is provided, comprising the following steps: S1: Acquire multi-source heterogeneous financial data throughout the entire aviation operation process, including structured data and unstructured data; S2: Preprocess the multi-source heterogeneous financial data to obtain standardized financial data; S3: Construct an aviation financial risk assessment model based on the robust optimization BERT pre-training method RoBERTa, introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model; S4: Input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk; S5: Output financial risk assessment results, including the financial risk level and the corresponding financial risk impact factors.

[0005] The structured data in step S1 includes real-time airfare data, route revenue statistics, fuel cost fluctuation data, and financing interest rate data; the unstructured data includes financial news texts, aviation policy document texts, and market analysis report texts.

[0006] The preprocessing in step S2 includes: segmenting unstructured text data, performing dynamic masking and semantic encoding; and removing outliers, filling in missing values, and normalizing structured data to ensure that all data meets the input format requirements of the RoBERTa model.

[0007] In step S3, a weight matrix of key features in the aviation field is introduced to optimize the multi-head self-attention layer of the RoBERTa model. The key features in the aviation field include route revenue volatility, fuel price volatility, and policy regulation and financial risk.

[0008] Step S4 further includes: based on real-time financial risk assessment results and actual financial risk disposal feedback data, including the effectiveness of stop-loss measures and feedback on financing plan adjustments, constructing a dynamic model update mechanism to optimize the parameters of the RoBERTa model through online fine-tuning.

[0009] In step S4, the financial risk assessment dimensions include aviation financing financial risk, route operation revenue risk, and aviation asset valuation risk. The financial risk values ​​of each dimension are calculated by a weighted summation algorithm to form a comprehensive financial risk level.

[0010] In step S5, the output evaluation results also include a financial risk evolution trend curve and key financial risk source location information. The financial risk evolution trend curve is generated based on a time series prediction algorithm, and the key financial risk sources include financial risk sources such as declining revenue of specific routes, soaring fuel prices, and tightening policy regulation.

[0011] The process also includes step S6: setting a financial risk threshold range. When the overall financial risk level exceeds the preset threshold, a corresponding graded early warning signal is triggered. The graded early warning signal includes four levels: blue, yellow, orange, and red, which correspond to different levels of financial risk response strategies, including adjusting route pricing, optimizing financing structure, and activating stop-loss plans.

[0012] According to another aspect of the present invention, an aviation dynamic financial risk assessment device is provided, comprising: a data acquisition module, a preprocessing module, a parameter determination module, a dynamic assessment module, and an assessment result output module; The data acquisition module is used to acquire multi-source heterogeneous financial data throughout the entire aviation operation process. The multi-source heterogeneous financial data includes structured data and unstructured data. The preprocessing module is used to preprocess the multi-source heterogeneous financial data to obtain standardized financial data. The parameter determination module is used to construct an aviation financial risk assessment model based on the robust optimization BERT pre-training method RoBERTa, introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model. The dynamic assessment module is used to input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk. The assessment result output module is used to output financial risk assessment results, which include financial risk level and corresponding financial risk impact factors.

[0013] According to another aspect of the present invention, a computer device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the aviation dynamic financial risk assessment method as described in any of the preceding claims.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the aviation dynamic financial risk assessment method as described in any of the preceding claims.

[0015] It can be seen that the above solutions fundamentally address the core deficiencies of existing technologies: by integrating multi-source heterogeneous financial data to cover both structured and unstructured data related to aviation finance, this solution overcomes the limitation of traditional methods relying on a single data source, ensuring the comprehensiveness of financial risk identification; the RoBERTa model's bidirectional self-attention mechanism can accurately capture the semantic relationships in unstructured financial data, including regulatory signals in policy texts and risk warnings in market reports, solving the problem of insufficient processing capabilities of traditional models for unstructured financial data and improving the ability to identify hidden financial risks; by fine-tuning the model incorporating knowledge from the aviation finance field, the model adapts to the risk characteristics of aviation finance scenarios, significantly improving the accuracy of capturing nonlinear coupled financial risks, including the risk of fuel price and financing cost superposition, compared to empirical rule models; the dynamic assessment mode overcomes the drawbacks of poor timeliness of static models, achieving real-time synchronization between financial risk assessment and the operational status of aviation finance, effectively improving the accuracy of financial risk identification compared to traditional methods, providing technical support for early intervention in financial risks, and improving the accuracy of dynamic financial risk assessment in aviation while reducing false alarm rates, thus better supporting the refined and forward-looking management needs of aviation financial security.

