A big data-based enterprise financial risk dynamic identification and early warning method

Abstract

The present application relates to the field of financial risk early warning, and more particularly to a kind of enterprise financial risk dynamic identification and early warning method based on big data, the method comprises: obtaining the basic characteristic parameter set for risk transmission modeling;Through the nonlinear evaluation of the synchronism of enterprise operating cash flow fluctuation, obtain the industry cycle resonance factor for modifying risk transmission intensity;Through the space-time coupling evaluation of enterprise debt maturity structure and liquidity defense capability, obtain the space-time phase damping factor for modifying risk bearing capacity;Through the fusion adjustment of initial static risk transmission weight, obtain dynamic risk transmission matrix and execute iterative solution;Through the decoupling analysis of converged enterprise risk score, obtain dynamic classification risk early warning signal, to solve the problem that the existing graph iteration risk propagation algorithm based on static edge weight is distorted in risk assessment due to ignoring industry resonance and term mismatch in complex supply chain scenarios.

Description

Technical Field

[0001] This invention relates to the field of financial risk early warning technology, and in particular to a method for dynamic identification and early warning of corporate financial risks based on big data. Background Technology

[0002] In the field of corporate financial risk management and credit monitoring, with the widespread application of business models such as supply chain finance, corporate mutual guarantees, and related-party transactions, the risk correlation between enterprises is increasingly strengthened. Financial anomalies in a single enterprise often spread to other enterprises through transaction relationships, guarantee relationships, or fund transfers, thereby triggering chain defaults or even systemic risks. Therefore, how to model and assess the risk transmission process within a group of enterprises has become a key technical issue of concern for financial institutions and regulatory authorities. Existing technologies typically employ graph-based risk propagation analysis methods for identifying risks within a group of enterprises. These methods generally abstract enterprises as nodes in a graph structure, and the guarantee, lending, or transaction relationships between enterprises as directed edges. Risk propagation weights are then constructed to characterize the diffusion process of risk between nodes. In practice, risk propagation weights are often calculated based on static financial indicators such as the amount of guarantees, transaction amounts, or accounts receivable and payable between enterprises. An iterative propagation algorithm is used to update the risk status of each enterprise, ultimately obtaining a risk score or risk level for each enterprise.

[0003] However, in practical applications, the aforementioned risk propagation models based on static weights reveal significant shortcomings in complex financial scenarios. Firstly, these methods typically determine the intensity of risk transmission solely based on the scale of financial connections between enterprises, neglecting the synchronous changes in enterprise operating conditions over time. When upstream, downstream, or related enterprises are simultaneously in an industry downturn or experiencing the same macroeconomic shocks, their operating cash flows often exhibit highly synchronized fluctuations. In this case, the intensity of risk transmission between enterprises is significantly amplified, and risk weights constructed solely based on static financial scale cannot reflect this cyclical resonance effect, easily leading to an underestimation of systemic risk. Secondly, existing technologies typically assume that risk shocks are transmitted instantly and proportionally between enterprises, failing to fully consider the enterprises' own buffering capacity in both time and financial dimensions. In reality, an enterprise's ability to withstand external risk shocks is not only related to related relationships but also closely linked to its debt maturity structure, repayment schedule, and quick asset reserves. When a company's main debt is concentrated in the long term and it possesses strong short-term liquidity defense capabilities, it has a certain buffering and absorption capacity against upstream risk shocks. However, existing methods often fail to effectively distinguish such situations, easily misjudging companies with self-regulating capabilities as high-risk nodes, leading to distorted early warning results. Furthermore, existing risk propagation models lack a balancing mechanism between static risk relationships and dynamic risk characteristics during weight construction and propagation. Introducing a single dynamic adjustment factor can easily lead to drastic fluctuations in risk weights or the complete severing of risk paths, thus affecting the stability and reliability of risk assessment results. Therefore, how to dynamically adjust the intensity of risk transmission and risk-bearing capacity while maintaining the objective existence of corporate relationships remains a key unresolved issue in current technologies.

[0004] In summary, existing risk identification and early warning methods based on enterprise correlation diagrams have shortcomings in identifying industry cycle resonance, characterizing risk-bearing capacity, and dynamically adjusting risk weights. There is an urgent need for a technical solution for identifying and warning of enterprise financial risks that can comprehensively consider the synchronous characteristics of enterprise operating cash flow, debt maturity structure, and liquidity defense capabilities, and achieve stable and controllable dynamic adjustment during the risk propagation process. Summary of the Invention

[0005] In view of this, the present invention aims to propose a dynamic identification and early warning method for corporate financial risks based on big data, in order to solve the problem of risk assessment distortion caused by the neglect of industry resonance and maturity mismatch in complex supply chain scenarios by existing graph iterative risk propagation algorithms based on static edge weights.

[0006] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0007] A method for dynamic identification and early warning of enterprise financial risks based on big data, the method comprising:

[0008] Step S1: By collecting and preprocessing the basic data related to corporate finance and transactions, a set of basic feature parameters for risk transmission modeling is obtained;

[0009] Step S2: Obtain the industry cycle resonance factor for correcting the intensity of risk transmission by nonlinearly assessing the synchronicity of fluctuations in corporate operating cash flow;

[0010] Step S3: By conducting a spatiotemporal coupling assessment of the corporate debt maturity structure and liquidity defense capabilities, obtain the spatiotemporal phase damping factor used to correct the risk-bearing capacity;

[0011] Step S4: By fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor, the dynamic risk transmission matrix is ​​obtained and iterative solution is performed;

[0012] Step S5: Obtain dynamic classification risk warning signals by performing decoupling analysis on the converged enterprise risk scores.

[0013] Furthermore, the process of collecting and preprocessing basic financial and transaction-related data of enterprises to obtain a set of basic feature parameters for risk transmission modeling includes:

[0014] By connecting and synchronously collecting data from enterprise information systems, financial interfaces, and external credit data sources, basic data on enterprise finance and transactions are obtained. Specifically, for each target enterprise within the scope of risk identification, historical transaction records between enterprises, mutual guarantee contract amounts, accounts receivable balances, accounts payable balances, total net assets, and total liabilities are collected, and the above data are used as the basic data for constructing enterprise relationships.

[0015] By collecting the inflow and outflow records of enterprise operating funds in a time series manner, the target enterprise's net operating cash flow time series data in multiple consecutive statistical periods is obtained. The net operating cash flow time series data covers at least a preset length of historical time window to form raw cash flow data for characterizing the fluctuation characteristics of enterprise operations.

[0016] By collecting information on the asset and liability structure of enterprises, the total quick assets and total current liabilities of the target enterprise are obtained. The total quick assets include at least cash, trading financial assets and accounts receivable that can be quickly converted into cash. The total current liabilities include short-term loans, accounts payable and non-current liabilities due within one year. The corresponding data are used as the basic financial data to characterize the liquidity status of the enterprise.

