Comprehensive risk assessment method and device based on interaction between enterprise and external environment

Abstract

The invention discloses a comprehensive risk assessment method and device based on interaction between an enterprise and an external environment. The method comprises the following steps: acquiring enterprise internal data and external environment system data; the collected data are preprocessed, and data extraction is carried out; respectively inputting the extracted data into an enterprise internal data risk model and an external environment data risk model so as to respectively calculate and obtain a preliminary internal risk score and a preliminary external risk score; inputting the initial internal risk score and the initial external risk score into a dynamic risk interaction model to calculate and obtain a dynamic interaction risk score, and optimizing the initial internal risk score and the initial external risk score; and calculating a comprehensive risk score according to the optimized internal risk score, the optimized external risk score and the dynamic interaction risk score. According to the method, the complexity of enterprise risk sources can be captured, potential risks caused by the external environment can be identified, and the method has wide applicability and high efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of financial risk control methods, and in particular to a comprehensive risk assessment method and device based on the interaction between an enterprise and an external environment. Background Art

[0002] The financial risk of an enterprise is not only affected by the internal financial data and operating conditions of the enterprise, but also closely related to the external environment, especially external factors such as macroeconomic fluctuations, industry cyclical changes, market competition and policy changes. This multi-level external factor has a significant impact on the operation and financial health of the enterprise. It is difficult to fully assess the risk based on the enterprise's own data alone. The pain points of enterprise risk assessment are as follows: 1. Limitations of a single data source: Traditional enterprise risk assessments mostly rely on the company's financial statements, credit records, and historical transaction data. These data can only reflect the company's current or historical financial health, and cannot timely reflect the potential impact of changes in the external environment on the company; 2. Complexity of external factors: The survival and development of enterprises depend on the entire market ecosystem. External factors such as policy changes, industry competition, and global supply chain disruptions often have a significant impact on corporate risks. Traditional risk models usually ignore these dynamic factors. 3. Insufficient ability to cope with extreme market fluctuations: During economic crises or drastic market fluctuations, the financial health of companies often deteriorates sharply. Real-time ecosystem interaction analysis can help financial institutions or investors identify potential risks of companies in advance and take corresponding countermeasures. Summary of the invention

[0003] The purpose of the present invention is to provide a comprehensive risk assessment method and device based on the interaction between an enterprise and its external environment in view of the deficiencies in the prior art.

[0004] To achieve the above objectives, in a first aspect, the present invention provides a comprehensive risk assessment method based on the interaction between an enterprise and an external environment, comprising: Step 1: Collect enterprise internal data and external environment system data; Step 2: pre-process the collected internal enterprise data and external environment system data, extract risk-related features from the internal enterprise data, and extract key macroeconomic and industry indicators from the external environment system data; Step 3: Input the extracted risk-related characteristics, macroeconomic and industry key indicators into the enterprise internal data risk model and external environment data risk model respectively, so as to calculate and obtain the preliminary internal risk score and the preliminary external risk score respectively; Step 4: Input the preliminary internal risk score and the external risk score into the dynamic risk interaction model to calculate the dynamic interaction risk score, and optimize the preliminary internal risk score and the preliminary external risk score; Step 5: Calculate the comprehensive risk score based on the optimized internal risk score, external risk score, and dynamic interaction risk score:

[0005] in, is the calculated comprehensive risk score, is the optimized internal risk score, is the optimized external risk score, is the dynamic interaction risk score, t is the time index, They are the weights of internal risk score, external risk score and dynamic interaction risk score respectively.

[0006] Furthermore, the internal enterprise data includes financial statements, transaction data and credit records; the external environment system data includes industry data, global economic data and policy change data.

[0007] Furthermore, the preprocessing method is specifically as follows: Align and interpolate data at different time scales to align the time series; Standardize or normalize the data.

[0008] Furthermore, the risk-related characteristics include debt-to-asset ratio, current ratio, sales growth rate and cash flow coverage ratio.

