Enterprise fund scheduling optimization method and system based on artificial intelligence

Abstract

The invention provides an enterprise fund scheduling optimization method and system based on artificial intelligence. The method comprises the following steps: S1, acquiring and processing internal financial historical data of an enterprise; s2, training enterprise historical fund flow data by using a machine learning model, and establishing a fund demand prediction model; s3, extracting high-relevancy features in the external data, and optimizing the prediction model; s4, integrating internal and external detection analysis results, and generating real-time risk assessment early warning based on a set threshold value; s5, generating an optimized fund scheduling scheme based on fund demand prediction and risk assessment results; and S6, executing the fund scheduling scheme, monitoring an execution process in real time, detecting an abnormal condition, adjusting and optimizing the fund scheduling scheme, summarizing an execution condition and an abnormal processing result, generating and feeding back a scheduling report, and providing decision support. The technical problem that the prior art lacks flexibility, real-time performance, dynamic prediction, model optimization and automatic and intelligent decision support is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise fund scheduling, and in particular to an optimization method and system for enterprise fund scheduling based on artificial intelligence. Background Art

[0002] Currently, enterprises mainly rely on manual experience, traditional enterprise fund management systems, and financial intelligent optimization platforms in fund scheduling and management. Among them, traditional fund management systems and financial intelligent optimization platforms mainly optimize fund flow, fund scheduling, repayment plans, etc. through pre-set rules and models. However, these systems and platforms often perform scheduling based on static rules or simple optimization algorithms, lacking intelligent data mining and prediction capabilities; therefore, in scenarios such as fund scheduling and fund flow prediction, they lack real-time performance and cannot dynamically respond to rapid changes in the external market and within the enterprise; moreover, scheduling decisions usually rely on manual analysis and decision-making, lacking automated and intelligent decision support, and the analysis of fund scheduling history is limited to basic statistical functions, failing to deeply explore the laws in the data and being difficult to optimize the analysis and provide decision support for future scheduling. Summary of the Invention

[0003] The purpose of the present invention is to provide an optimization method and system for enterprise fund scheduling based on artificial intelligence, which solves the technical problems of the existing technology lacking flexibility and real-time performance, dynamic prediction and model optimization, as well as automated and intelligent decision support.

[0004] In the first aspect, the present invention provides an optimization method for enterprise fund scheduling based on artificial intelligence, including the following steps: S1. Obtain the internal financial historical data of the enterprise and process the internal data into an available format; S2. Use a machine learning model to train the enterprise historical fund flow data, establish a fund demand prediction model, and use the internal historical data to train the model and adjust the parameters; S3. Obtain and process external financial data, extract the influencing factors in the external data and perform a correlation analysis with the historical fund demand data, extract the features with high correlation, supplement them to the machine learning model to optimize the prediction model, and output the fund demand prediction for each future time period through the prediction model; S4. Real-time collect and read the fund flow data of the enterprise internal financial system to calculate the fund status, use an algorithm to detect abnormal fund flow; through external data, identify and analyze external risks and calculate the external risk index; based on the comprehensive internal and external detection and analysis results, generate a real-time risk assessment warning and delimit the warning level based on the set threshold; S5. Generate an optimized fund scheduling plan based on the fund demand prediction and risk assessment results, and use a genetic algorithm to generate the best fund scheduling plan; S6. Execute the optimized fund scheduling plan, monitor the execution process in real time, detect abnormal situations, adjust and optimize the fund scheduling plan, summarize the execution situation and the results of exception handling, generate and feedback a scheduling report, and provide decision-making support.

[0005] Further, in step S3, the impact factors in the external data are extracted, including extracting the information that potentially affects the fund demand, performing semantic parsing on the text to extract keywords, and evaluating the sentiment tendency of the external data to determine the external risks that may affect the fund demand.

[0006] Further, the keywords and sentiment tendency extracted externally are added to the feature set of the prediction model, the features from different sources are standardized and weighted, the multi-model integration method is used to improve the prediction accuracy, and the fund demand for each future time period is predicted based on the comprehensive model output.

[0007] Further, in step S4, algorithms are used to detect abnormal fund flows, including setting the ratio of outflows exceeding the balance or the number of days of inflow delay as a fixed threshold, using unsupervised algorithms to detect abnormalities and time series prediction abnormalities, establishing a baseline model for normal fund flows, performing deviation detection on real-time data, and marking events outside the range.

