Intelligent financial statement analysis system based on cloud AI technology

Abstract

The invention discloses an intelligent financial statement analysis system based on a cloud AI technology, and relates to the technical field of data management. According to the intelligent financial statement analysis system based on the cloud AI technology, the collected data are classified through preprocessing, the processed data are synchronously extracted to realize association of various evaluation indexes, a machine learning prediction model is established, and real-time data are introduced into the machine learning prediction model to predict and evaluate financial conditions. The artificial intelligence technology is utilized to carry out deep mining and analysis on financial data, interference of human factors is reduced, accuracy and objectivity of analysis results are improved, association of financial index calculation and machine learning analysis is comprehensively utilized to form a machine learning prediction model, financial information analysis is deeply carried out, the financial condition of an enterprise is monitored in real time, and the economic benefit of enterprises is improved. Potential financial risks can be found in time, and decision support is provided for enterprise managers.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data management, in particular to an intelligent financial statement analysis system based on cloud AI technology. BACKGROUND

[0002] With the rapid development of cloud computing and artificial intelligence technology, new opportunities have been brought to financial statement analysis. Cloud computing technology can provide powerful computing power and storage resources, which can handle massive financial data. Machine learning and natural language processing methods in artificial intelligence technology can deeply mine and analyze financial data to discover potential patterns and rules in the data.

[0003] Referring to the patent entitled "Intelligent financial statement analysis system based on cloud AI technology" (Patent Publication No. CN116542800A, Patent Publication Date: 2023-08-04), it includes a cloud architecture unit, a financial statement extraction unit, a data processing unit, an index analysis unit, and an intelligent decision-making unit. The cloud architecture unit is used to build a cloud storage for financial statement data. The user can issue voice commands to extract corresponding financial statement data through the financial statement extraction unit. The index analysis unit can analyze and calculate the financial indicators of the financial statement data corresponding to the user's voice command. This improves the control effect and facilitates targeted analysis. At the same time, the intelligent decision-making unit can visualize the financial statement data to improve accuracy.

[0004] Based on the above description, the existing financial statement analysis mainly relies on manual analysis, which is easily affected by the professional level and subjective factors of the analysts, making it difficult to ensure the accuracy and objectivity of the analysis results. Moreover, traditional analysis methods have obvious shortcomings in data processing capacity and mining depth when facing massive and complex financial data, and cannot timely discover potential financial risks and business problems. Therefore, the present application provides an intelligent financial statement analysis system based on cloud AI technology. SUMMARY

[0005] To overcome the shortcomings of the prior art, the present application provides an intelligent financial statement analysis system based on cloud AI technology, which solves the problem of obvious shortcomings in data processing capacity and mining depth of existing analysis methods, and cannot timely discover potential financial risks and business problems.

[0006] To achieve the above purpose, the present application realizes the following technical scheme: an intelligent financial statement analysis system based on cloud AI technology, comprising: A financial data acquisition module is responsible for collecting various financial information of enterprises and collecting publicly disclosed financial statement data of external enterprises. The data is compressed and transmitted. The cloud storage and AI analysis module receives and decompresses the data, classifies the collected data through preprocessing, stores the collected financial data into the cloud server using distributed storage technology, synchronously extracts the processed data to realize the correlation of various evaluation indicators, and establishes a machine learning prediction model. Real-time data is introduced into the machine learning prediction model to make a prediction and evaluation of the financial situation. The intelligent early warning module realizes different levels of early warning operations based on the results of the prediction and evaluation, and generates corresponding processing strategies for reference by enterprise personnel. The visual display module provides an interactive interface for data query operations, and controls data access and operation at different levels according to the roles and permissions of users.

[0007] Preferably, the operation of classifying the collected data through preprocessing in the cloud storage and AI analysis module is: The intelligent table is composed of column titles and row titles, the content of the column titles is the item category, and the content of the row titles is the timestamp. The intelligent table realizes data extraction and confirmation through the set correlation verification model and numerical processing window. The numerical processing window is used to clean, denoise and normalize the collected data. Data cleaning is used to handle missing values, outliers and repeated values in the data. Data denoising is used to eliminate random errors or irrelevant signals in the data. Data normalization is used to scale the data to a unified range. The processed data is classified in the intelligent table and used with the index module.