[0016] Furthermore, the above solution, by clearly defining the scope of data collection across all dimensions of aviation finance, ensures that the data sources for financial risk assessment cover key aspects of aviation finance operations, including revenue, costs, financing, and policies. This addresses the problem of missed financial risk assessments caused by incomplete data coverage in traditional methods. Specifically, the introduction of fuel cost fluctuation data and financing interest rate data can accurately capture dynamic financial risks in the market environment; the inclusion of financial news texts and policy documents can uncover early signs of financial risks related to policy regulation and market expectations, effectively improving the comprehensiveness of financial risk assessment and providing a complete data foundation for subsequent accurate model analysis.

[0017] Furthermore, the above solutions enhance the model's deep understanding of financial semantics through dynamic masking of unstructured financial text data, including semantic mining of policy terms and market jargon, avoiding the loss of financial semantics in traditional text processing. Outlier removal and normalization of structured financial data effectively reduce the interference of noisy data, including abnormally fluctuating ticket prices, on the model's evaluation results, thus significantly improving data quality. The targeted design of the preprocessing workflow ensures that standardized financial data accurately matches the input requirements of the RoBERTa model, providing data assurance for the model's inference accuracy. Compared to solutions without targeted preprocessing, the error in financial risk assessment is effectively reduced.

[0018] Furthermore, the above solution addresses the issue of insufficient adaptability of the general-purpose RoBERTa model to specialized aviation finance scenarios by optimizing the weights of key features in the aviation finance field. This allows the model to focus on core risk points in aviation finance operations, including fluctuations in route revenue and the impact of policy adjustments. Compared to the unoptimized model, the sensitivity to identifying key financial risks such as declining route revenue, soaring fuel prices, and tightening policy controls is effectively improved. This effectively avoids feature extraction biases that occur in general-purpose models within specialized aviation finance scenarios, further enhancing the accuracy of financial risk assessment.

[0019] Furthermore, the above scheme overcomes the rigidity of traditional static models, which become fixed once trained. It enables the model to continuously learn new aviation financial risk characteristics, including risk changes brought about by new financial policies and the revenue fluctuation patterns of emerging routes, adapting to the dynamic changes in the aviation financial operating environment. For example, when new financial risk sources such as carbon trading policies are added, the model can quickly incorporate these risk factors into its assessment dimensions through online fine-tuning, effectively improving the accuracy of early warnings. Simultaneously, the closed-loop optimization mode continuously enhances the model's generalization ability, keeping the false alarm rate below a controllable range during long-term operation, ensuring the long-term effectiveness of the assessment model.

[0020] Furthermore, the above solutions break through the limitations of traditional assessment methods that focus solely on a single type of risk, achieving comprehensive coverage and refined quantification of core aviation financial risks, including financing, returns, and asset valuation. The separate output of financial risk values ​​across various dimensions meets the differentiated needs of different departments within airlines, such as finance, operations, and investment. For example, finance focuses on financing risk, while operations focus on return risk. The comprehensive financial risk rating provides management with a holistic decision-making basis, enabling financial risk management to shift from single-point prevention to comprehensive measures, effectively improving operational efficiency and reducing direct financial losses.

[0021] Furthermore, the above solutions address the limitations of traditional assessments, which only provide the current state of the financial risk evolution curve. This allows operators to anticipate the development of financial risks, including future quarterly route revenue risk trends, effectively advancing the warning time. The key financial risk source identification function avoids the blindness of financial risk management, helping staff quickly identify core issues, such as accurately locating a decline in revenue on a specific route as the main risk source and a surge in fuel prices as the core cost risk. This effectively shortens the time for handling financial risks and significantly improves emergency response efficiency, including quickly adjusting route pricing and optimizing fuel procurement plans.