[0008] Further, in step S4, external risks are identified and analyzed, and an external risk index is calculated, including analyzing the key information and sentiment tendency based on external news and policy data, classifying the analysis results into risks, calculating the risk index based on the sentiment score and event frequency, and determining the risk index threshold.

[0009] Further, in step S5, the genetic algorithm includes gene coding, initialization of cluster generation, fitness function design, and genetic operations; in the gene coding, the time of fund scheduling, the corresponding fund source ID, the amount of scheduling, and the fund corresponding demand are used as the time point, fund source, fund amount, and usage target included in each gene coding in turn; in the initialization of cluster generation, some initial solutions are generated based on the results of risk assessment and demand prediction to improve the quality of the initial population.

[0010] Further, in step S6, the optimized fund scheduling plan is executed to detect abnormal situations, including using the enterprise financial system API to execute the fund transfer of internal accounts, calling the external financial institution or payment gateway API to complete transfer and payment operations, parsing the specific execution results according to the status code returned by the API, recording the operation content, and identifying abnormal situations.

[0011] Further, step S6 summarizes the execution status and exception handling results to provide decision-making support, including summarizing the scheduling execution success rate, the number and reasons of failed tasks, and the execution effect after adjustment, providing visual data to display the overall execution status, listing all exception events and their handling processes, and providing specific suggestions for failed tasks.

[0012] In a second aspect, the present invention also provides a system for implementing the above-mentioned enterprise fund scheduling optimization method based on artificial intelligence, including: A collection and processing module: used to obtain and structurally process the internal financial historical data and external financial data of the enterprise; A prediction and modeling module: built with a machine learning model, establishing a fund demand prediction model by training historical fund flow data, and optimizing the model parameters by integrating highly relevant features extracted from external data; A risk assessment module: including an anomaly detection unit and an external risk assessment unit. The anomaly detection unit uses an unsupervised algorithm and time series to establish a baseline model to identify fund flow deviations. The external risk assessment unit analyzes key information and sentiment tendencies based on natural language processing technology, calculates a risk index, and finally comprehensively evaluates the fund scheduling risk in real time based on the internal and external analysis results; A scheduling optimization module: based on the input predicted fund demand and risk assessment information, using a genetic algorithm to optimize the fund scheduling plan; An execution monitoring module: executes the optimized fund scheduling plan, monitors the execution process in real time, and adjusts and optimizes the scheduling plan when an anomaly occurs; A generation and feedback module: generates and feedbacks a scheduling report for decision-making use.

[0013] Beneficial effects:

[0014] The present invention provides an enterprise fund scheduling optimization method and system based on artificial intelligence. By introducing advanced technologies such as artificial intelligence technology, machine learning models, and natural language processing, it solves multiple problems in traditional fund scheduling systems and realizes the intelligence, automation, and real-time of fund scheduling. The system can optimize the fund scheduling plan, reduce costs, enhance risk management capabilities, automatically generate reports, and has strong scalability and flexibility, providing efficient and accurate fund scheduling services for enterprises, and helping enterprises maintain the health and stability of fund flow in a dynamic market environment.

[0015] Specifically, the system integrates internal financial data with external financial information, constructs dynamic prediction and optimization models using machine learning and genetic algorithms, significantly improves the accuracy of capital demand prediction and scheduling efficiency; combines a real-time risk assessment mechanism to achieve anomaly detection and external risk quantification warning, and responds to potential threats at different levels; the full-process automated execution and closed-loop feedback system ensure the flexibility and transparency of capital scheduling, maximize resource utilization while reducing manual intervention, and provide data-driven intelligent decision-making support for enterprises. Brief Description of the Drawings

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0017] Figure 1 It is an application block diagram of the enterprise capital scheduling optimization system provided in this embodiment; Figure 2 It is a structural block diagram of the enterprise capital scheduling optimization system provided in this embodiment; Figure 3 It is a flow block diagram of the enterprise capital scheduling optimization method provided in this embodiment. Detailed Embodiments

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following will describe in detail some embodiments of the present invention with reference to the drawings.