[0008] Preferably, the operation of realizing data extraction and confirmation through the set correlation verification model and numerical processing window in the intelligent table is: The numerical processing window sets multiple moving correlation windows on the intelligent table to perform traversal operations. The traversal trajectory starts from the first row title and moves along the column title direction until the last column title with content in the current direction is traversed. Then the correlation window moves to the next row title for traversal. After the multiple correlation windows realize data extraction, the correlation verification model is introduced based on the extracted content to trace the data credentials, and the data abnormality is analyzed and determined to be fed back to the numerical processing window for processing.

[0009] Preferably, the multiple correlation windows perform traversal operations on the intelligent table are: The plurality of associated windows include a row title window (a1, b1), a column title window (a2, b2) and a numerical result window (a2, b1), the row title window (a1, b1) is only used for identifying the row title content in the table, the column title window (a2, b2) is only used for identifying the column title content in the table, and the numerical result window (a2, b1) is only used for identifying the content in the table extending across the corresponding column title and row title; When the row title window (a1, b1) is fixed, the column title window (a2, b2) and the numerical result window (a2, b1) keep moving at the same frequency, and when the column title window (a2, b2) returns to the initial column title table, the numerical result window (a2, b1) keeps moving at the same frequency while also moving at the same frequency with the row title window (a1, b1); And the vertical width b1 of the row title window (a1, b1) and the numerical result window (a2, b1) is expanded synchronously, and the horizontal length a2 of the column title window (a2, b2) and the numerical result window (a2, b1) is expanded synchronously.

[0010] Preferably, the operation of introducing the associated verification model based on the extracted content traceability data certificate is: The extracted item category result under the corresponding timestamp is matched with the corresponding storage data in the cloud server; If the content extracted by the numerical result window is blank, it is identified whether there is a corresponding timestamp under the corresponding item category, if no corresponding timestamp is generated, the current numerical result is blank, otherwise the current data column is abnormal, the certificate seal and transaction certificate under the corresponding item category and timestamp are verified, and when the certificate seal and transaction certificate are correct, the numerical result extraction is filled into the data column where the numerical result window is located, and the expansion of the data column is realized according to the specific numerical result; If the content extracted by the numerical result window is a normal value, the certificate seal and transaction certificate under the corresponding item category and timestamp are identified and verified, and when the certificate seal and transaction certificate are correct, the extracted normal value is compared with the traceable numerical result, and if they are consistent, there is no abnormality, otherwise the current data column is abnormal, the numerical result is extracted and replaced with the data in the data column where the numerical result window is located, and the expansion of the data column is realized according to the specific numerical result; If the content extracted by the numerical result window is a wrong code, the current data column is abnormal, the certificate seal and transaction certificate under the corresponding item category and timestamp are identified and verified, and when the certificate seal and transaction certificate are correct, the traceable numerical result is extracted and replaces the wrong code in the data column where the numerical result window is located, and the expansion of the data column is realized according to the specific numerical result; When the certificate seal and transaction certificate are abnormal, data checking needs to be fed back to the corresponding income and expenditure party for data extraction.

[0011] Preferably, the operation of expanding the data column according to the specific numerical result is as follows: Extract the numerical result, and determine the expansion distance of the data column and the numerical result window (a2, b1) according to the number of characters of the numerical result; Set the number of characters as c, the single horizontal distance of the character as f, the single vertical distance of the character as g, the spacing between characters as d, and the maximum horizontal length of the data column as e. When [c×f+ (c+1)×d]≤a2, the current data column does not need to be expanded. When a2<[c×f+ (c+1)×d]≤e, the current data column is only horizontally expanded, and the horizontal expansion length is [c×(f+d)-a2]. When [c×f+ (c+1)×d]>e, the current data column is synchronously expanded horizontally and vertically, and the horizontal expansion length is e, and the vertical expansion width is [[c×(f+d) / e]-1]×(g+d)+(g+2d). Then, the expansion operation of the data column is realized according to the expansion conditions. After the horizontal distance of the current data column is expanded, the horizontal length of all vertical data columns and the horizontal length of the numerical result window (a2, b1) corresponding to the current data column are synchronously expanded. After the vertical distance of the current data column is expanded, the vertical width of all horizontal data columns and the vertical width of the numerical result window (a2, b1) corresponding to the current data column are synchronously expanded.