[0022] Furthermore, the above-mentioned solutions, through a tiered early warning mechanism, enable precise policy implementation for aviation financial risk prevention and control, avoiding the resource waste or insufficient early warning issues caused by the traditional one-size-fits-all approach. The linkage between the four-level early warning signals and the aviation financial emergency response mechanism includes adjusting pricing for blue warnings and activating stop-loss plans for red warnings, ensuring that different levels of financial risks receive appropriate resources for handling, and shortening the time for handling major financial risks to within the preset time. At the same time, the clear financial risk threshold standards make the early warning more operational, comply with the standardized requirements of the aviation financial risk management system, effectively reduce the incidence of major aviation financial losses, and strengthen the first line of defense for aviation financial security. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating an embodiment of the aviation dynamic financial risk assessment method of the present invention; Figure 2 This is a schematic diagram of the structure of an embodiment of the aviation dynamic financial risk assessment device of the present invention; Figure 3 This is a schematic diagram of the structure of an embodiment of the computer device of the present invention. Detailed Implementation

[0025] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be particularly noted that the following embodiments are for illustrative purposes only and do not limit the scope of the invention. Similarly, the following embodiments are only some, not all, embodiments of the present invention, and all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This invention provides a method for assessing dynamic financial risks in aviation, which can improve the accuracy and false alarm rate of dynamic financial risk assessment in aviation, so as to better support the needs of refined and forward-looking management of aviation financial security.

[0027] Please see Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the aviation dynamic financial risk assessment method of the present invention. It should be noted that if substantially the same result is obtained, the method of the present invention is not necessarily identical. Figure 1 The illustrated process sequence is limited. For example... Figure 1 As shown, the method includes the following steps: S1: Acquire multi-source heterogeneous financial data throughout the entire aviation operation process, including structured data and unstructured data; S2: Preprocess the multi-source heterogeneous financial data to obtain standardized financial data; S3: Construct an aviation financial risk assessment model based on robust optimization BERT pre-training method (RoBERTa), introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model; S4: Input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk; S5: Output financial risk assessment results, including the financial risk level and the corresponding financial risk impact factors.

[0028] In this embodiment, the solution uses the RoBERTa pre-trained model as its core. It first integrates structured data (such as ticket price data and financing interest rate data) and unstructured data (such as policy document texts and market analysis reports) in aviation finance operations. After standardized preprocessing, the RoBERTa model is fine-tuned using knowledge in the field of aviation finance. Through its bidirectional self-attention mechanism, the inherent correlation of multi-dimensional financial risk factors (such as the correlation between policy regulation and route revenue, and the coupling relationship between fuel costs and financing costs) is explored, so as to realize real-time dynamic aviation financial risk assessment and result output.

[0029] It can be seen that in this embodiment, the solution fundamentally solves the core defects of the existing technology: by integrating multi-source heterogeneous financial data, it breaks through the limitation of single data source in traditional methods, ensuring the comprehensiveness of financial risk identification; the bidirectional self-attention mechanism of the RoBERTa model can accurately capture the semantic associations of unstructured financial data, solving the problem of insufficient processing capability of traditional models for unstructured financial data, and improving the ability to identify hidden financial risks; the model fine-tuning combined with knowledge in the aviation finance field makes the model adaptable to the risk characteristics of aviation finance scenarios, and significantly improves the accuracy of capturing nonlinear coupled financial risks compared with empirical rule models; the dynamic evaluation mode overcomes the shortcomings of poor timeliness of static models, realizes real-time synchronization of financial risk assessment and aviation finance operation status, effectively improves the accuracy of financial risk identification compared with traditional methods, provides technical support for early intervention in financial risks, and can improve the accuracy of dynamic financial risk assessment in aviation and reduce false alarm rate, so as to better support the refined and forward-looking management and control needs of aviation financial security.