[0019] This embodiment provides an artificial intelligence-based enterprise capital scheduling optimization method. Please refer to Figure 3 As shown, this method obtains raw data from multiple internal and external data sources. Among them, the enterprise internal financial system includes ERP, CRM, report systems, etc. Clean and format the obtained unstructured data (text files (PDF, Word), scanned documents, email contents, etc.) and semi-structured data (JSON, XML, Excel spreadsheets), and convert them into structured data; extract keywords from the text data through natural language processing (NLP) and output them to the specified storage databases MySQL, local files, or data warehouses for use by subsequent modules.

[0020] Use machine learning models to train historical fund flow data and establish a fund demand prediction model. The models used are: time series model (ARIMA), regression model (XGBoost), and regression enhanced model based on time series features (HybridModels); use internal historical data to train the model, adjust parameters, consider external features and add them to the prediction model to enhance the robustness of the model, and use cross-validation to optimize the model effect.

[0021] Use QWEN to analyze external financial institution data, extract information that may potentially affect fund demand, clean and organize external data (such as policy announcements, economic news), use QWEN to perform semantic parsing on the text, extract keywords and entities (such as "interest rate hike", "monetary policy", etc.), evaluate the sentiment tendency (positive, negative) in financial news and industry reports, and determine external risks that may affect fund demand; conduct a correlation analysis between the factors extracted from external data and historical fund demand data, extract highly correlated features, and supplement them to the machine learning model.

[0022] Integrate internal and external data to generate future fund demand predictions; add external analysis results (such as keywords and their sentiment extracted by QWEN) to the feature set of the prediction model, and perform standardization and weighting processing on features from different sources; use multi-model integration methods (such as weighted average, stacking model) to improve prediction accuracy, and output future fund demand predictions for each time period based on the integrated model.

[0023] Provide an intuitive presentation of fund demand predictions, including showing the time trend of fund demand predictions (line chart), the changing trends of superimposed external influencing factors, the influence weights of key factors on the predictions (bar chart, pie chart), summarizing the fund demand prediction results and risk warnings.

[0024] Real-time analyze potential risks in fund scheduling (such as fund shortage, repayment delay, etc.), monitor the current fund flow situation, and calculate the fund surplus / deficit in real time; combine QWEN to perform sentiment analysis on market news, policy changes, etc. to identify potential risks; use algorithms (such as anomaly detection) to identify abnormal situations in fund flow; automatically generate risk warning reports to alert management of potential risks.

[0025] Connect to the enterprise's internal financial system, read fund flow data in real time (API or database subscription mechanism), collect data at the second or minute level, perform fund status calculations, fund surplus / deficit (fund status = current balance + expected inflows - expected outflows), fund status = current balance + expected inflows - expected outflows.

[0026] Use algorithms to detect abnormal fund flows, set fixed thresholds (such as the ratio of outflows exceeding the balance or the number of days of inflow delay), use unsupervised algorithms (IsolationForest) to detect anomalies and time series prediction anomalies (LSTMAutoEncoder), establish a baseline model for normal fund flows, perform deviation detection on real-time data, and mark events outside the range, including excessive short-term fund outflows, expected inflows not arriving on time, and abnormal fluctuations not seen in historical patterns.

[0027] Through external news and policy data, identify potential market risks, automatically scrape market news and policy announcements (RSS, news API), use the QWEN model to analyze the sentiment of news content (positive, neutral, negative), identify key information such as events, institutions, and policy names (such as "interest rate hike"), classify the analysis results into macroeconomic risks, industry policy risks, or market sentiment risks, and calculate the market risk index based on sentiment scores and event frequencies.

[0028] Integrate internal and external analysis results to generate real-time warnings. Use the proportion or absolute value of the fund gap exceeding the set value and the frequent occurrence of abnormal events as thresholds. When the external market sentiment risk index reaches the set threshold, increase the internal risk level. The warning levels are divided into green (normal, no obvious risk), yellow (warning, there are certain potential risks), and red (urgent, serious risks need to be dealt with immediately).

[0029] Regularly or immediately generate risk warning reports. The report content includes an overview of the real-time fund status, detailed records of abnormal detection events, a summary of external sentiment analysis risks, a comprehensive risk rating, and response suggestions, and is visually presented with surplus / deficit trend charts, time axes of abnormal flow events, and curves of changes in external risk indices.