[0012] Preferably, the operation of correlating the evaluation indexes of the processed data in the cloud storage and AI analysis module and establishing a machine learning prediction model is as follows: Extract the processed data, and uniformly divide the data into a training set and a test set in a ratio of 8:2; Calculate the solvency index, profitability index, and operating capacity index from the historical data in the training set within the same period, compare the calculated index results with the market index threshold interval to determine the financial situation in the current period, and correlate the solvency index, profitability index, and operating capacity index to form a machine learning prediction model; Introduce the test set into the machine learning prediction model for testing and complete the optimization operation.

[0013] Preferably, the calculation method of the solvency index, profitability index, and operating capacity index is as follows: In the solvency index, the current ratio, quick ratio, and asset-liability ratio are calculated and evaluated. The current ratio = total current assets / total current liabilities, the quick ratio = (current assets - inventory) / total current liabilities, and the asset-liability ratio = total liabilities / total assets × 100%; Among the profitability indicators, the net profit margin, the net asset yield and the cost of profit margin are calculated and evaluated, the net profit margin = net profit / sales revenue x 100%, the net asset yield = net profit / average net assets x 100%, the cost of profit margin = total profit / cost of total expenses x 100%, Among the operating capacity indicators, the inventory turnover rate, the accounts receivable turnover rate and the total asset turnover rate are calculated and evaluated, the inventory turnover rate = sales cost / average inventory balance, the accounts receivable turnover rate = net accounts receivable balance / average accounts receivable balance, the total asset turnover rate = sales revenue / average total assets.

[0014] Preferably, the calculated indicator results are compared with the market indicator threshold intervals to determine the current period financial situation operation: The indicator results under the solvency indicator, the profitability indicator and the operating capacity indicator are compared with the indicator threshold intervals in sequence; The calculated results of the current ratio, the quick ratio and the asset-liability ratio are compared with the set indicator threshold values, for example, the current ratio, the value in the period is located in the indicator threshold interval, the current indicator value is normal and the score is m1, while the value in the period is lower than the minimum threshold value, the current indicator value has a lower risk and the score is m2, while the value in the period is greater than the maximum threshold value and less than the exceeding threshold value, the current indicator value has a medium risk and the score is m3, while the value in the period is greater than the exceeding threshold value, the current indicator value has a higher risk and the score is m4, and m1 The current ratio, the quick ratio and the asset-liability ratio are weighted according to the degree of risk impact to obtain the solvency indicator value P1 = x x m u +y x n v +z x l w ; P1 is the solvency indicator value, m u is the corresponding current ratio score after comparison, x is the weight value of the current ratio score after comparison, n v is the corresponding quick ratio score after comparison, y is the weight value of the quick ratio score after comparison, l w is the corresponding asset-liability ratio score after comparison, z is the weight value of the asset-liability ratio score after comparison, and the profitability indicator value and the operating capacity indicator value are P2 and P3; According to historical data, the early warning threshold and the exceeding threshold are determined, and the sum of P1, P2 and P3 is less than the early warning threshold, so that the current financial situation has a low risk and is classified as a third-level early warning, and the sum of P1, P2 and P3 is greater than the early warning threshold and less than the exceeding threshold, and the exceeding threshold is greater than the maximum threshold of the index, so that the current financial situation has a medium risk and is classified as a second-level early warning, and the sum of P1, P2 and P3 is greater than the exceeding threshold, so that the current financial situation has a high risk and is classified as a first-level early warning, and the risk degree of the first-level early warning to the third-level early warning decreases in turn, and when a higher risk score is generated in the process of index evaluation, it is directly classified as a first-level early warning, and a corresponding early warning mapping table is generated according to the results of each index to match the corresponding processing strategy.