[0009] Furthermore, the calculation method of the preliminary internal risk score is as follows:

[0010] in, is the calculated preliminary internal risk score, is the debt-to-asset ratio, is the current ratio, is the sales growth rate, is the cash flow coverage ratio, , , , They are the weights of the debt-to-asset ratio, current ratio, sales growth rate and cash flow coverage ratio respectively.

[0011] Furthermore, the key macroeconomic and industry indicators include GDP growth rate, market volatility and event variables.

[0012] Furthermore, the calculation method of the preliminary external risk score is as follows:

[0013] in, is the calculated preliminary external risk score, is the GDP growth rate, is the volatility, is the event variable, , , They are the weights of GDP growth rate, market volatility and event variables respectively.

[0014] Furthermore, the dynamic interaction risk score is calculated as follows:

[0015] in, is the calculated dynamic interaction risk score, Indicates the use of long short-term memory network for processing, are the intercept terms, are the regression coefficients of internal risk scores, are the regression coefficients of external risk scores, are the error terms, is the optimized internal risk score, is the optimized external risk score, is the predicted value of the internal risk score calculated based on the vector autoregression model, is the predicted value of the external risk score calculated based on the vector autoregression model.

[0016] Furthermore, the optimized internal risk scores and external risk scores are as follows: .

[0017] In a second aspect, the present invention provides a comprehensive risk assessment device based on the interaction between an enterprise and an external environment, comprising a storage medium and a processor, wherein the storage medium stores a computer program, and the computer program is used to implement the above method when executed by the processor.

[0018] Beneficial effects: 1. The present invention innovatively integrates internal enterprise data (such as financial statements, credit records, transaction data) with external environment system data (such as macroeconomics, industry dynamics, policy changes, market fluctuations) in multiple dimensions, and dynamically integrates these data into a unified risk assessment input through feature extraction and standardization; this multi-source data fusion method significantly improves the comprehensiveness of enterprise risk assessment and can capture the complexity and diversity of enterprise risk sources; compared with traditional models that rely solely on internal enterprise data, the present invention can identify potential risks caused by the external environment and provide enterprises with more accurate global risk assessments; it is particularly suitable for risk prediction of multi-industry and multi-regional enterprises, and has wide applicability and high efficiency.

[0019] 2. The present invention proposes a dynamic risk interaction analysis based on a vector autoregression model (VAR) and a deep learning model (LSTM), which effectively captures the time series interaction relationship between the internal data of an enterprise and the external ecological data; by modeling the lag effect and nonlinear relationship between the enterprise and the external environmental data, it can predict the dynamic transmission effect of external environmental changes (such as industry fluctuations, policy adjustments) on the future risk level of the enterprise; this dynamic risk interaction analysis solves the limitation that traditional static models cannot cope with complex dynamic environments, provides a forward-looking perspective for risk assessment, and improves the accuracy and timeliness of risk prediction; 3. The present invention supports enterprise risk prediction based on scenario simulation. By constructing different scenarios of external environmental changes (such as market fluctuations, policy adjustments, and economic crises), it predicts the risk evolution trend of enterprises in specific environments. It can dynamically adjust external ecological variables and generate risk scores and prediction results for enterprises in different scenarios. Scenario-based analysis helps financial institutions identify potential risks in complex environments in advance and provide data support for credit decisions, risk management, etc. This capability makes the present invention particularly outstanding in dealing with risk assessments in extreme environments, providing more accurate risk response strategies for financial institutions and corporate decision makers. 4. The present invention has designed a real-time monitoring and rapid early warning mechanism, which can timely update the risk assessment results of enterprises by continuously tracking the dynamic changes of external environmental data (such as fluctuations in economic indicators, updates of industry data, policy releases, etc.); when the external environment undergoes major changes and may have a significant impact on the enterprise's risks, it can quickly adjust the comprehensive risk score and issue dynamic early warning suggestions; compared with the lag of traditional risk models, the real-time monitoring and rapid response mechanism of the present invention significantly improves the ability to respond to emergencies, helping financial institutions to identify potential crises in advance and optimize risk control and business planning; 5. The present invention has a high degree of industry adaptability and can flexibly adjust external data variables and model parameters according to the industry characteristics (such as manufacturing, service, and financial industries) and regional characteristics (such as domestic and foreign markets, regional policies) of the enterprise. Through customized risk assessment solutions, it can provide personalized risk assessment and prediction for enterprises in different industries. This scalability makes the present invention not only suitable for corporate credit management of financial institutions, but also can serve supply chain management, cross-border trade enterprises and other fields, which fully reflects the adaptability and wide application value of the technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flowchart of a comprehensive risk assessment method based on the interaction between the enterprise and the external environment. DETAILED DESCRIPTION