[0017] By collecting details of each outstanding debt of a company, data on the principal amount and remaining maturity time of each debt of the target company can be obtained, forming data on the company's debt maturity structure. At the same time, by monitoring and collecting information from publicly disclosed information, judicial announcements and financial risk detection systems, information on default events and the time of default related to the company's credit status can be obtained.

[0018] After data collection is completed, outlier removal, missing value completion, and consistency verification are performed on the collected basic data related to corporate finance and transactions. Based on the processed data, quick ratio, weighted average remaining debt maturity, first-order difference data of net operating cash flow, and initial static risk transmission weights are calculated as the basic feature parameter set for risk transmission modeling.

[0019] Furthermore, the process of obtaining an industry cycle resonance factor for correcting the intensity of risk transmission through a nonlinear assessment of the synchronicity of fluctuations in corporate operating cash flow includes:

[0020] By performing directional consistency and amplitude correlation processing on the time-series differential data of operating cash flows of related enterprises, data representing the synchronicity of cash flow fluctuations are obtained.

[0021] By performing downside-sensitive weighting and nonlinear mapping on the cash flow fluctuation synchronicity characterization data, an industry cycle resonance factor is obtained to correct the intensity of risk transmission.

[0022] Furthermore, the step of obtaining cash flow fluctuation synchronization data by performing directional consistency and amplitude correlation processing on the time-series difference data of operating cash flows of related enterprises includes:

[0023] Set a historical time window; for any target company and its related companies, extract the time series data of net operating cash flow of the target company and its related companies in the historical time window from the set of basic feature parameters, perform first-order difference calculation on the net operating cash flow time data of adjacent statistical periods, and obtain the time series difference data of operating cash flow of the target company and its related companies.

[0024] By determining the sign consistency of the time-series difference data of the operating cash flow of the target company and its related companies in the same statistical period, we can identify whether the cash flow of the target company and its related companies changes in the same direction in the statistical period. Statistical periods with consistent cash flow changes are marked as synchronous change periods, and statistical periods with inconsistent cash flow changes are marked as asynchronous change periods.

[0025] Within the synchronous change period, the absolute value of the product of the time-series difference data of the operating cash flow of the target company and the related companies is used as the correlation assessment value of the cash flow fluctuation amplitude, and the correlation assessment value of the cash flow fluctuation amplitude is used as the amplitude correlation assessment data to characterize the degree of correlation between the cash flow fluctuation amplitude of the target company and the related companies within the statistical period.

[0026] Based on the distribution of synchronous change cycles of the target company and related companies within historical time windows and the corresponding amplitude correlation assessment data, the directional consistency judgment results and amplitude correlation assessment results within multiple statistical periods are accumulated and summarized to obtain cash flow fluctuation synchronization characterization data used to characterize the cash flow fluctuation synergy of the target company and related companies.

[0027] Furthermore, the process of obtaining an industry cycle resonance factor for correcting the intensity of risk transmission by performing downside-sensitive weighting and nonlinear mapping on the cash flow fluctuation synchronicity characterization data includes:

[0028] Within the historical time window, for the target company and its related companies, the operating cash flow time series difference data corresponding to each statistical period is used to determine the downward state, identify whether the target company and its related companies are simultaneously in a state of declining cash flow in the same statistical period, and mark the statistical period corresponding to the cash flow time series difference data of the target company and its related companies that are both negative in the same statistical period as the downward resonance period, and mark the other statistical periods as non-downward resonance periods.

[0029] For any given statistical period, the product of the time-series difference data of the operating cash flows of the target company and its related companies is used as the fluctuation correlation product assessment. The sign of the fluctuation correlation product assessment is determined to identify whether the cash flow fluctuations of the target company and its related companies are consistent within the statistical period. When the fluctuation correlation product assessment is positive, the square root of the absolute value of the fluctuation correlation product assessment is used as the basic correlation strength assessment for that statistical period. When the fluctuation correlation product assessment is negative, the square root of the absolute value of the fluctuation correlation product assessment, combined with sign reversal processing, is used as the basic correlation strength assessment for that statistical period.

[0030] For the assessment of basic correlation strength corresponding to the downtrend resonance cycle, a downtrend sensitive weight coefficient is introduced for weighting; for the assessment of basic correlation strength corresponding to non-downtrend resonance cycle, a benchmark weight coefficient is introduced for weighting to obtain weighted correlation strength assessment data for each statistical cycle.

[0031] Based on the weighted correlation strength assessment data corresponding to each statistical period within the historical time window, the weighted correlation strength assessment data within multiple statistical periods are accumulated and summarized. Combined with the overall fluctuation range of the time series difference data of the operating cash flow of the target company and related companies, the data is normalized to obtain the cyclical resonance strength assessment value of the target company and related companies within the historical time window. By performing exponential nonlinear mapping on the cyclical resonance strength assessment value, an industry cyclical resonance factor is obtained to correct the risk transmission strength between the target company and related companies.

[0032] Furthermore, the method of conducting a spatiotemporal coupled assessment of the corporate debt maturity structure and liquidity defense capabilities to obtain a spatiotemporal phase damping factor for correcting risk-bearing capacity includes:

[0033] By performing time phase difference analysis on corporate debt maturity structure data, we can obtain time buffer representation data of risk shocks.

[0034] By assessing a company's solvency through data on its quick assets and current liabilities, data representing the company's liquidity defense capabilities can be obtained.

[0035] By performing spatiotemporal coupling damping processing on time buffer characterization data and liquidity defense capability characterization data, a spatiotemporal phase damping factor for correcting risk absorption capability is obtained.

[0036] Furthermore, the step of obtaining time buffer representation data of risk shocks by performing time phase difference analysis on corporate debt maturity structure data includes:

[0037] For any target company, extract the principal amount and remaining maturity time of each outstanding debt from the set of basic feature parameters. Then, use the principal amount of each outstanding debt as the weight to calculate the weighted average remaining maturity time of the debt, and obtain the weighted average remaining maturity time of the debt for the target company.

[0038] For the target company's related companies, risk event timing data corresponding to the related companies are extracted from the basic feature parameter set. The risk event timing data includes the predicted default outbreak time of the related companies or the default time corresponding to the default events that have already occurred. By calculating the time difference between the target company's weighted average remaining debt maturity time and the risk event timing data of the related companies, time phase difference assessment data representing the relative chronological relationship between the target company's debt maturity time and the risk impact time of the related companies is obtained. The time phase difference assessment data is used as time buffer representation data to characterize the target company's buffering capacity against risk impacts from related companies in the time dimension.