[0015] Preferably, the operation of introducing real-time data into the machine learning prediction model in the cloud storage and AI analysis module to make a prediction evaluation on the financial situation is: extracting real-time data into a machine learning prediction model, calculating and evaluating according to the data values, and matching the obtained index results with the early warning mapping table; After determining the risk situation, the abnormal data points are traced back, the combination of words and colors is realized according to different early warning levels, and the user is fed back and informed, and the processing strategy is matched for the user to refer to.

[0016] The application provides an intelligent financial statement analysis system based on cloud AI technology. 1、The intelligent financial statement analysis system based on cloud AI technology classifies the collected data through preprocessing, synchronously extracts the processed data to realize the correlation of each evaluation index, and establishes a machine learning prediction model, introduces real-time data into the machine learning prediction model to make a prediction evaluation on the financial situation, deeply mines and analyzes the financial data by using artificial intelligence technology, reduces the interference of human factors, improves the accuracy and objectivity of the analysis results, comprehensively uses the correlation of financial index calculation and machine learning analysis to form a machine learning prediction model, and thus deeply analyzes the financial information, monitors the financial status of the enterprise in real time, discovers potential financial risks in time, and provides decision support for enterprise managers.

[0017] 2、The intelligent financial statement analysis system based on cloud AI technology realizes data classification through setting an intelligent form, realizes data extraction and confirmation through the correlation verification model and the numerical processing window set in the intelligent form, greatly shortens the time of financial statement analysis through the automatic data collection, processing and analysis process, improves the work efficiency, adopts cloud storage and encryption technology to ensure the safety and reliability of the financial data, prevents data leakage and loss, and cleans the collected original financial data to remove repeated, incorrect and invalid data, unifies the data format and range, and provides a high-quality data basis for subsequent analysis.

[0018] 3、The intelligent financial statement analysis system based on cloud AI technology calculates the solvency index, profitability index and operating capacity index by calculating the historical data in the training set and the same period, compares the calculated index results with the market index threshold interval to determine the financial situation in the current period, and establishes a machine learning prediction model by associating the solvency index, profitability index and operating capacity index, automatically calculates various financial indicators, and compares and analyzes them with the average level of the same industry, uses artificial intelligence technology to deeply mine and analyze financial data, and thus monitors the financial situation in real time and discovers potential financial risks in a timely manner.

[0019] 4、The intelligent financial statement analysis system based on cloud AI technology realizes the comparison of each index result and index threshold interval under the solvency index, profitability index and operating capacity index in turn, extracts real-time data into a machine learning prediction model, calculates and evaluates according to the data values, and matches the obtained index results with the early warning mapping table, the system monitors the financial data in real time, automatically triggers the early warning mechanism when a certain risk index exceeds the set threshold, timely notifies the relevant personnel through various ways, and provides intuitive and timely decision-making basis for enterprise managers, helping enterprises make more scientific and reasonable decisions. BRIEF DESCRIPTION OF DRAWINGS

[0020] Fig. 1 The principle of the intelligent financial statement analysis system of the present application; Fig. 2 The logic flowchart of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] Please refer to Figs. 1-2 The present application provides two technical solutions: Embodiment one, an intelligent financial statement analysis system based on cloud AI technology, comprising: A financial data acquisition module is responsible for collecting various financial information of an enterprise and collecting publicly disclosed financial statement data of external enterprises, and compressing and transmitting the data; The cloud storage and AI analysis module receives and completes decompression of the data, classifies the collected data through preprocessing, stores the collected financial data into the cloud server using distributed storage technology, synchronously extracts the processed data to realize the correlation of various evaluation indexes and establish a machine learning prediction model, and introduces real-time data into the machine learning prediction model to make a prediction and evaluation of the financial situation. The intelligent early warning module realizes different levels of early warning operations based on the results of the prediction and evaluation, and generates corresponding processing strategies for reference by enterprise personnel. The visual display module provides an interactive interface for data query operations, and performs different levels of data access and operation control according to the roles and permissions of users. Rich report templates and visualization components are provided, and users can customize report content and formats according to their own needs, support multiple visualization methods such as column chart, line chart, pie chart, and heat map, display complex financial data in intuitive chart form, facilitate users to quickly understand data meaning and trends, and provide customized control interfaces for enterprise management, real-time display of key financial indicators and business data, and through interactive operation, management can deeply understand data details, timely discover problems and make decisions.