[0030] In this embodiment, the structured data in step S1 includes real-time airfare data, route revenue statistics, fuel cost fluctuation data, and financing interest rate data; the unstructured data includes financial news texts, aviation policy document texts, and market analysis report texts.

[0031] In this embodiment, the specific types of multi-source heterogeneous data in the aviation finance field are clearly defined. Structured data covers quantitative indicators of the core elements of aviation finance (revenue, cost, financing), while unstructured data focuses on text records of scenarios such as policy regulation, market expectations, and industry analysis, so as to realize the data collection of all elements of aviation finance operation.

[0032] It can be observed that, in this embodiment, by clearly defining the scope of data collection across all dimensions of aviation finance, the data sources for financial risk assessment are ensured to cover key aspects of aviation finance operations, thus resolving the problem of missed financial risk assessments caused by incomplete data coverage in traditional methods. Specifically, the introduction of fuel cost fluctuation data and financing interest rate data can accurately capture dynamic financial risks in the market environment; the inclusion of financial news texts and policy document texts can uncover early signs of financial risks related to policy regulation and market expectations, effectively improving the comprehensiveness of financial risk assessment and providing a complete data foundation for subsequent accurate model analysis.

[0033] In this embodiment, the preprocessing in step S2 includes: performing word segmentation, dynamic masking, and semantic encoding on unstructured text data; and performing outlier removal, missing value completion, and normalization on structured data to ensure that all data meets the input format requirements of the RoBERTa model.

[0034] In this embodiment, a differentiated preprocessing process is designed for the characteristics of different types of data in the aviation finance field: unstructured financial text data adopts RoBERTa-adapted word segmentation and dynamic masking strategies (such as accurate word segmentation of policy terms and financial terms) to preserve the integrity of financial semantics; structured financial data eliminates dimensional differences through data cleaning and normalization (such as normalizing data of different magnitudes such as ticket prices and interest rates) to ensure that the data quality meets the requirements of model training and inference.

[0035] It can be observed that, in this embodiment, the dynamic masking processing of unstructured financial text data enhances the model's ability to deeply understand financial semantics, avoiding the problem of lost financial semantics in traditional text processing. Outlier removal and normalization of structured financial data effectively reduce the interference of noisy data on the model evaluation results, thus significantly improving data quality. The targeted design of the preprocessing workflow ensures that standardized financial data can accurately adapt to the input requirements of the RoBERTa model, providing data assurance for the model's inference accuracy. Compared to solutions without targeted preprocessing, the error in financial risk assessment is effectively reduced.

[0036] In this embodiment, in step S3, a weight matrix of key features in the aviation field is introduced to optimize the multi-head self-attention layer of the RoBERTa model. The key features in the aviation field include route revenue volatility features, fuel price volatility features, and policy regulation and financial risk features.

[0037] In this embodiment, based on the multi-head self-attention mechanism of the RoBERTa model, professional knowledge in the field of aviation finance is incorporated to construct a key feature weight matrix, which assigns higher attention weights to core financial risk features such as route revenue volatility, fuel price volatility, and policy regulation, thereby optimizing the direction of financial feature extraction in the model.

[0038] It can be observed that, in this embodiment, by optimizing the weights of key features in the aviation finance field, the problem of insufficient adaptability of the general RoBERTa model to specialized aviation finance scenarios is solved, enabling the model to focus on the core risk points of aviation finance operations. Compared to the unoptimized model, the sensitivity to identifying key financial risks such as declining route revenue, soaring fuel prices, and tightening policy controls is effectively improved, effectively avoiding feature extraction biases that occur in general models in specialized aviation finance scenarios, and further enhancing the accuracy of financial risk assessment.

[0039] In this embodiment, step S4 further includes: constructing a dynamic update mechanism for the model based on real-time financial risk assessment results and actual financial risk disposal feedback data (such as the effectiveness of stop-loss measures and feedback on financing plan adjustments), and optimizing the parameters of the RoBERTa model through online fine-tuning.