[0021] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. The present embodiments are implemented based on the technical solution of the present invention. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0022] like Figure 1 As shown, an embodiment of the present invention provides a comprehensive risk assessment method based on the interaction between an enterprise and an external environment, including: Step 1: Collect internal enterprise data and external environment system data. Internal enterprise data includes: Financial statements: including balance sheet, income statement, cash flow statement, etc., which reflect the overall financial health of the company.

[0023] Transaction data: the company’s historical orders, purchase records, sales, accounts receivable and accounts payable, etc.

[0024] Credit record: the company’s credit score, loan repayment record, etc.

[0025] External environment system data includes: Industry data: industry prosperity index, market demand, supply and demand relationship, etc. For example, the manufacturing PMI (Purchasing Managers Index) is an important indicator reflecting the health of the industry.

[0026] Global economic data: including GDP growth rate, inflation rate, unemployment rate, exchange rate fluctuations, etc., reflecting the potential impact of the macroeconomic environment on corporate operations.

[0027] Policy changes: National and regional policy adjustments, such as tax policies, trade policies, interest rate adjustments, environmental protection policies, etc., have direct impacts on enterprises.

[0028] Step 2: Preprocess the collected internal enterprise data and external environment system data, extract risk-related features from the internal enterprise data, and extract key macroeconomic and industry indicators from the external environment system data.

[0029] The above preprocessing method is as follows: Time series alignment: Internal enterprise data is usually recorded on a monthly or quarterly basis, while external economic data may fluctuate on an annual, quarterly or daily basis. Data of these different time scales need to be aligned and interpolated. For example, aligning the monthly enterprise data with the quarterly GDP growth rate for analysis.

[0030] Standardization and normalization: The units and scales of external data and enterprise data vary greatly. By standardizing or normalizing the data, you can ensure that different data sources are comparable.

[0031] Step 3: Input the extracted risk-related characteristics, macroeconomic and industry key indicators into the enterprise internal data risk model and external environment data risk model respectively to calculate the preliminary internal risk score and preliminary external risk score respectively.

[0032] Specifically, the above risk-related characteristics include asset-liability ratio, current ratio, sales growth rate and cash flow coverage ratio, and their calculation methods are all existing technologies and will not be repeated here. Correspondingly, the calculation method of the preliminary internal risk score is as follows:

[0033] in, is the calculated preliminary internal risk score, is the debt-to-asset ratio, is the current ratio, is the sales growth rate, is the cash flow coverage ratio, , , , They are the weights of the debt-to-asset ratio, current ratio, sales growth rate and cash flow coverage ratio respectively.

[0034] The above key indicators of macroeconomics and industries include GDP growth rate, market volatility and event variables. Among them, market volatility can be the volatility of the stock market or the fluctuation range of the market index. The common calculation method is standard deviation:

[0035] in, is the market return rate on the i-th day, is the average rate of return, and N is the number of days in the time window. Event variables such as interest rate adjustments, tariff changes, and new regulations can be represented by 0 or 1 to indicate whether the event has occurred, or scored according to the impact of the policy. Correspondingly, the calculation method of the preliminary external risk score is as follows:

[0036] in, is the calculated preliminary external risk score, is the GDP growth rate, is the volatility, is the event variable, They are the weights of GDP growth rate, market volatility and event variables respectively.