[0023] The collected data is classified through preprocessing, the processed data is synchronously extracted to realize the correlation of various evaluation indexes and establish a machine learning prediction model, real-time data is introduced into the machine learning prediction model to make a prediction and evaluation of the financial situation, artificial intelligence technology is used to deeply mine and analyze financial data, human factors are reduced, the accuracy and objectivity of the analysis results are improved, and the machine learning prediction model is formed by comprehensively using financial indicator calculation and machine learning analysis, thereby deeply analyzing financial information, real-time monitoring the financial status of enterprises, timely discovering potential financial risks, and providing decision support for enterprise managers.

[0024] In the embodiment of the application, the operation of classifying the collected data through preprocessing in the cloud storage and AI analysis module is as follows: An intelligent table is set, the intelligent table is composed of column titles and row titles, the content of the column titles is the item category, the content of the row titles is the timestamp, and the intelligent table realizes extraction and confirmation of the collected data through the set correlation verification model and the numerical processing window; The numerical processing window is used to clean, denoise and normalize the collected data, the data cleaning operation is used to process missing values, abnormal values and repeated values in the data, the data denoising operation is used to eliminate random errors or irrelevant signals in the data, and the data normalization operation is used to scale the data to a unified range; The processed data is classified in the intelligent table and used with the index module.

[0025] For data denoising and data normalization operation, existing data processing technology is adopted to ensure the accuracy of data and reduce resource consumption.

[0026] In the embodiment of the application, the operation of extracting and confirming the collected data in the intelligent table through the set correlation verification model and the numerical processing window is as follows: The numerical processing window moves a plurality of correlation windows on the intelligent table to perform traversal operation, and the traversal trajectory moves along the column title direction from the first row title to the last column title with content in the current direction, and then moves to the next row title for traversal. After the plurality of correlation windows extract data, the correlation verification model is introduced based on the extracted content to trace the data credentials, and the abnormal data is analyzed and determined to be fed back to the numerical processing window for processing operation.

[0027] In the embodiment of the application, the plurality of correlation windows move on the intelligent table to perform traversal operation as follows: The plurality of correlation windows include a row title window (a1, b1), a column title window (a2, b2) and a numerical result window (a2, b1), the row title window (a1, b1) is only used to identify the row title content in the table, the column title window (a2, b2) is only used to identify the column title content in the table, and the numerical result window (a2, b1) is only used to identify the content in the table corresponding to the column title and the row title. When the row title window (a1, b1) is stationary, the column title window (a2, b2) and the numerical result window (a2, b1) move at the same frequency, and when the column title window (a2, b2) returns to the initial column title table, the numerical result window (a2, b1) moves at the same frequency while also moving with the row title window (a1, b1) at the same frequency. And the vertical width b1 of the row title window (a1, b1) and the numerical result window (a2, b1) is expanded synchronously, and the horizontal length a2 of the column title window (a2, b2) and the numerical result window (a2, b1) is expanded synchronously.