[0039] Furthermore, the process of assessing a company's solvency by analyzing its quick assets and current liabilities to obtain data characterizing its liquidity defense capabilities includes:

[0040] For any target company, extract the total quick assets and total current liabilities data corresponding to the target company from the set of basic feature parameters. The total quick assets data includes at least cash, trading financial assets and accounts receivable that can be quickly converted into cash, and the total current liabilities data includes short-term loans, accounts payable and non-current liabilities due within one year.

[0041] By calculating the ratio between the total quick assets and the total current liabilities of the target company, using the total quick assets as the numerator and the total current liabilities as the denominator, the corresponding quick ratio assessment data of the target company is obtained.

[0042] When the target company's total current liabilities are zero, the target company's quick ratio assessment data is set to a preset high constant to indicate that the target company has strong short-term solvency.

[0043] The quick ratio assessment data corresponding to the target company is used as a liquidity defense capability representation data to characterize the target company's ability to defend against external risk shocks in terms of funding.

[0044] Furthermore, the step of performing spatiotemporal coupling damping processing on time buffer characterization data and liquidity defense capability characterization data to obtain a spatiotemporal phase damping factor for correcting risk absorption capability includes:

[0045] Set a time sensitivity normalization constant, and for any target company and its related companies, obtain time buffer characterization data and liquidity defense capability characterization data; by performing time sensitivity normalization processing on the time buffer characterization data, calculate the ratio of the time buffer characterization data to the time sensitivity normalization constant, and obtain normalized time difference assessment data that characterizes the degree of influence of time phase difference.

[0046] By performing a logarithmic mapping on the sum of the absolute values ​​of the time buffer characterization data and the natural constant, logarithmic evaluation data of the time buffer is obtained to characterize the scale effect of the time buffer. This logarithmic evaluation data is then multiplicatively modulated with liquidity defense capability characterization data and a preset liquidity effectiveness amplification coefficient to obtain time-coupling evaluation data of funds, characterizing the defense effectiveness of liquidity under time buffer conditions. Normalized time difference evaluation data is summed with time-coupling evaluation data to obtain spatiotemporal coupling driving force evaluation data. Finally, by performing an exponential nonlinear mapping on the spatiotemporal coupling driving force evaluation data and adding the mapping result to a constant 1 and taking the reciprocal, a spatiotemporal phase damping factor is obtained to correct the target company's ability to absorb risk shocks from related companies.

[0047] Furthermore, the process of fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor to obtain the dynamic risk transmission matrix and performing iterative solutions includes:

[0048] For any target company and its related companies, the initial static risk transmission weights between the target company and its related companies are extracted from the set of basic feature parameters, and the industry cycle resonance factor and the spatiotemporal phase damping factor are obtained.

[0049] A static risk retention coefficient is set and used as a rigid risk retention term. The difference between the constant 1 and the static risk retention coefficient is used as a dynamic adjustment weight. By multiplying the dynamic adjustment weight with the industry cycle resonance factor and the spatiotemporal phase damping factor, and then summing it with the static risk retention term, a fusion adjustment coefficient is obtained to adjust the initial static risk transmission weight. By multiplying the initial static risk transmission weight with the fusion adjustment coefficient, the dynamic risk transmission weight between the target enterprise and its related enterprises is obtained. Based on the dynamic risk transmission weights obtained between each target enterprise and its related enterprises, a dynamic risk transmission matrix is ​​constructed among all enterprise nodes.

[0050] Initialize the risk status value of each enterprise node, and perform risk iterative propagation operation based on the dynamic risk transmission matrix. In each iteration, the enterprise risk status value of the current round is superimposed and updated with the risk transmission result corresponding to the dynamic risk transmission matrix, and the updated risk status value is subject to upper limit constraint processing. Repeat the risk iterative propagation operation until the change of enterprise risk status value obtained by two adjacent iterations is less than the preset convergence threshold or the preset number of iterations is reached. The final converged set of enterprise risk status values ​​is taken as the enterprise risk score result.

[0051] Compared with the prior art, the present invention has the following advantages:

[0052] This invention presents a dynamic identification and early warning method for enterprise financial risks based on big data. By introducing a nonlinear risk adjustment mechanism based on the synchronous characteristics of operating cash flow, the intensity of risk transmission can adaptively adjust with industry cycle changes. When upstream and downstream enterprises are in the same downward phase of the industrial chain or are subjected to common macroeconomic shocks, the system can keenly identify the risk resonance effect caused by the synchronous deterioration of cash flow and dynamically amplify the relevant risk transmission relationships. This significantly improves the sensitivity of identifying systemic risks, chain defaults, and avalanche-like risk diffusion, avoiding the technical defects of traditional static models that generally underestimate risks during overall industry downturns, and making the risk warning results closer to the actual market operation. Furthermore, this invention incorporates debt maturity structure and liquidity defense capabilities into a unified spatiotemporal coupled assessment framework, dynamically characterizing risk absorption capacity. While maintaining the continuity of risk transmission, it effectively suppresses false alarms caused by maturity mismatch and unidentified funding buffers. For enterprises with sufficient time windows or strong quick asset reserves, risk shocks can be reasonably attenuated or even absorbed in the model, thus avoiding misjudging enterprises with self-rescue capabilities as high-risk nodes. Through the above-mentioned technical means, this invention not only enhances the ability to capture systemic risks, but also achieves more refined and interpretable risk warning results, providing more reliable technical support for financial institutions to formulate differentiated risk control strategies. Attached Figure Description

[0053] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0054] Figure 1 This is a flowchart illustrating a method for dynamic identification and early warning of enterprise financial risks based on big data, as described in an embodiment of the present invention. Detailed Implementation

[0055] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0056] See Figure 1 This is a flowchart of a method for dynamic identification and early warning of enterprise financial risks based on big data, provided in Embodiment 1 of the present invention. Figure 1 As shown, a method for dynamic identification and early warning of corporate financial risks based on big data may include:

[0057] Step S1 involves collecting and preprocessing basic data related to corporate finance and transactions to obtain a set of basic feature parameters for risk transmission modeling.

[0058] First, by connecting and synchronously collecting data from enterprise information systems, financial interfaces, and external credit data sources, basic data on enterprise finance and transactions are obtained. Specifically, for each target enterprise within the scope of risk identification, historical transaction records between enterprises, mutual guarantee contract amounts, accounts receivable balances, accounts payable balances, total net assets, and total liabilities are collected, and the above data are used as the basic data for constructing enterprise relationships.

[0059] By collecting time-series data on the inflow and outflow of operating funds of enterprises, the target enterprise's net operating cash flow data is obtained over multiple consecutive statistical periods. The time-series data on net operating cash flow covers at least a preset historical time window to form raw cash flow data that characterizes the fluctuations in the enterprise's operations.

[0060] By collecting information on the asset and liability structure of enterprises, the total quick assets and total current liabilities of the target enterprise are obtained. The total quick assets include at least cash, trading financial assets and accounts receivable that can be quickly converted into cash. The total current liabilities include short-term loans, accounts payable and non-current liabilities due within one year. The corresponding data are used as the basic financial data to characterize the liquidity status of the enterprise.