[0040] In this embodiment, a closed-loop mechanism of assessment-feedback-optimization is established in the field of aviation finance. The real-time financial risk assessment results are linked with the data on the effectiveness of subsequent financial risk disposal. When a new financial risk event occurs (such as sudden policy regulation) or the characteristics of financial risk change (such as changes in the fluctuation pattern of fuel prices), the model is automatically fine-tuned online to update the model parameters to adapt to the new evolution pattern of financial risk.

[0041] As can be seen, in this embodiment, the dynamic update mechanism overcomes the rigidity of traditional static models that become fixed once trained, enabling the model to continuously learn new aviation financial risk characteristics and adapt to the dynamic changes in the aviation financial operating environment. For example, when new financial risk sources such as carbon trading policies are added, the model can quickly incorporate the assessment dimensions of this risk factor through online fine-tuning, effectively improving the early warning accuracy. At the same time, the closed-loop optimization mode continuously enhances the model's generalization ability, keeping the false alarm rate below a controllable range in long-term operation, ensuring the long-term effectiveness of the assessment model.

[0042] In this embodiment, in step S4, the financial risk assessment dimensions include aviation financing financial risk, route operation revenue risk, and aviation asset valuation risk. The financial risk values ​​of each dimension are calculated by a weighted summation algorithm to form a comprehensive financial risk level.

[0043] In this embodiment, aviation financial risk is divided into three core dimensions: financing, returns, and asset valuation. Differentiated financial risk assessment indicator systems are designed for each dimension (e.g., financing risk includes indicators such as interest rate fluctuations and debt repayment capacity, while return risk includes indicators such as ticket price fluctuations and passenger flow scale). The weight of each dimension is determined by the analytic hierarchy process (combining the experience of aviation finance experts with historical risk data). The comprehensive financial risk level is obtained by weighted summation, thereby achieving a quantitative assessment of multi-dimensional aviation financial risk.

[0044] It can be seen that this embodiment breaks through the limitations of traditional assessment methods that focus solely on a single type of risk, achieving comprehensive coverage and refined quantification of core aviation financial risks. The separate output of financial risk values ​​for each dimension meets the differentiated needs of different departments within the airline, and the comprehensive financial risk level provides management with a basis for overall decision-making, enabling financial risk management to shift from single-point prevention and control to comprehensive measures, effectively improving operational efficiency and reducing direct financial losses.

[0045] In this embodiment, the output evaluation results in step S5 also include the financial risk evolution trend curve and the location information of key financial risk sources. The financial risk evolution trend curve is generated based on a time series prediction algorithm.

[0046] In this embodiment, based on the output of financial risk level, historical financial risk data and real-time assessment results are analyzed through time series prediction algorithms to generate an evolution trend curve of aviation financial risk over a future period (such as the trend of route revenue risk over the next 6 months). At the same time, the feature attribution analysis function of the RoBERTa model is used to locate the key financial risk sources and impact paths that lead to the current financial risk level (such as a policy adjustment that leads to a decline in route revenue, which in turn leads to an increase in comprehensive financial risk).

[0047] It can be seen that in this embodiment, the financial risk evolution trend curve solves the limitation of traditional assessments that can only provide the current state, enabling operators to predict the development trend of financial risks in advance and effectively advance the warning time; the key financial risk source positioning function avoids the blindness of financial risk management and control, helps staff to quickly identify core issues, effectively shortens the financial risk disposal time, and greatly improves emergency response efficiency.

[0048] In this embodiment, step S6 is also included: setting a financial risk threshold range. When the comprehensive financial risk level exceeds the preset threshold, a corresponding graded early warning signal is triggered. The graded early warning signal includes four levels: blue, yellow, orange, and red.

[0049] In this embodiment, based on aviation finance industry safety standards and airline historical financial risk data, four levels of financial risk threshold ranges are set, corresponding to different financial emergency response levels (blue warning: minor risk, adjust pricing strategy; yellow warning: general risk, optimize financing structure; orange warning: significant risk, reduce non-core routes; red warning: major risk, activate stop-loss plan). When the comprehensive financial risk level or a single dimension of financial risk value exceeds the corresponding threshold, an early warning signal is automatically triggered and simultaneously pushed to relevant departments of the airline such as finance, operations, and decision-making.