[0028] In the embodiment of the application, the operation of introducing the correlation verification model based on the extracted content to trace the data credentials is as follows: The extracted item category result corresponding to the timestamp is matched with the corresponding stored data in the cloud server. If the content extracted by the numerical result window is blank, it is identified whether there is a corresponding timestamp under the corresponding item category, and if no corresponding timestamp is generated, the current numerical result is blank, otherwise the current data column is abnormal, the voucher seal and transaction voucher under the corresponding item category and timestamp are verified, and when the voucher seal and transaction voucher are correct, the numerical result extraction is filled into the data column where the current numerical result window is located, and the data column is expanded according to the specific numerical result condition; If the content extracted by the numerical result window is a normal value, the voucher seal and transaction voucher under the corresponding item category and timestamp are identified and verified, and when the voucher seal and transaction voucher are correct, the numerical result of the trace source is compared with the extracted normal value, and when they are consistent, there is no abnormality, otherwise the current data column is abnormal, the numerical result is extracted and replaces the data in the data column where the current numerical result window is located, and the data column is expanded according to the specific numerical result condition; If the content extracted by the numerical result window is an error code, the current data column is abnormal, the voucher seal and transaction voucher under the corresponding item category and timestamp are identified and verified, and when the voucher seal and transaction voucher are correct, the numerical result of the trace source is extracted and replaces the error code in the data column where the current numerical result window is located, and the data column is expanded according to the specific numerical result condition; When the voucher seal and transaction voucher are abnormal, data checking needs to be fed back to the corresponding income and expenditure party, and data extraction is realized again.

[0029] Among them, the intelligent table is set to realize the classification of data, and the associated verification model and numerical processing window in the intelligent table are set to realize the extraction and confirmation of the collected data. Through the automatic data collection, processing and analysis process, the time of financial statement analysis is greatly shortened, the work efficiency is improved, and the cloud storage and encryption technology is adopted to ensure the safety and reliability of the financial data, prevent data leakage and loss, and clean the collected original financial data, remove repeated, incorrect and invalid data, unify the data format and range, and provide high-quality data basis for subsequent analysis.

[0030] In the embodiment of the application, the operation of expanding the data column according to the specific numerical result condition is: The numerical result is extracted, and the expansion distance of the data column and the numerical result window (a2, b1) is determined according to the character number of the numerical result; Set the number of characters c, and the single horizontal distance of the character f, the single vertical distance of the character g, and the spacing between characters d, and the maximum horizontal length of the data column is set as e, then when [c×f+ (c+1)×d]≤a2, then the current data column does not need to be expanded, when a2< [c×f+ (c+1)×d]≤e, then the current data column only performs horizontal expansion, and the horizontal expansion length is [c× (f+d)-a2], when [c×f+ (c+1)×d]>e, then the current data column performs synchronous expansion in horizontal and vertical directions, and the horizontal expansion length is e, and the vertical expansion width is [[c× (f+d) / e]-1]× (g+d)+ (g+2d). Then, according to the expansion of each item, the expansion operation of the data column is realized, and after the horizontal distance of the current data column is expanded, the horizontal length of all vertical data columns corresponding to the current data column and the horizontal length of the numerical result window (a2, b1) are synchronously expanded, and after the vertical distance of the current data column is expanded, the vertical width of all horizontal data columns corresponding to the current data column and the vertical width of the numerical result window (a2, b1) are synchronously expanded.

[0031] In the embodiment of the application, the operation of correlating the evaluation indexes of the extracted and processed data in the cloud storage and AI analysis module and establishing a machine learning prediction model is: The processed data is extracted, and the data is uniformly divided into a training set and a test set in a ratio of 8:2; The historical data in the training set within the same period is calculated to obtain the solvency index, the profitability index and the operating capacity index, and the calculated index results are compared with the market index threshold interval to determine the financial situation in the current period, and the solvency index, the profitability index and the operating capacity index are correlated to form a machine learning prediction model; The test set is introduced into the machine learning prediction model for testing and optimization.

[0032] The historical data in the training set within the same period is calculated to obtain the solvency index, the profitability index and the operating capacity index, and the calculated index results are compared with the market index threshold interval to determine the financial situation in the current period, and the solvency index, the profitability index and the operating capacity index are correlated to form a machine learning prediction model, which automatically calculates various financial indexes and compares them with the average level of the same industry, and uses artificial intelligence technology to deeply mine and analyze the financial data, so as to monitor the financial situation in real time and timely discover potential financial risks.