[0061] By collecting details of each outstanding debt of a company, data on the principal amount and remaining maturity date of each debt of the target company are obtained, forming data on the company's debt maturity structure. At the same time, by monitoring and collecting information from publicly disclosed information, judicial announcements and financial risk detection systems, information on default events related to the company's credit status and the time of default are obtained.

[0062] After data collection is completed, outlier removal, missing value completion, and consistency verification are performed on the collected basic data related to corporate finance and transactions. Based on the processed data, quick ratio, weighted average remaining debt maturity, first-order difference data of net operating cash flow, and initial static risk transmission weights are calculated as the basic feature parameter set for risk transmission modeling.

[0063] This completes the process of collecting and preprocessing basic data related to corporate finance and transactions to obtain a set of basic feature parameters for risk transmission modeling.

[0064] Step S2: By nonlinearly evaluating the synchronicity of fluctuations in the company's operating cash flow, an industry cycle resonance factor is obtained to correct the intensity of risk transmission.

[0065] In the process of iteratively calculating enterprise risk using a directed weighted graph model, the first issue to address is the underestimation of risk transmission intensity caused by existing algorithms neglecting the resonance effect of industry cycles. In real supply chain finance scenarios, the interconnected risks between enterprises are not isolated but are deeply modulated by macroeconomic industry cycles and market environments. When two enterprises in the upstream and downstream of the supply chain (i.e., the risk transmitter and the risk bearer) belong to the same industrial chain link or are affected by the same macroeconomic factors (such as a sharp drop in raw material prices or policy-driven production restrictions), their operating cash flow fluctuations often exhibit a high degree of synchronicity. This synchronicity means that when one party experiences a financial crisis, the other party is likely to be in a vulnerable state of cash flow depletion at the same time. At this point, when the risk flows from the transmitter to the bearer, it is no longer a simple linear superposition but a resonance-like phenomenon, resulting in a non-linear multiplication of risk destructive power. However, existing weighted calculation methods based on static amount proportions merely regard the amount of funds between the two as the width of the transmission channel, completely ignoring this resonance amplification effect caused by the synchronicity of time series fluctuations. This leads to the risk value calculated by the algorithm being far lower than the actual systemic collapse risk when the industry as a whole is declining. Therefore, this step analyzes the degree of synchronous coupling of cash flow changes between upstream and downstream enterprises by collecting operating cash flow data within the enterprise's historical time window. In particular, it identifies adverse operating conditions where both parties are simultaneously in a downward trend, and constructs an industry cycle resonance factor to dynamically amplify the edge weights that are in a resonance state, thereby truly restoring the intensity of systemic risk contagion.

[0066] In summary, this invention first obtains data representing the synchronicity of cash flow fluctuations by processing the directional consistency and amplitude correlation of the time-series difference data of operating cash flow of related enterprises. Specifically, a historical time window is set; in this embodiment, the length of the historical time window is set to 18 months. This length can be adjusted according to the actual scenario and is not required. For any target enterprise and its related enterprises, the time-series data of net operating cash flow of the target enterprise and its related enterprises within the historical time window are extracted from the basic feature parameter set. The time-series difference data of net operating cash flow is processed by first-order difference calculation of adjacent statistical periods to obtain the time-series difference data of operating cash flow of the target enterprise and its related enterprises. By determining the sign consistency of the time-series difference data of operating cash flow of the target enterprise and its related enterprises within the same statistical period, the synchronicity of cash flow fluctuations between the target enterprise and its related enterprises is identified. The statistical period is used to determine whether the direction of cash flow changes is consistent. Statistical periods with consistent cash flow changes are marked as synchronous change periods, while those with inconsistent cash flow changes are marked as asynchronous change periods. Within synchronous change periods, the absolute value of the product of the time-series difference data of the operating cash flow of the target company and its related companies is used as the correlation assessment value of cash flow fluctuation amplitude. This correlation assessment value is then used as the amplitude correlation assessment data to characterize the degree of correlation between the cash flow fluctuation amplitudes of the target company and its related companies within that statistical period. Based on the distribution of synchronous change periods and the corresponding amplitude correlation assessment data obtained by the target company and its related companies within historical time windows, the results of the direction consistency determination and amplitude correlation assessment within multiple statistical periods are accumulated and summarized to obtain the cash flow fluctuation synchronization characterization data used to characterize the synergy of cash flow fluctuations between the target company and its related companies.

[0067] After obtaining the data representing the synchronicity of cash flow fluctuations, the data is further processed by downside-sensitive weighting and nonlinear mapping to obtain an industry cycle resonance factor for correcting the intensity of risk transmission. Specifically, within a historical time window, for the target company and its related companies, the operating cash flow time-series difference data for each statistical period is used to determine the downward trend. This identifies whether the target company and its related companies are simultaneously in a declining cash flow state within that statistical period. The statistical period corresponding to the cash flow time-series difference data of both the target company and its related companies being negative within the same statistical period is marked as a downside resonance period, while the remaining statistical periods are marked as non-downside resonance periods. For any statistical period, the product of the operating cash flow time-series difference data of the target company and its related companies is used as the volatility correlation product for evaluation, and the volatility correlation is then assessed. The product assessment performs sign determination to identify whether the cash flow fluctuation direction of the target company and related companies is consistent within the statistical period. When the fluctuation correlation product assessment is positive, the square root of the absolute value of the fluctuation correlation product assessment is used as the basic correlation strength assessment corresponding to the statistical period. When the fluctuation correlation product assessment is negative, the square root of the absolute value of the fluctuation correlation product assessment, combined with sign reversal processing, is used as the basic correlation strength assessment corresponding to the statistical period. For the basic correlation strength assessment corresponding to the downward resonance period, a downward sensitive weight coefficient is introduced for weighting. In this embodiment of the invention, the downward sensitive weight coefficient for the downward resonance period is set to 4, and the downward sensitive weight coefficient for other periods is set to 1. For the basic correlation strength assessment corresponding to non-downward resonance periods, a benchmark weight coefficient is introduced for weighting to obtain the weighted correlation strength assessment data corresponding to each statistical period.

[0068] Based on the weighted correlation strength assessment data corresponding to each statistical period within the historical time window, the weighted correlation strength assessment data within multiple statistical periods are accumulated and summarized. Combined with the overall fluctuation range of the time series difference data of the operating cash flow of the target company and related companies, the data is normalized to obtain the cyclical resonance strength assessment value of the target company and related companies within the historical time window. By performing exponential nonlinear mapping on the cyclical resonance strength assessment value, an industry cyclical resonance factor is obtained to correct the risk transmission strength between the target company and related companies.