[0030] Based on the results of fund demand forecasting and risk assessment, generate an optimized fund scheduling plan. Use the genetic algorithm (GA) to generate the best scheduling plan and provide the optimized fund scheduling plan for subsequent execution. The core design of the genetic algorithm is as follows: 1. Gene encoding: Each gene encoding contains a time point (the time of fund scheduling), a fund source (the corresponding fund source ID), an amount of funds (the amount of scheduling), and a usage target (the demand corresponding to the funds).

[0031] 2. Initial cluster generation: Randomly generate several legal fund scheduling plans, and generate some initial plans based on the results of risk assessment and demand forecasting to improve the quality of the initial population.

[0032] 3. Fitness Function Design: According to the optimization objective weights, design the comprehensive fitness function F = w1 * cost optimization + w2 * risk aversion + w3 * liquidity guarantee. Calculate the total cost based on the scheduling plan, and calculate the total risk value according to the risk of the funding source and the amount of scheduling.

[0033] 4. Genetic Operations: Use the roulette wheel selection method or the tournament selection method to ensure that individuals with high fitness have a higher probability of being selected. Perform single-point crossover or multi-point crossover to combine two parent individuals to generate new offspring individuals. Randomly adjust some genes (such as changing the funding source or amount). The mutation rate needs to be dynamically adjusted according to the problem scale. Retain several individuals with the highest fitness in the current population to prevent excellent solutions from being eliminated.

[0034] Automatically execute the fund scheduling operations (such as fund transfer, payment, etc.) according to the plan of the fund scheduling optimization module. Analyze the optimization plan, decompose the fund scheduling task into executable units (such as multiple transfer tasks), call the enterprise financial system API to execute the fund allocation of internal accounts, and call the external financial institution or payment gateway API to complete operations such as transfer and payment. Monitor the scheduling execution process in real-time (waiting, in execution, successful, failed), detect whether there are problems, and analyze the specific execution results according to the status code returned by the external API. When an exception occurs (such as insufficient funds, payment failure), record the start time, end time, execution status, and feedback information of each operation, and automatically identify exceptions such as insufficient funds, payment failure, and timeout not completed.

[0035] When an exception occurs, adjust the scheduling plan and resume the operation. Optimize the existing fund scheduling plan (such as adjusting the priority, reallocating funds). For payment failure tasks, set automatic retry (limiting the number of times and time intervals), start the emergency fund pool for supplementary allocation, and push the exception reasons and handling measures to the management personnel (email, SMS, dashboard notification).

[0036] Summarize the execution situation and exception handling results to provide decision support, including the scheduling execution success rate, the number and reasons of failed tasks, and the execution effect after adjustment. Provide visual data (pie chart, trend chart) to display the overall execution status, list all exception events and their handling processes, and provide specific suggestions for failed tasks (such as contacting the bank for verification).

[0037] Automatically generate reports including fund demand forecasting, fund flow, risk assessment, etc., aggregate the output data of each module. Integrate them into a unified format, connect to the APIs or databases of the fund demand forecasting, fund flow monitoring, and risk assessment modules; extract key content such as fund surplus / gap, abnormal event description, risk level, etc., and aggregate data to generate summary information. Generate a comprehensive report based on the data, and conduct analysis of the current fund scheduling and feasibility analysis of potential problems through line charts, pie charts, and bar charts.

[0038] Use the email service (SMTP) to send reports, integrate enterprise messaging tools (such as Slack, DingTalk, WeChat Enterprise Edition), distribute different report versions according to the roles and permissions of the decision-making layer, support on-demand sending (real-time generation) and scheduled sending (daily, weekly); record the sending status (success, failure), record the sending status (success, failure). Record the whole process of report generation and sending, detailed records of each report generation and sending operation, including timestamps, operation status, save snapshots of the generated report content for historical traceability; trigger an alarm when sending fails (notify the administrator by email or text message).