[0050] It can be observed that in this embodiment, a tiered early warning mechanism enables precise measures for aviation financial risk prevention and control, avoiding the resource waste or insufficient early warning caused by the traditional one-size-fits-all approach. The linkage between the four-level early warning signals and the emergency response mechanism ensures that different levels of financial risks receive appropriate resources for handling, shortening the handling time for major financial risks to within a preset time. At the same time, clear threshold standards make the early warning more operable, comply with the standardized requirements of the aviation financial risk management system, effectively reduce the incidence of major aviation financial losses, and strengthen the first line of defense for aviation financial security.

[0051] Please see Figure 2 , Figure 2 This is a schematic diagram of an embodiment of the aviation dynamic financial risk assessment device of the present invention. In this embodiment, the aviation dynamic financial risk assessment device 20 includes a data acquisition module 21, a preprocessing module 22, a parameter determination module 23, a dynamic assessment module 24, and an assessment result output module 25.

[0052] The data acquisition module 21 is used to acquire multi-source heterogeneous financial data in the entire aviation operation process, including structured data and unstructured data. The preprocessing module 22 is used to preprocess the multi-source heterogeneous financial data to obtain standardized financial data; The parameter determination module 23 is used to construct an aviation financial risk assessment model based on the robust optimization BERT pre-training method RoBERTa, introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model. The dynamic assessment module 24 is used to input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk. The assessment result output module 25 is used to output financial risk assessment results, which include financial risk level and corresponding financial risk impact factors.

[0053] Each unit module of the aviation dynamic financial risk assessment device 20 can execute the corresponding steps in the above method embodiment, so the details of each unit module will not be elaborated here. Please refer to the description of the corresponding steps above for details.

[0054] This invention also provides a computer device, such as... Figure 3 As shown, it includes: at least one processor 31; and a memory 32 communicatively connected to at least one processor 31; wherein the memory 32 stores instructions executable by at least one processor 31, the instructions being executed by at least one processor 31 to enable at least one processor 31 to execute the above-described aviation dynamic financial risk assessment method.

[0055] The memory 32 and processor 31 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 31 and memory 32. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 31 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 31.

[0056] Processor 31 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 32 can be used to store data used by processor 31 during operation.

[0057] The present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the above-described method embodiments.

[0058] It can be seen that the above solutions fundamentally address the core deficiencies of existing technologies: by integrating multi-source heterogeneous financial data to cover both structured and unstructured data related to aviation finance, this solution overcomes the limitation of traditional methods relying on a single data source, ensuring the comprehensiveness of financial risk identification; the RoBERTa model's bidirectional self-attention mechanism can accurately capture the semantic relationships in unstructured financial data, including regulatory signals in policy texts and risk warnings in market reports, solving the problem of insufficient processing capabilities of traditional models for unstructured financial data and improving the ability to identify hidden financial risks; by fine-tuning the model incorporating knowledge from the aviation finance field, the model adapts to the risk characteristics of aviation finance scenarios, significantly improving the accuracy of capturing nonlinear coupled financial risks, including the risk of fuel price and financing cost superposition, compared to empirical rule models; the dynamic assessment mode overcomes the drawbacks of poor timeliness of static models, achieving real-time synchronization between financial risk assessment and the operational status of aviation finance, effectively improving the accuracy of financial risk identification compared to traditional methods, providing technical support for early intervention in financial risks, and improving the accuracy of dynamic financial risk assessment in aviation while reducing false alarm rates, thus better supporting the refined and forward-looking management needs of aviation financial security.

[0059] Furthermore, the above solution, by clearly defining the scope of data collection across all dimensions of aviation finance, ensures that the data sources for financial risk assessment cover key aspects of aviation finance operations, including revenue, costs, financing, and policies. This addresses the problem of missed financial risk assessments caused by incomplete data coverage in traditional methods. Specifically, the introduction of fuel cost fluctuation data and financing interest rate data can accurately capture dynamic financial risks in the market environment; the inclusion of financial news texts and policy documents can uncover early signs of financial risks related to policy regulation and market expectations, effectively improving the comprehensiveness of financial risk assessment and providing a complete data foundation for subsequent accurate model analysis.