[0069] In one implementation, assume the first The company in the The time-series difference data of operating cash flow at each time step are as follows: ;No. The company in the The time-series difference data of operating cash flow at each time step are as follows: The length of the historical time window is The sign function is: 1 for input values ​​greater than 0, -1 for input values ​​less than 0, and 0 for input values ​​equal to 0. Downlink sensitivity weighting coefficient is Then the first The company and the first The formula for calculating the industry cycle resonance factor of an individual enterprise is as follows:

[0070]

[0071] in, Indicates the first The company and the first Industry cycle resonance factor for individual enterprises; This represents an exponential function with the natural constant e as the base. This represents the downlink sensitivity weighting coefficient; This is a sign function that takes 1 when the input value is greater than 0, -1 when it is less than 0, and 0 when it is equal to 0. Indicates the first The company in the Time-series difference data of operating cash flow at each time step; Indicates the first The company in the Time-series difference data of operating cash flow at each time step; Indicates the length of the historical time window; This represents the absolute value function.

[0072] It should be noted that, firstly, regarding the question of how to capture the synchronicity of fluctuations, the numerator of the formula... The item served as a direction identifier. When enterprises and enterprises When the cash flows change in the same direction (i.e., rise or fall together), this item is positive, contributing positively to the index. If the fluctuations of the two are complementary (one rises and the other falls), the term will be negative. The reduction reflects the risk hedging effect. This design enables the algorithm to distinguish between two different risk states: resonance and hedging. Secondly, to address the asymmetric risk characteristic where bad news is more lethal than good news, the formula introduces a downside-sensitive weight coefficient. In actual risk control, the probability of risk transmission when both parties experience a simultaneous sharp drop in cash flow (a double whammy) is far higher than when both parties experience a simultaneous increase. Therefore, when a double whammy is detected... and At that time, Set to a value greater than The penalty value is set to [value], while in other cases it is set to [value]. This conditional triggering mechanism ensures that the factor can keenly capture and amplify the impact of periods of severe resonance, preventing the algorithm from reacting sluggishly during widespread industry declines. Furthermore, to accurately quantify the energy magnitude of the resonance, the molecule employs... The square root operation is introduced as a non-linear compression of numerical amplitude. This is because extreme outliers in financial data often introduce significant statistical noise. Square root processing can preserve fluctuation amplitude information while suppressing excessive interference from individual outliers on the overall resonance trend, ensuring the robustness of factor calculation. Finally, the denominator ensures that the resonance intensity calculated in the numerator is a relative quantity relative to the company's own fluctuation scale, eliminating the difference in data dimensions between companies of different sizes. The mapping of the exponential function further distributes the data from the original distribution... The linear correlation between them is transformed into The multiplicative factor. This means that when the resonance is extremely strong, It will grow exponentially, thereby driving the algorithm to significantly increase the risk transmission weight of this path in subsequent iterations, accurately simulating the avalanche effect when systemic risks erupt.

[0073] Thus, the nonlinear assessment of the synchronicity of fluctuations in corporate operating cash flow has been completed, and an industry cycle resonance factor has been obtained to correct the intensity of risk transmission.

[0074] Step S3 involves conducting a spatiotemporal coupling assessment of the company's debt maturity structure and liquidity defense capabilities to obtain a spatiotemporal phase damping factor for correcting risk-bearing capacity.

[0075] After addressing the underestimation of risk transmission intensity by introducing an industry cycle resonance factor in step S2, the algorithm can identify avalanche-like resonance risks. However, a further detailed problem remains: false alarms in risk warnings caused by the neglect of maturity mismatch and liquidity defense. Specifically, the existing calculation logic assumes that risk transmission is instantaneous and rigid. However, in real business scenarios, even if upstream companies experience severe financial crises (or even resonate with downstream companies), downstream companies (the risk bearers) may not collapse immediately. This depends on two key spatiotemporal factors: First, the time phase difference, i.e., the time distance between the downstream company's own debt maturity date and the upstream company's default point. If the downstream company's debt repayment pressure is mainly concentrated in the long term, while the upstream shock occurs in the present, then the downstream company has sufficient time to restructure its debt or find alternative suppliers. This time misalignment naturally constitutes a risk firewall. Second, liquidity defense, i.e., the downstream company's current quick asset reserves. Existing algorithms lack coupling analysis of these two factors. For example, when a company has extremely high cash reserves and its debt is mainly due in three years, even if its upstream partners default, the actual impact on the company should be significantly reduced (damping effect). If the algorithm ignores this and still calculates based on the weight amplified by the industry cycle resonance factor, it will misjudge companies with strong self-rescue capabilities as high-risk nodes, creating a false prosperity in the early warning signal. Therefore, this step needs to further collect data on the company's debt maturity structure and quick asset data based on step S2 to construct a spatiotemporal phase damping factor. The mechanism of this factor is: when it detects that the risk-bearing party has sufficient time buffer and funds, it forcibly reduces the risk transmission weight, thereby filtering out pseudo-risk nodes that, although on the resonance chain, are actually immune to shocks.

[0076] In summary, this invention first obtains time buffer representation data for risk shocks by performing time phase difference analysis on corporate debt maturity structure data. Specifically, for any target company, the principal amount and remaining maturity time of each outstanding debt are extracted from the basic feature parameter set. A weighted average is then calculated using the principal amount of each outstanding debt as a weight to obtain the weighted average remaining maturity time of the target company's debt. For related companies of the target company, risk event time data is extracted from the basic feature parameter set. This risk event time data includes the predicted default time of the related company or the default time corresponding to a default event that has already occurred. By calculating the time difference between the weighted average remaining maturity time of the target company's debt and the risk event time data of the related companies, time phase difference assessment data representing the relative chronological relationship between the target company's debt maturity time and the risk shock time of the related companies is obtained. This time phase difference assessment data is used as time buffer representation data to characterize the target company's ability to buffer against risk shocks from related companies over time.

[0077] After obtaining the time buffer representation data of risk shocks, the solvency assessment of the enterprise's quick assets and current liabilities is further performed to obtain the liquidity defense capability representation data of the enterprise. Specifically, for any target enterprise, the total quick assets data and total current liabilities data of the target enterprise are extracted from the basic feature parameter set. The total quick assets data includes at least cash, trading financial assets, and accounts receivable that can be quickly converted into cash. The total current liabilities data includes short-term loans, accounts payable, and non-current liabilities due within one year. By calculating the ratio of the total quick assets data and the total current liabilities data of the target enterprise, the quick ratio assessment data of the target enterprise is obtained by using the total quick assets data as the numerator and the total current liabilities data as the denominator. When the total current liabilities data of the target enterprise is zero, the quick ratio assessment data of the target enterprise is set to a preset high constant. In this embodiment of the invention, the high constant is set to 20 to represent that the target enterprise has strong short-term solvency. The quick ratio assessment data of the target enterprise is used as the liquidity defense capability representation data to represent the target enterprise's ability to defend against external risk shocks in terms of funds.