[0037] Step 4: Input the preliminary internal risk score and the external risk score into the dynamic risk interaction model to calculate the dynamic interaction risk score, and optimize the preliminary internal risk score and the preliminary external risk score.

[0038] Specifically, there is a dynamic interaction between the internal financial situation of an enterprise and its external ecological environment. For example, when the market fluctuates greatly, the sales growth rate of an enterprise may be affected; similarly, policy changes may change the cash flow and financial leverage of an enterprise. Therefore, a dynamic risk interaction model is needed to capture the interaction between these factors.

[0039] In order to deal with the time series interaction between enterprise and external ecosystem data, a vector autoregression model (VAR) can be used, which can capture the interaction between multiple time series variables. ) and external risk score ( ) Regression analysis of these two time series can predict the internal and external risk levels of an enterprise at a certain point in the future.

[0040] If you want to handle more complex interactions and nonlinear features, you can use a long short-term memory network (LSTM). LSTM is suitable for processing time series data with long-term dependencies, such as capturing the impact of corporate operating conditions on future risks. The input of the LSTM model is corporate data and external data at multiple time steps, and the output is the comprehensive risk score of the enterprise in the future.

[0041] First, LSTM captures complex nonlinear relationships and long-term dependencies by introducing memory cells and gating mechanisms: 1. Hidden state update formula:

[0042] in, is the hidden layer state, which is used to save the information of the current time step. is the memory cell state, used to store long-term dependencies. is the output gate, controlling the update of the hidden state, is the hyperbolic tangent function.

[0043] 2. Memory unit update formula:

[0044] in, is the forget gate, which determines how much old information to discard. is the input gate, which determines the importance of new information. for new candidate memories.

[0045] 3. Time step input:

[0046] The input vector of LSTM contains both the internal risk score and the external risk score of the enterprise to capture the dynamic nonlinear relationship between the two.

[0047] Secondly, the model association formula: from linear to nonlinear. The linear output of VAR can be used as the baseline input of LSTM to further capture nonlinear relationships. This association is reflected in the following steps: 1. Linear modeling (VAR output as baseline)

[0048]

[0049] in, is the predicted value of the internal risk score calculated based on the vector autoregression model, is the predicted value of the external risk score calculated based on the vector autoregression model; are the intercept terms, , are the regression coefficients of internal risk scores, are the regression coefficients of external risk scores, are the error terms, is the optimized internal risk score, is the optimized external risk score, and t is the time index.

[0050] 2. Nonlinear expansion (LSTM captures residuals) We input the residual part of the VAR into the LSTM model to further capture the nonlinear relationship:

[0051] The input of the LSTM model is the residual of the VAR output and the original variable:

[0052] The LSTM model further optimizes the prediction by learning these nonlinear residuals:

[0053] in, Dynamic interaction risk scoring (e.g. future and joint forecast of 2017).

[0054] Finally, through the joint modeling of VAR and LSTM, a more complete risk assessment optimization formula can be obtained: 1. Internal risk scoring optimization:

[0055] 2. External risk scoring optimization:

[0056] The LSTM model can capture the complex relationship between the enterprise and the external ecosystem and achieve risk prediction through a multi-layer network structure.

[0057] Step 5: Calculate the comprehensive risk score based on the optimized internal risk score, external risk score, and dynamic interaction risk score:

[0058] in, is the calculated comprehensive risk score, They are the weights of internal risk score, external risk score and dynamic interaction risk score respectively.

[0059] The present invention can also automatically trigger an early warning when the enterprise's risk score exceeds a certain threshold, and recommend financial institutions or investors to take countermeasures, such as adjusting credit limits, improving risk control measures, etc. The enterprise's risk score can be dynamically adjusted by monitoring real-time changes in external economic data and enterprise operating data. It can also predict the enterprise's risk level in the future environment by simulating different policy changes or market fluctuations, and issue early warnings.

[0060] Based on the above embodiments, those skilled in the art can easily understand that the present invention also provides a comprehensive risk assessment device based on the interaction between an enterprise and an external environment, including a storage medium and a processor, wherein the storage medium stores a computer program, and the computer program is used to implement the above method when executed by the processor.