[0060] Furthermore, the above solutions enhance the model's deep understanding of financial semantics through dynamic masking of unstructured financial text data, including semantic mining of policy terms and market jargon, avoiding the loss of financial semantics in traditional text processing. Outlier removal and normalization of structured financial data effectively reduce the interference of noisy data, including abnormally fluctuating ticket prices, on the model's evaluation results, thus significantly improving data quality. The targeted design of the preprocessing workflow ensures that standardized financial data accurately matches the input requirements of the RoBERTa model, providing data assurance for the model's inference accuracy. Compared to solutions without targeted preprocessing, the error in financial risk assessment is effectively reduced.

[0061] Furthermore, the above solution addresses the issue of insufficient adaptability of the general-purpose RoBERTa model to specialized aviation finance scenarios by optimizing the weights of key features in the aviation finance field. This allows the model to focus on core risk points in aviation finance operations, including fluctuations in route revenue and the impact of policy adjustments. Compared to the unoptimized model, the sensitivity to identifying key financial risks such as declining route revenue, soaring fuel prices, and tightening policy controls is effectively improved. This effectively avoids feature extraction biases that occur in general-purpose models within specialized aviation finance scenarios, further enhancing the accuracy of financial risk assessment.

[0062] Furthermore, the above scheme overcomes the rigidity of traditional static models, which become fixed once trained. It enables the model to continuously learn new aviation financial risk characteristics, including risk changes brought about by new financial policies and the revenue fluctuation patterns of emerging routes, adapting to the dynamic changes in the aviation financial operating environment. For example, when new financial risk sources such as carbon trading policies are added, the model can quickly incorporate these risk factors into its assessment dimensions through online fine-tuning, effectively improving the accuracy of early warnings. Simultaneously, the closed-loop optimization mode continuously enhances the model's generalization ability, keeping the false alarm rate below a controllable range during long-term operation, ensuring the long-term effectiveness of the assessment model.

[0063] Furthermore, the above solutions break through the limitations of traditional assessment methods that focus solely on a single type of risk, achieving comprehensive coverage and refined quantification of core aviation financial risks, including financing, returns, and asset valuation. The separate output of financial risk values ​​across various dimensions meets the differentiated needs of different departments within airlines, such as finance, operations, and investment. For example, finance focuses on financing risk, while operations focus on return risk. The comprehensive financial risk rating provides management with a holistic decision-making basis, enabling financial risk management to shift from single-point prevention to comprehensive measures, effectively improving operational efficiency and reducing direct financial losses.

[0064] Furthermore, the above solutions address the limitations of traditional assessments, which only provide the current state of the financial risk evolution curve. This allows operators to anticipate the development of financial risks, including future quarterly route revenue risk trends, effectively advancing the warning time. The key financial risk source identification function avoids the blindness of financial risk management, helping staff quickly identify core issues, such as accurately locating a decline in revenue on a specific route as the main risk source and a surge in fuel prices as the core cost risk. This effectively shortens the time for handling financial risks and significantly improves emergency response efficiency, including quickly adjusting route pricing and optimizing fuel procurement plans.

[0065] Furthermore, the above-mentioned solutions, through a tiered early warning mechanism, enable precise policy implementation for aviation financial risk prevention and control, avoiding the resource waste or insufficient early warning issues caused by the traditional one-size-fits-all approach. The linkage between the four-level early warning signals and the aviation financial emergency response mechanism includes adjusting pricing for blue warnings and activating stop-loss plans for red warnings, ensuring that different levels of financial risks receive appropriate resources for handling, and shortening the time for handling major financial risks to within the preset time. At the same time, the clear financial risk threshold standards make the early warning more operational, comply with the standardized requirements of the aviation financial risk management system, effectively reduce the incidence of major aviation financial losses, and strengthen the first line of defense for aviation financial security.

[0066] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection of apparatuses or units, and may be electrical, mechanical, or other forms.