[0039] This embodiment also provides a system for implementing the above-mentioned enterprise fund scheduling optimization method based on artificial intelligence. Please refer to Figure 2 As shown, it includes: a collection and processing module: used to obtain and structurally process the enterprise's internal financial historical data and external financial data; a prediction and modeling module: built-in with machine learning models, establish a fund demand forecasting model by training historical fund flow data, and optimize the model parameters by integrating highly relevant features extracted from external data; a risk assessment module: including an anomaly detection unit and an external risk assessment unit. The anomaly detection unit uses unsupervised algorithms and time series to establish a baseline model to identify fund flow deviations. The external risk assessment unit analyzes key information and sentiment tendencies based on natural language processing technology, calculates the risk index, and finally comprehensively evaluates the fund scheduling risk in real time based on the internal and external analysis results; a scheduling optimization module: based on the input predicted fund demand and risk assessment information, use genetic algorithms to optimize the fund scheduling plan; an execution and monitoring module: execute the optimized fund scheduling plan, monitor the execution process in real time, and adjust and optimize the scheduling plan when anomalies occur; a generation and feedback module: generate and feedback scheduling reports for decision-making use.

[0040] Specifically, the system constructs a full-process intelligent closed-loop for enterprise fund scheduling through multi-module collaboration: The acquisition and processing module integrates internal and external data and conducts structured processing to provide high-quality input for decision-making; The prediction and modeling module combines machine learning with externally highly relevant features (such as policy text sentiment analysis) to enhance the dynamics and accuracy of fund demand prediction; The risk assessment module innovatively combines unsupervised algorithms (identifying abnormal fund flows) with natural language processing (quantifying external public opinion risk indices) to achieve two-dimensional three-dimensional perception of internal and external risks; The scheduling optimization module dynamically weighs costs, risks, and liquidity based on genetic algorithms to generate a globally optimal solution; The execution and monitoring module tracks the scheduling execution status in real time and automatically triggers an abnormal adjustment and optimization mechanism to ensure the smooth execution of the plan; The generation and feedback module reduces the decision-making threshold and improves the response efficiency through visual reports and multi-role push mechanisms.

[0041] In addition, as Figure 1 shown, the system adopts the release mode of WEB applications, and the working steps of the computer application are as follows: 1. Data acquisition and preprocessing: Obtain data from the enterprise's internal financial system (such as cash flow data, borrowing information, repayment plans, etc.) and external financial institutions (such as settlement data, fund balances, market exchange rates, etc.); Use the PC-side service to clean the input unstructured data, remove redundant information, and format it into structured data. By docking with an AI model (such as QWENNLP), process the data, extract the key information, and analyze the emotions and important information in the text data using sentiment analysis and keyword extraction techniques.

[0042] 2. The system uses machine learning algorithms to train the enterprise's historical fund flow data through the historical financial data analysis module to predict future fund demands. The system combines the prediction results output by QWEN with the internal financial data to generate a time series prediction of fund demands, persists it in the database, and provides a basis for subsequent scheduling, generates a review report, and pushes it to the decision-makers.

[0043] 3. Enterprise users can access the PC-side application service by entering the website address through the browser and clicking on the website address. The PC-side application has built-in reports such as cash flow prediction, fund scheduling analysis, and settlement situation analysis for enterprise users to view. At the same time, it has a built-in fund scheduling function, providing intelligent scheduling and manual scheduling methods, and connecting to external financial institutions through bank-enterprise direct connection for real-time payment and query.

[0044] 4. Based on the natural language generation technology of the AI model, the PC-side service enables the system to achieve intelligent interaction actions, facilitating the understanding and decision-making of enterprise management.

Claims

1. An optimization method for enterprise fund scheduling based on artificial intelligence, characterized in that, It includes the following steps: S1. Obtain the internal financial historical data of the enterprise and process the internal data into an available format; S2. Use a machine learning model to train the enterprise's historical fund flow data, establish a fund demand prediction model, and use the internal historical data to train the model and adjust the parameters; S3. Obtain and process external financial data, extract the influencing factors in the external data, conduct a correlation analysis with the historical fund demand data, extract the highly correlated features, supplement them to the machine learning model to optimize the prediction model, and output the fund demand predictions for future time periods through the prediction model; S4. Real-time collect and read the fund flow data of the enterprise's internal financial system for fund status calculation, and use an algorithm to detect abnormal fund flows; Identify and analyze external risks through external data and calculate the external risk index; Based on the combined internal and external detection and analysis results and based on the set thresholds, generate a real-time risk assessment warning and delimit the warning level; S5. Based on the fund demand prediction and risk assessment results, generate an optimized fund scheduling plan, and use a genetic algorithm to generate the best fund scheduling plan; S6. Execute the optimized fund scheduling plan, monitor the execution process in real time, detect abnormal situations, adjust and optimize the fund scheduling plan, summarize the execution situation and the results of abnormal handling, generate and feedback a scheduling report, and provide decision support.