[0061] For example: Case 1: Green transformation assessment of a traditional steel company A traditional steel company produces steel products in a high-energy and high-pollution manner and has been under pressure from the low-carbon requirements of the international market in recent years. In order to meet the carbon neutrality goal, the company plans to invest in transforming the production line and introducing electric arc furnace technology to reduce carbon emissions. However, the transformation project faces the following challenges: 1. High initial investment: The cost of renovation is as high as 1 billion yuan, and a large loan is required.

[0062] 2. Technical and market risks: It is unclear whether the transformation can significantly reduce emissions while maintaining market competitiveness.

[0063] 3. Policy dependence: Enterprises hope to enjoy the benefits of green transformation financial policies, but they need to prove the feasibility of their transformation plans.

[0064] This application will be used to assess the risks and potential benefits of the company's green transformation and support financial institutions in making lending decisions.

[0065] Implementation steps: 1. Multi-source data collection Internal data of the enterprise Financial data: current debt-to-asset ratio (55%), current ratio (1.3).

[0066] 2) Expected cash flow after transformation, based on the operational forecast for the next five years.

[0067] Production data: 1) Annual steel output: 1 million tons.

[0068] 2) After the transformation, unit energy consumption is expected to drop by 30%.

[0069] 3) Current carbon emission intensity: 2.5 tons CO 2 / ton of steel.

[0070] External ecological data 1) Policy environment: The government promises to provide a 10% subsidy for green transformation projects, and the carbon emission rights trading price is about 200 yuan / ton.

[0071] 2) Market demand: Downstream customers (such as the automotive industry) have increased demand for low-carbon steel. The international market imposes a 15% carbon border tax on high-emission products.

[0072] 3) Industry dynamics: The number of transformed enterprises accounts for 20%, and their market competitiveness has been significantly enhanced after the transformation.

[0073] 2. Multidimensional feature extraction and processing Internal risk characteristics: 1) Debt-to-asset ratio (DAR) before transformation: DAR = Total Liabilities / Total Assets = 55%;

[0074] Predicted asset - liability ratio after transformation:

[0075] DAR after transformation = 65% Cash - flow coverage ratio (CFCR):

[0076] 1) External ecological characteristics: Change in carbon cost: Calculation: Reduction in carbon cost = (2.5 - 1.75) * 1000000 * 200 = 150 million yuan / year Change in market share: The demand for low - carbon steel by downstream customers increases by 10%. After the enterprise transformation, it is expected to increase its market share by 5%.

[0077] Risk interaction modeling 1) VAR model (short - term interaction modeling) Input: Financial data (such as current ratio, cash flow), carbon emission intensity, and industry dynamics.

[0078] Output: The impact of transformation on the enterprise's cash flow and risk score in the short term.

[0079] Model result: After the transformation is completed, the enterprise's cash flow is expected to increase by 150 million yuan, and the risk score drops by 10%.

[0080] LSTM model (long - term prediction modeling) Input: Time - series data such as carbon emission trading price, policy subsidy changes, etc.

[0081] Output: Changes in the enterprise's risk score and profitability in the next 5 years.

[0082] Model result: For every 10% increase in carbon price, the enterprise's profitability increases by 3%. The risk score shows a downward trend in the next 5 years, dropping from the current 60 (medium - high risk) to 40 (low risk).

[0083] 4. Comprehensive risk assessment By integrating internal and external risk data, calculate the comprehensive risk score:

[0084] The weights are set to: α=0.4: internal risk is more important.

[0085] β=0.4: External policy and market risks are equally important.

[0086] γ=0.2: interaction effect between internal and external factors.

[0087] Comprehensive rating results: , , , ; Comprehensive risk score result: 46 (medium-low risk).

[0088] 5. Opportunity identification and optimization suggestions 1) Policy opportunities: The project is eligible to apply for a 10% government subsidy. Carbon emissions are reduced by 30%, and it is expected to make a profit of 150 million yuan in the carbon trading market each year.