[0033] In the embodiment of the application, the calculation method of the solvency index, the profitability index and the operating capacity index is: Among the solvency indicators, the current ratio, quick ratio and asset-liability ratio are calculated and evaluated. The current ratio = total current assets / total current liabilities, the quick ratio = (current assets - inventory) / total current liabilities, and the asset-liability ratio = total liabilities / total assets x 100%. Evaluation criteria: The current ratio is generally considered reasonable in the range of 150%-200% in China, and the quick ratio is generally considered reasonable in the range of 90%-100% in China. This indicator excludes inventory and more accurately reflects the immediate solvency. If the quick ratio is consistently below 90%, there may be a short-term liquidity crisis. If it is higher than 120%, the opportunity cost may increase due to idle funds. The asset-liability ratio is generally considered reasonable in the range of 50%-60% in China, reflecting the proportion of debt in total assets. If the asset-liability ratio exceeds 70%, there may be pressure to repay debts. If it is lower than 40%, the financial leverage may not be fully utilized to improve earnings.

[0034] Among the profitability indicators, the net profit margin, return on equity and cost of profit margin are calculated and evaluated. The net profit margin = net profit / sales revenue x 100%, the return on equity = net profit / average net assets x 100%, and the cost of profit margin = total profit / total cost of expenses x 100%. Evaluation criteria: The net profit margin reflects the ability of the enterprise to convert each unit of income into net profit. If the net profit margin consistently exceeds the industry average (such as 5%-10% for manufacturing), it indicates that the enterprise has strong profitability. If it is lower than 3%, it needs to be concerned about cost control or product pricing issues. The return on equity (ROE) indicator reflects the return on shareholder capital. If ROE consistently exceeds 15%, it indicates that the enterprise has high capital utilization efficiency. If it is lower than 8%, it may need to optimize asset allocation or improve operational efficiency. The cost of profit margin reflects the profit created by each unit of cost and expense. If the ratio is higher than 30%, it indicates that the enterprise has effective cost control. If it is lower than 15%, it needs to optimize the cost structure or improve operational efficiency.

[0035] Among the operating capacity indicators, the inventory turnover rate, accounts receivable turnover rate and total asset turnover rate are calculated and evaluated. The inventory turnover rate = sales cost / average inventory balance, the accounts receivable turnover rate = net credit sales / average accounts receivable balance, and the total asset turnover rate = sales revenue / average total assets.

[0036] Evaluation criteria: The inventory turnover rate reflects the speed of inventory turnover. If the turnover rate is higher than the industry average (such as the retail industry is usually 6-8 times / year), it indicates that the company's inventory management is efficient. If it is lower than 3 times / year, it may face the risk of inventory backlog. The accounts receivable turnover rate reflects the speed of accounts receivable collection. If the turnover rate is higher than 10 times / year, it indicates that the company's collection efficiency is high; if it is lower than 5 times / year, it is necessary to pay attention to customer credit policies or account aging management. The total asset turnover rate reflects the operating efficiency of the company's total assets. If the turnover rate is higher than 1 time / year, it indicates that the company's assets are fully utilized. If it is lower than 0.5 times / year, it may be necessary to optimize the asset structure or improve operational efficiency.

[0037] In the embodiment of the present invention, the calculated indicator results are compared with the threshold ranges of various market indicators to determine the financial situation in the current period. The operation is as follows: Compare the results of various indicators under the debt-paying ability indicator, profitability indicator and operating ability indicator with the indicator threshold range in sequence; Compare the calculation results of the current ratio, quick ratio, and debt-to-asset ratio with the set indicator thresholds. Take the current ratio as an example. If the value of the current ratio within the period is within the indicator threshold range, the current indicator value is normal and the score is m1. If the value of the current ratio within the period is lower than the minimum threshold, the current indicator value has a low risk and is scored as m2. If the value of the current ratio within the period is greater than the maximum threshold and less than the excess threshold, and the excess threshold is greater than the maximum threshold, the current indicator value has a medium risk and is scored as m3. If the value of the current ratio within the period is greater than the excess threshold, the current indicator value has a high risk and is scored as m4, and m1<m2<m3<m4; According to the degree of risk impact, the current ratio, quick ratio and debt-to-asset ratio are weighted to obtain the debt-paying ability index value: P1=x×m u +y×n v +z×l w ; P1 is the solvency index value, m u is the score after the current ratio comparison, x is the weight value of the score after the current ratio comparison, n v is the score after the quick ratio comparison, y is the weight value of the score after the quick ratio comparison, l w is the score after the asset-liability ratio comparison, z is the weight value of the score after the asset-liability ratio comparison, and similarly, the profitability index value and operating capacity index value are P2 and P3; According to historical data, the early warning threshold and the exceeding threshold are determined, and the sum of P1, P2 and P3 is less than the early warning threshold, so that the current financial situation has a low risk and is classified as a third-level early warning, and the sum of P1, P2 and P3 is greater than the early warning threshold and less than the exceeding threshold, and the exceeding threshold is greater than the maximum threshold of the index, so that the current financial situation has a medium risk and is classified as a second-level early warning, and the sum of P1, P2 and P3 is greater than the exceeding threshold, so that the current financial situation has a high risk and is classified as a first-level early warning, and the risk degree of the first-level early warning to the third-level early warning decreases in turn, and when a higher risk score is generated in the process of index evaluation, it is directly classified as a first-level early warning, and a corresponding early warning mapping table is generated according to the results of each index to match the corresponding processing strategy.