2. The optimization method for enterprise fund scheduling based on artificial intelligence according to claim 1, characterized in that In step S3, the influencing factors in the external data are extracted, including extracting the information that has a potential impact on the fund demand, conducting semantic parsing on the text to extract keywords, and evaluating the sentiment tendency of the external data to determine the external risks that may affect the fund demand.

3. The optimization method for enterprise fund scheduling based on artificial intelligence according to claim 2, characterized in that Add the keywords and sentiment tendency extracted externally to the feature set of the prediction model, perform standardization and weighting processing on the features from different sources, use the multi-model integration method to improve the prediction accuracy, and output the fund demand predictions for future time periods based on the comprehensive model.

4. An optimization method for enterprise fund scheduling based on artificial intelligence according to claim 1, characterized in that, In step S4, an algorithm is used to detect abnormal fund flows, including setting the ratio of the outflow exceeding the balance or the number of days of inflow delay as a fixed threshold, using unsupervised algorithms to detect anomalies and time series prediction anomalies, establishing a baseline model for normal fund flows, conducting deviation detection on the real-time data, and marking the events that exceed the range.

5. The optimization method for enterprise fund scheduling based on artificial intelligence according to claim 4, characterized in that In step S4, external risks are identified and analyzed and the external risk index is calculated, including analyzing the key information and sentiment tendency in the external news and policy data based on the external news and policy data, classifying the analysis results into risks, calculating the risk index based on the sentiment score and event frequency, and determining the risk index threshold.

6. The enterprise fund scheduling optimization method based on artificial intelligence according to claim 1, characterized in that In step S5, the genetic algorithm includes gene encoding, initialization of cluster generation, fitness function design, and genetic operations; in the gene encoding, the time of fund scheduling, the corresponding fund source ID, the amount of scheduling, and the demand corresponding to the funds are used as the time point, fund source, fund amount, and usage target included in each gene encoding in turn; in the initialization of cluster generation, some initial plans are generated based on the results of risk assessment and demand prediction to improve the quality of the initial population.

7. An optimization method for enterprise fund scheduling based on artificial intelligence according to claim 1, characterized in that, Step S6 executes the optimized fund scheduling plan and detects abnormal situations, including performing fund transfers of internal accounts using the enterprise financial system API, invoking external financial institutions or payment gateway APIs to complete transfer and payment operations, parsing the specific execution results according to the status codes returned by the APIs, recording the operation content and identifying abnormal situations.

8. A method for optimizing enterprise fund scheduling based on artificial intelligence according to claim 1, characterized in that Step S6 summarizes the execution situation and the results of exception handling to provide decision-making support, including summarizing the success rate of scheduling execution, the number and reasons of failed tasks, and the execution effect after adjustment, providing visual data to display the overall execution status, listing all abnormal events and their handling processes, and providing specific suggestions for failed tasks.

9. A system for implementing the optimization method of enterprise fund scheduling based on artificial intelligence as claimed in any one of claims 1-8, characterized in that, Including: Collection and processing module: used to obtain and structurally process the enterprise's internal financial historical data and external financial data; Prediction modeling module: built-in machine learning model, establishing a fund demand prediction model by training historical fund flow data, and optimizing the model parameters by integrating highly relevant features extracted from external data; Risk assessment module: includes an anomaly detection unit and an external risk assessment unit. The anomaly detection unit uses unsupervised algorithms and time series to establish a baseline model to identify fund flow deviations. The external risk assessment unit analyzes key information and sentiment tendencies based on natural language processing technology, calculates the risk index, and finally comprehensively evaluates the fund scheduling risk in real time based on the internal and external analysis results; Scheduling optimization module: based on the input predicted fund demand and risk assessment information, using genetic algorithms to optimize the fund scheduling plan; Execution monitoring module: executes the optimized fund scheduling plan, monitors the execution process in real time, and adjusts and optimizes the scheduling plan when anomalies occur; Generation and feedback module: generates and feedbacks scheduling reports for decision-making use.

Images