[0089] 2) Market opportunities: After the transformation, the products meet the low-carbon standards of the international market, avoid carbon border taxes, and reduce export costs by 15%. Downstream customer demand increased by 10%, which will help further expand market share.

[0090] 3) Risk control suggestions Financial risk control: It is recommended to control the debt-to-asset ratio below 60% to avoid over-financing.

[0091] Technical implementation guarantee: Require technology suppliers to provide performance guarantees to ensure that the transformation is completed on schedule.

[0092] 4) Loan decision: Recommend banks to approve loans and support enterprises to participate in the issuance of green financial bonds.

[0093] 5) Risk control measures: Companies are required to disclose their transformation progress on a regular basis.

[0094] Use carbon trading revenue as a source of security for loan repayment.

[0095] Case 2: Risk assessment and prediction for a manufacturing enterprise A manufacturing company mainly engages in export business, and its main income comes from the European market. The company has recently been hit by multiple external risks: 1) Macroeconomic risks: The European market is affected by the energy crisis, inflation is rising and market demand is weak.

[0096] 2) Policy adjustment risk: Increased import tariffs increase corporate costs.

[0097] 3) Industry competition risk: Intensified competition among domestic peers and price wars further squeezed profit margins.

[0098] The company applied for a loan from a financial institution to expand its production scale. The financial institution hopes to conduct a risk assessment on the company through this application to determine whether to approve the loan and the required risk control measures.

[0099] Implementation steps 1. Multi-source data collection 1) Internal data of the enterprise Financial statements: debt-to-asset ratio (60%), current ratio (1.2), sales growth rate (10% growth in the past three years).

[0100] Credit record: Good, no overdue payment record.

[0101] Cash flow data: Cash flow from operating activities is on a downward trend.

[0102] 2) External ecological data Macroeconomic data: European market GDP growth rate: 0.2% (significant slowdown). Inflation rate: 7.5% (continued to rise).

[0103] Industry data: Manufacturing PMI (Purchasing Managers Index): 48 (below the boom-bust line, industry demand is declining).

[0104] Policy changes: Europe imposes new tariffs on some raw materials, increasing tariffs by 5%.

[0105] 2. Multidimensional feature extraction and processing 1) Internal risk characteristics: Debt-to-Asset Ratio (DAR): DAR = 60% Current Ratio (CR): CR = 1.2 Sales Growth Rate (SGR): SGR = 10% 2) External ecological characteristics: GDP growth rate, inflation rate, and PMI data are input into the model as continuous time series.

[0106] Tariff changes are used as event variables, taking values ​​of 1 (occurred) or 0 (did not occur).

[0107] 3. Risk interaction modeling 1) VAR model (short-term interaction modeling) Input variables: internal risk scores of enterprises (such as DAR, CR) and external risk scores (such as GDP growth rate, PMI).

[0108] The VAR model captures the lagged effects of internal and external data, such as:

[0109] The model results show that the internal risk score is significantly affected by the external ecological data at a lag of one period (γ1=0.8\gamma_1 = 0.8γ1=0.8), especially the decline in PMI has a negative impact on sales.

[0110] 2) LSTM model (non-linear relationship modeling) Input: Time series data of internal and external ratings of enterprises.

[0111] Output: Risk score trend for the next 6 months.

[0112] The model shows that external inflation and tariff adjustments have had a significant impact on corporate cash flow in the long term. The corporate risk score will rise from 60 to 85 (100 is high risk) in the next three months.

[0113] 4. Comprehensive risk assessment By integrating internal and external risk data, a comprehensive risk score is calculated:

[0114] The weights are set to: α=0.4: internal risk is more important.

[0115] β=0.4: External policy and market risks are equally important.

[0116] γ=0.2: interaction effect between internal and external factors.

[0117] Comprehensive rating results: , , , ; Comprehensive risk score result: 76 (medium-high risk).