[0078] After obtaining the enterprise liquidity defense capability characterization data, the spatiotemporal coupling damping processing is performed on the time buffer characterization data and the liquidity defense capability characterization data to obtain the spatiotemporal phase damping factor used to correct the risk-bearing capability. Specifically, a time sensitivity normalization constant is set, which is set to 30 (days) in this embodiment. For any target enterprise and its related enterprises, time buffer characterization data and liquidity defense capability characterization data are obtained. By performing time sensitivity normalization processing on the time buffer characterization data, the ratio of the time buffer characterization data to the time sensitivity normalization constant is calculated to obtain normalized time difference assessment data that characterizes the degree of influence of time phase difference. Logarithmic mapping processing is performed on the sum of the absolute value of the time buffer characterization data and the natural constant. The method involves obtaining logarithmic time buffer assessment data to characterize the scale effect of the time buffer, and then performing multiplicative modulation processing on the logarithmic time buffer assessment data, liquidity defense capability characterization data, and a preset liquidity efficiency amplification coefficient to obtain time-coupling assessment data of funds characterizing the defense efficiency of liquidity under time buffer conditions. In this embodiment of the invention, the liquidity efficiency amplification coefficient is set to 0.8 to avoid the fund item having too high a proportion in the index, which would lead to numerical instability. The method also involves summing the normalized time difference assessment data and the time-coupling fund assessment data to obtain spatiotemporal coupling driving quantity assessment data. By performing exponential nonlinear mapping processing on the spatiotemporal coupling driving quantity assessment data and adding the mapping result to a constant 1 and taking the reciprocal, the method obtains the spatiotemporal phase damping factor used to correct the target enterprise's ability to withstand risk shocks from related enterprises.

[0079] In one implementation, assume the first The weighted average remaining debt maturity of the enterprises is ;No. Risk event timing data for each related enterprise is as follows: The time sensitivity normalization constant is: The liquidity efficiency amplification factor is: ;No. The liquidity defense capability characterization data for each enterprise is as follows: Then the first The company and the first The formula for calculating the spatiotemporal phase damping factor of an enterprise is as follows:

[0080]

[0081] in, Indicates the first The company and the first Spatiotemporal phase damping factor of an individual enterprise; Indicates the first Weighted average remaining debt maturity for each enterprise; Indicates the first Risk event timing data for each related enterprise; Indicates the liquidity efficiency amplification factor; This represents the time sensitivity normalization constant; Indicates the first Data characterizing the liquidity defense capabilities of individual enterprises; Let e ​​represent the natural constant.

[0082] It should be noted that this formula constructs a nonlinear damped filter with an inverse gating mechanism, and its value range is strictly limited to [value range missing]. Between these values, the smaller the value, the stronger the damping effect (i.e., the more risk is blocked). Firstly, the exponential term in the formula... This reflects the time lag in the crisis. When Time (i.e., enterprise) The repayment date is in the distant future, and the company (I'll default now), this term is a large positive number. Because this term is in the exponent of the denominator, a large positive number will cause the denominator to become extremely large, thus making the entire... Tend to This mathematically encapsulates the fact that the more time available, the less impact one experiences, thus achieving time-based risk mitigation. Secondly, to couple the role of financial defense, the formula introduces [a specific factor] into the index. This item will improve liquidity. The product was performed with the logarithm of the time difference. This means that the funds... It doesn't operate in isolation, but rather as a time multiplier. If a business has a long time window ( (very large), then This allows for more efficient utilization (such as turnover and investment), thereby generating greater defensive potential. Conversely, if the time window is zero, the logarithmic term approaches... The defensive effectiveness of funds returns to the baseline level. Through this design, the formula achieves the following adjustment effect: when enterprises... Facing a tight timeframe ( )and When it is very low, the exponential term approaches The denominator approaches , Maintaining a high conduction level indicates damping failure and risk penetration. When the enterprise When there is ample time or a large amount of money, the exponential term in the denominator grows explosively, leading to... rapidly trending towards This indicates a soft landing, meaning that although the risks were transmitted, they were completely absorbed by the company's own immune system.

[0083] Thus, the spatiotemporal phase damping factor for correcting risk-bearing capacity was obtained by conducting a spatiotemporal coupling assessment of the corporate debt maturity structure and liquidity defense capabilities.

[0084] Step S4 involves fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor to obtain the dynamic risk transmission matrix and then performing iterative solutions.

[0085] To avoid the complete disruption of risk transmission paths due to excessive fluctuations in a single dynamic factor, thereby ignoring the objectively existing basic friction costs in supply chain relationships, this invention employs a hybrid weighted architecture combining static retention and dynamic adjustment to reconstruct the risk transmission coefficient. Specifically, for any target enterprise and its related enterprises, initial static risk transmission weights between the target enterprise and its related enterprises are extracted from the set of basic feature parameters, and industry cycle resonance factors and spatiotemporal phase damping factors are obtained. A static risk retention coefficient is set; in this embodiment, the static risk retention coefficient is set to 0.15, serving as a rigid risk retention term, and the difference between the constant 1 and the static risk retention coefficient is used as a dynamic adjustment weight. By multiplicatively modulating the dynamic adjustment weight with the industry cycle resonance factor and the spatiotemporal phase damping factor, and then summing it with the static risk retention term, a fusion adjustment coefficient for adjusting the initial static risk transmission weight is obtained. By multiplicatively modulating the initial static risk transmission weight with the fusion adjustment coefficient, the dynamic risk transmission weights between the target enterprise and its related enterprises are obtained. Based on the dynamic risk transmission weights obtained between each target enterprise and its related enterprises, a dynamic risk transmission matrix among all enterprise nodes is constructed.

[0086] Initialize the risk status value of each enterprise node, and perform risk iterative propagation operation based on the dynamic risk transmission matrix. In each iteration, the enterprise risk status value of the current round is superimposed and updated with the risk transmission result corresponding to the dynamic risk transmission matrix, and the updated risk status value is subject to upper limit constraint processing. Repeat the risk iterative propagation operation until the change of enterprise risk status value obtained by two adjacent iterations is less than the preset convergence threshold or the preset number of iterations is reached. The final converged set of enterprise risk status values ​​is taken as the enterprise risk score result.

[0087] Thus, the dynamic risk transmission matrix was obtained and iteratively solved by fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor.

[0088] Step S5: By performing decoupling analysis on the converged enterprise risk scores, dynamic classification risk warning signals are obtained.

[0089] Based on the enterprise risk score results output in step S4, enterprises within the system are classified and given early warnings. This invention generates interpretable early warning reports by decoupling and analyzing the dominant factors leading to increased risk.