[0067] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0068] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0069] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0070] The above description is only a part of the embodiments of the present invention and does not limit the scope of protection of the present invention. Any equivalent device or equivalent process transformation made based on the content of the present invention specification and drawings, or direct or indirect application in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for assessing dynamic financial risks in aviation, characterized in that, Includes the following steps: S1: Acquire multi-source heterogeneous financial data throughout the entire aviation operation process, including structured data and unstructured data; S2: Preprocess the multi-source heterogeneous financial data to obtain standardized financial data; S3: Construct an aviation financial risk assessment model based on the robust optimization BERT pre-training method RoBERTa, introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model; S4: Input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk; S5: Output financial risk assessment results, including the financial risk level and the corresponding financial risk impact factors.

2. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, The structured data in step S1 includes real-time airfare data, route revenue statistics, fuel cost fluctuation data, and financing interest rate data; the unstructured data includes financial news texts, aviation policy document texts, and market analysis report texts.

3. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, The preprocessing in step S2 includes: segmenting, dynamically masking, and semantically encoding unstructured text data; removing outliers, filling in missing values, and normalizing structured data to ensure that all data meets the input format requirements of the RoBERTa model.

4. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, In step S3, a weight matrix of key features in the aviation field is introduced to optimize the multi-head self-attention layer of the RoBERTa model. The key features in the aviation field include route revenue volatility, fuel price volatility, and policy regulation and financial risk.

5. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, Step S4 further includes: based on real-time financial risk assessment results and actual financial risk disposal feedback data, including the effectiveness of loss-stopping measures and feedback on financing plan adjustments, constructing a dynamic model update mechanism and optimizing the parameters of the RoBERTa model through online fine-tuning.

6. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, In step S4, the financial risk assessment dimensions include aviation financing financial risk, route operation revenue risk, and aviation asset valuation risk. The financial risk values ​​of each dimension are calculated by a weighted summation algorithm to form a comprehensive financial risk level.

7. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, In step S5, the output evaluation results also include the financial risk evolution trend curve and the location information of key financial risk sources. The financial risk evolution trend curve is generated based on a time series prediction algorithm, and the key financial risk sources include the decline in revenue of specific routes, the surge in fuel prices, and the source of financial risks caused by tightening policy regulation.

8. The aviation dynamic financial risk assessment method according to claim 1, characterized in that, It also includes step S6: setting a financial risk threshold range. When the overall financial risk level exceeds the preset threshold, a corresponding graded early warning signal is triggered. The graded early warning signal includes four levels: blue, yellow, orange, and red, which correspond to different intensities of financial risk response strategies, including adjusting route pricing, optimizing financing structure, and activating stop-loss plans.

9. An aviation dynamic financial risk assessment device, characterized in that, include: The module includes a data acquisition module, a preprocessing module, a parameter determination module, a dynamic evaluation module, and an evaluation result output module. The data acquisition module is used to acquire multi-source heterogeneous financial data throughout the entire aviation operation process. The multi-source heterogeneous financial data includes structured data and unstructured data. The preprocessing module is used to preprocess the multi-source heterogeneous financial data to obtain standardized financial data. The parameter determination module is used to construct an aviation financial risk assessment model based on the robust optimization BERT pre-training method RoBERTa, introduce aviation domain knowledge to pre-train and fine-tune the RoBERTa model, and determine the core parameters of the model. The dynamic assessment module is used to input the standardized financial data into the fine-tuned RoBERTa aviation financial risk assessment model, and capture the correlation characteristics between financial risk factors through the model's bidirectional self-attention mechanism to achieve dynamic assessment of aviation financial risk. The assessment result output module is used to output financial risk assessment results, which include financial risk level and corresponding financial risk impact factors.

10. A computer device, characterized in that, include: At least one processor; And a memory communicatively connected to at least one processor; wherein the memory stores instructions executable by at least one processor, the instructions being executed by at least one processor to enable at least one processor to perform the aviation dynamic financial risk assessment method as described in any one of claims 1 to 8.

Images