[0118] Step 5: Risk warning and decision support 1) Short-term risk warning: The sluggish GDP growth rate and the decline in PMI may lead to a further reduction in the order volume of enterprises in the short term. It is recommended that enterprises reduce their leverage ratio (the debt-to-asset ratio should be controlled below 50%).

[0119] 2) Long-term risk control: The long-term impact of inflation and tariffs may lead to a continuous deterioration in corporate cash flow. It is recommended to adjust the export market and reduce dependence on the European market.

[0120] 3) Loan decision-making: The bank is recommended to approve the loan, but with the following risk control measures: The enterprise is required to increase its working capital reserves. Regularly review the enterprise's cash flow and take timely measures when risks are found.

[0121] The above is only a preferred embodiment of the present invention. It should be noted that for ordinary technicians in this technical field, other parts not specifically described belong to the prior art or common knowledge. Several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

Claims

1. A comprehensive risk assessment method based on the interaction between an enterprise and the external environment, characterized in that: include: Step 1: Collect enterprise internal data and external environment system data; Step 2: pre-process the collected internal enterprise data and external environment system data, extract risk-related features from the internal enterprise data, and extract key macroeconomic and industry indicators from the external environment system data; Step 3: Input the extracted risk-related characteristics, macroeconomic and industry key indicators into the enterprise internal data risk model and external environment data risk model respectively, so as to calculate and obtain the preliminary internal risk score and the preliminary external risk score respectively; Step 4: Input the preliminary internal risk score and the external risk score into the dynamic risk interaction model to calculate the dynamic interaction risk score, and optimize the preliminary internal risk score and the preliminary external risk score; Step 5: Calculate the comprehensive risk score based on the optimized internal risk score, external risk score, and dynamic interaction risk score: ； in, is the calculated comprehensive risk score, is the optimized internal risk score, is the optimized external risk score, is the dynamic interaction risk score, t is the time index, They are the weights of internal risk score, external risk score and dynamic interaction risk score respectively.

2. According to claim 1, a comprehensive risk assessment method based on the interaction between an enterprise and the external environment is characterized in that: The internal enterprise data includes financial statements, transaction data and credit records; the external environment system data includes industry data, global economic data and policy change data.

3. The comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 1 is characterized in that: The pre-processing method is as follows: Align and interpolate data at different time scales to align the time series; Standardize or normalize the data.

4. A comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 1, characterized in that: The risk-related characteristics include debt-to-equity ratio, current ratio, sales growth rate and cash flow coverage ratio.

5. A comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 4, characterized in that: The preliminary internal risk score is calculated as follows: ； in, is the calculated preliminary internal risk score, is the debt-to-asset ratio, is the current ratio, is the sales growth rate, is the cash flow coverage ratio, , , , They are the weights of the debt-to-asset ratio, current ratio, sales growth rate and cash flow coverage ratio respectively.

6. A comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 1, characterized in that: The key macroeconomic and industry indicators include GDP growth rate, market volatility and event variables.

7. A comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 6, characterized in that: The preliminary external risk score is calculated as follows: ； in, is the calculated preliminary external risk score, is the GDP growth rate, is the volatility, is the event variable, , , They are the weights of GDP growth rate, market volatility and event variables respectively.

8. The comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 1 is characterized in that: The dynamic interaction risk score is calculated as follows: ； in, is the calculated dynamic interaction risk score, Indicates the use of long short-term memory network for processing, are the intercept terms, are the regression coefficients of internal risk scores, are the regression coefficients of external risk scores, are the error terms, is the optimized internal risk score, is the optimized external risk score, is the predicted value of the internal risk score calculated based on the vector autoregression model, is the predicted value of the external risk score calculated based on the vector autoregression model.

9. The comprehensive risk assessment method based on the interaction between an enterprise and the external environment according to claim 8 is characterized in that: The optimized internal risk scores and external risk scores are as follows: 。 10. A comprehensive risk assessment device based on the interaction between an enterprise and an external environment, comprising a storage medium and a processor, wherein the storage medium stores a computer program, characterized in that: When the computer program is executed by a processor, it is used to implement the method according to any one of claims 1 to 9.

Images