[0090] The specific logic is as follows: Set a high-risk warning threshold. In this embodiment of the invention, the high-risk warning threshold is set to 0.7. The system iterates through all enterprise nodes, and when a certain enterprise... When the final risk score exceeds the high-risk warning threshold, the system triggers an alarm. At this point, the system backtracks the high-weight in-degree edge data of that node and calculates the dynamic adjustment term. The contribution to the total weight is assessed and categorized: If backtracking analysis reveals that the main risk source edges at this node generally exhibit [a certain characteristic], then [the following is a possible interpretation]. In this situation, the system generates an industry cycle resonance-type risk warning. This alerts risk control personnel that the company is in a resonance zone where cash flow is deteriorating simultaneously across the entire industry chain, with the risk source being a macro-systemic shock. It recommends implementing industry-level hedging strategies. If backtracking analysis reveals that among the main risk sources at this point, there are... Furthermore, if the initial static risk transmission weight is relatively large (meaning that despite the time lag or funding, the damping has failed to take effect, or there is no defensive capability at all), the system generates a liquidity defense depletion risk warning. This alerts risk control personnel that although the company has not yet defaulted, it has lost its immunity to upstream risks due to the approaching debt maturity date and insufficient quick assets, and recommends immediate liquidity injection or debt rollover.

[0091] This completes the process of obtaining dynamic classification risk warning signals by decoupling the converged enterprise risk scores.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic identification and early warning of enterprise financial risks based on big data, characterized in that, The method includes: Step S1: By collecting and preprocessing the basic data related to corporate finance and transactions, a set of basic feature parameters for risk transmission modeling is obtained; Step S2: Obtain the industry cycle resonance factor for correcting the intensity of risk transmission by nonlinearly assessing the synchronicity of fluctuations in corporate operating cash flow; Step S3: By conducting a spatiotemporal coupling assessment of the corporate debt maturity structure and liquidity defense capabilities, obtain the spatiotemporal phase damping factor used to correct the risk-bearing capacity; Step S4: By fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor, the dynamic risk transmission matrix is ​​obtained and iterative solution is performed; Step S5: Obtain dynamic classification risk warning signals by performing decoupling analysis on the converged enterprise risk scores.

2. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 1, characterized in that, The process involves collecting and preprocessing basic financial and transaction-related data of enterprises to obtain a set of basic feature parameters for risk transmission modeling, including: By connecting and synchronously collecting data from enterprise information systems, financial interfaces, and external credit data sources, basic data on enterprise finance and transactions are obtained. Specifically, for each target enterprise within the scope of risk identification, historical transaction records between enterprises, mutual guarantee contract amounts, accounts receivable balances, accounts payable balances, total net assets, and total liabilities are collected, and the above data are used as the basic data for constructing enterprise relationships. By collecting the inflow and outflow records of enterprise operating funds in a time series manner, the target enterprise's net operating cash flow time series data in multiple consecutive statistical periods is obtained. The net operating cash flow time series data covers at least a preset length of historical time window to form raw cash flow data for characterizing the fluctuation characteristics of enterprise operations. By collecting information on the asset and liability structure of enterprises, the total quick assets and total current liabilities of the target enterprise are obtained. The total quick assets include at least cash, trading financial assets and accounts receivable that can be quickly converted into cash. The total current liabilities include short-term loans, accounts payable and non-current liabilities due within one year. The corresponding data are used as the basic financial data to characterize the liquidity status of the enterprise. By collecting details of each outstanding debt of a company, data on the principal amount and remaining maturity time of each debt of the target company can be obtained, forming data on the company's debt maturity structure. At the same time, by monitoring and collecting information from publicly disclosed information, judicial announcements and financial risk detection systems, information on default events and the time of default related to the company's credit status can be obtained. After data collection is completed, outlier removal, missing value completion, and consistency verification are performed on the collected basic data related to corporate finance and transactions. Based on the processed data, quick ratio, weighted average remaining debt maturity, first-order difference data of net operating cash flow, and initial static risk transmission weights are calculated as the basic feature parameter set for risk transmission modeling.

3. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 1, characterized in that, The process involves obtaining an industry cycle resonance factor for correcting the intensity of risk transmission through a nonlinear assessment of the synchronicity of fluctuations in corporate operating cash flow, including: By performing directional consistency and amplitude correlation processing on the time-series differential data of operating cash flows of related enterprises, data representing the synchronicity of cash flow fluctuations are obtained. By performing downside-sensitive weighting and nonlinear mapping on the cash flow fluctuation synchronicity characterization data, an industry cycle resonance factor is obtained to correct the intensity of risk transmission.

4. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 3, characterized in that, The process of obtaining data representing the synchronicity of cash flow fluctuations by performing directional consistency and amplitude correlation processing on the time-series differential data of operating cash flows of related enterprises includes: Set a historical time window; for any target company and its related companies, extract the time series data of net operating cash flow of the target company and its related companies in the historical time window from the set of basic feature parameters, perform first-order difference calculation on the net operating cash flow time data of adjacent statistical periods, and obtain the time series difference data of operating cash flow of the target company and its related companies. By determining the sign consistency of the time-series difference data of the operating cash flow of the target company and its related companies in the same statistical period, we can identify whether the cash flow of the target company and its related companies changes in the same direction in the statistical period. Statistical periods with consistent cash flow changes are marked as synchronous change periods, and statistical periods with inconsistent cash flow changes are marked as asynchronous change periods. Within the synchronous change period, the absolute value of the product of the time-series difference data of the operating cash flow of the target company and the related companies is used as the correlation assessment value of the cash flow fluctuation amplitude, and the correlation assessment value of the cash flow fluctuation amplitude is used as the amplitude correlation assessment data to characterize the degree of correlation between the cash flow fluctuation amplitude of the target company and the related companies within the statistical period. Based on the distribution of synchronous change cycles of the target company and related companies within historical time windows and the corresponding amplitude correlation assessment data, the directional consistency judgment results and amplitude correlation assessment results within multiple statistical periods are accumulated and summarized to obtain cash flow fluctuation synchronization characterization data used to characterize the cash flow fluctuation synergy of the target company and related companies.

5. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 3, characterized in that, The process of obtaining an industry cycle resonance factor for correcting the intensity of risk transmission by performing downside-sensitive weighting and nonlinear mapping on the cash flow fluctuation synchronicity characterization data includes: Within the historical time window, for the target company and its related companies, the operating cash flow time series difference data corresponding to each statistical period is used to determine the downward state, identify whether the target company and its related companies are simultaneously in a state of declining cash flow in the same statistical period, and mark the statistical period corresponding to the cash flow time series difference data of the target company and its related companies that are both negative in the same statistical period as the downward resonance period, and mark the other statistical periods as non-downward resonance periods. For any given statistical period, the product of the time-series difference data of the operating cash flows of the target company and its related companies is used as the fluctuation correlation product assessment. The sign of the fluctuation correlation product assessment is determined to identify whether the cash flow fluctuations of the target company and its related companies are consistent within the statistical period. When the fluctuation correlation product assessment is positive, the square root of the absolute value of the fluctuation correlation product assessment is used as the basic correlation strength assessment for that statistical period. When the fluctuation correlation product assessment is negative, the square root of the absolute value of the fluctuation correlation product assessment, combined with sign reversal processing, is used as the basic correlation strength assessment for that statistical period. For the assessment of basic correlation strength corresponding to the downtrend resonance cycle, a downtrend sensitive weight coefficient is introduced for weighting; for the assessment of basic correlation strength corresponding to non-downtrend resonance cycle, a benchmark weight coefficient is introduced for weighting to obtain weighted correlation strength assessment data for each statistical cycle. Based on the weighted correlation strength assessment data corresponding to each statistical period within the historical time window, the weighted correlation strength assessment data within multiple statistical periods are accumulated and summarized. Combined with the overall fluctuation range of the time series difference data of the operating cash flow of the target company and related companies, the data is normalized to obtain the cyclical resonance strength assessment value of the target company and related companies within the historical time window. By performing exponential nonlinear mapping on the cyclical resonance strength assessment value, an industry cyclical resonance factor is obtained to correct the risk transmission strength between the target company and related companies.

6. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 1, characterized in that, The method involves a spatiotemporal coupling assessment of the company's debt maturity structure and liquidity defense capabilities to obtain a spatiotemporal phase damping factor for correcting risk-bearing capacity, including: By performing time phase difference analysis on corporate debt maturity structure data, we can obtain time buffer representation data of risk shocks. By assessing a company's solvency through data on its quick assets and current liabilities, data representing the company's liquidity defense capabilities can be obtained. By performing spatiotemporal coupling damping processing on time buffer characterization data and liquidity defense capability characterization data, a spatiotemporal phase damping factor for correcting risk absorption capability is obtained.

7. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 6, characterized in that, The method of obtaining time buffer representation data of risk shocks by performing time phase difference analysis on corporate debt maturity structure data includes: For any target company, extract the principal amount and remaining maturity time of each outstanding debt from the set of basic feature parameters. Then, use the principal amount of each outstanding debt as the weight to calculate the weighted average remaining maturity time of the debt, and obtain the weighted average remaining maturity time of the debt for the target company. For the target company's related companies, risk event timing data corresponding to the related companies are extracted from the basic feature parameter set. The risk event timing data includes the predicted default outbreak time of the related companies or the default time corresponding to the default events that have already occurred. By calculating the time difference between the target company's weighted average remaining debt maturity time and the risk event timing data of the related companies, time phase difference assessment data representing the relative chronological relationship between the target company's debt maturity time and the risk impact time of the related companies is obtained. The time phase difference assessment data is used as time buffer representation data to characterize the target company's buffering capacity against risk impacts from related companies in the time dimension.

8. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 6, characterized in that, The process involves assessing a company's solvency by analyzing its quick assets and current liabilities to obtain data characterizing its liquidity defense capabilities, including: For any target company, extract the total quick assets and total current liabilities data corresponding to the target company from the set of basic feature parameters. The total quick assets data includes at least cash, trading financial assets and accounts receivable that can be quickly converted into cash, and the total current liabilities data includes short-term loans, accounts payable and non-current liabilities due within one year. By calculating the ratio between the total quick assets and the total current liabilities of the target company, using the total quick assets as the numerator and the total current liabilities as the denominator, the corresponding quick ratio assessment data of the target company is obtained. When the target company's total current liabilities are zero, the target company's quick ratio assessment data is set to a preset high constant to indicate that the target company has strong short-term solvency. The quick ratio assessment data corresponding to the target company is used as a liquidity defense capability representation data to characterize the target company's ability to defend against external risk shocks in terms of funding.

9. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 6, characterized in that, The process of performing spatiotemporal coupling damping processing on time buffer characterization data and liquidity defense capability characterization data to obtain a spatiotemporal phase damping factor for correcting risk absorption capability includes: Set a time sensitivity normalization constant, and for any target company and its related companies, obtain time buffer characterization data and liquidity defense capability characterization data; by performing time sensitivity normalization processing on the time buffer characterization data, calculate the ratio of the time buffer characterization data to the time sensitivity normalization constant, and obtain normalized time difference assessment data that characterizes the degree of influence of time phase difference. By performing a logarithmic mapping on the sum of the absolute values ​​of the time buffer characterization data and the natural constant, logarithmic evaluation data of the time buffer is obtained to characterize the scale effect of the time buffer. This logarithmic evaluation data is then multiplicatively modulated with liquidity defense capability characterization data and a preset liquidity effectiveness amplification coefficient to obtain time-coupling evaluation data of funds, characterizing the defense effectiveness of liquidity under time buffer conditions. Normalized time difference evaluation data is summed with time-coupling evaluation data to obtain spatiotemporal coupling driving force evaluation data. Finally, by performing an exponential nonlinear mapping on the spatiotemporal coupling driving force evaluation data and adding the mapping result to a constant 1 and taking the reciprocal, a spatiotemporal phase damping factor is obtained to correct the target company's ability to absorb risk shocks from related companies.

10. The method for dynamic identification and early warning of enterprise financial risks based on big data according to claim 1, characterized in that, The process of fusing and adjusting the initial static risk transmission weights with the industry cycle resonance factor and the spatiotemporal phase damping factor to obtain the dynamic risk transmission matrix and performing iterative solutions includes: For any target company and its related companies, the initial static risk transmission weights between the target company and its related companies are extracted from the set of basic feature parameters, and the industry cycle resonance factor and the spatiotemporal phase damping factor are obtained. A static risk retention coefficient is set and used as a rigid risk retention term. The difference between the constant 1 and the static risk retention coefficient is used as a dynamic adjustment weight. By multiplying the dynamic adjustment weight with the industry cycle resonance factor and the spatiotemporal phase damping factor, and then summing it with the static risk retention term, a fusion adjustment coefficient is obtained to adjust the initial static risk transmission weight. By multiplying the initial static risk transmission weight with the fusion adjustment coefficient, the dynamic risk transmission weight between the target enterprise and its related enterprises is obtained. Based on the dynamic risk transmission weights obtained between each target enterprise and its related enterprises, a dynamic risk transmission matrix is ​​constructed among all enterprise nodes. Initialize the risk status value of each enterprise node, and perform risk iterative propagation operation based on the dynamic risk transmission matrix. In each iteration, the enterprise risk status value of the current round is superimposed and updated with the risk transmission result corresponding to the dynamic risk transmission matrix, and the updated risk status value is subject to upper limit constraint processing. Repeat the risk iterative propagation operation until the change of enterprise risk status value obtained by two adjacent iterations is less than the preset convergence threshold or the preset number of iterations is reached. The final converged set of enterprise risk status values ​​is taken as the enterprise risk score result.

Images