[0038] In the embodiment of the application, the operation of introducing real-time data into the machine learning prediction model in the cloud storage and AI analysis module to make a prediction and evaluation of the financial situation is as follows: The real-time data is extracted into the machine learning prediction model, the data values are calculated and evaluated, and the obtained index results are matched with the early warning mapping table; After determining the risk situation, the abnormal data points are traced back, the combination of text and color corresponding to different early warning levels is realized, and the user is fed back and informed, and the processing strategy is matched for the user to refer to.

[0039] Among them, the index results under the solvency index, profitability index and operating capacity index are compared with the index threshold interval in turn, the real-time data is extracted into the machine learning prediction model, the data values are calculated and evaluated, and the obtained index results are matched with the early warning mapping table, the system monitors the financial data in real time, when a certain risk index exceeds the set threshold, the early warning mechanism is automatically triggered, the relevant personnel are timely informed through various ways, and an intuitive and timely decision basis is provided for enterprise managers, helping enterprises to make more scientific and reasonable decisions.

[0040] Embodiment two, compared with the embodiment, the difference lies in that the financial test data of multiple enterprises is analyzed by the traditional financial statement analysis system and the financial statement analysis system of the application, and the cycle time of financial analysis, the accuracy rate of financial risk prediction and the occupation situation of data resource analysis are recorded and compared, and the specific results are shown in Table 1: Table 1 record result table

[0041] The results show that the intelligent financial statement analysis system based on the cloud AI technology is significantly better than the existing financial statement analysis system in the embodiment of various parameter data, and proves that the intelligent financial statement analysis system has shorter financial analysis cycle time, higher financial risk prediction accuracy and lower data resource analysis occupancy rate in actual application, so that the financial statement analysis operation can be effectively and quickly realized.

[0042] Meanwhile, the contents not described in detail in the specification all belong to the prior art known by the person skilled in the art.

[0043] It should be noted that, in this paper, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0044] Although the embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent financial statement analysis system based on cloud AI technology, characterized by: include: The financial data collection module is responsible for collecting various financial information of the enterprise, as well as collecting the financial statement data disclosed by enterprises outside the market, and compressing and transmitting the data; The cloud storage and AI analysis module receives and decompresses the data, then pre-processes and classifies the collected data. It then uses distributed storage technology to store the collected financial data on cloud servers. It then simultaneously extracts the processed data to correlate various evaluation indicators and establish a machine learning prediction model. This model then uses real-time data to make predictions and assessments of financial performance. The intelligent early warning module implements different levels of early warning operations based on the results of prediction and evaluation, and generates corresponding processing strategies for reference by enterprise personnel; The visualization display module provides an interactive interface for data query operations, and performs different levels of data access and operation control based on user roles and permissions.

2. The intelligent financial statement analysis system based on cloud AI technology according to claim 1, characterized in that: The operations for pre-processing and classifying the collected data in the cloud storage and AI analysis module are as follows: Set up a smart table. The smart table consists of column headers and row headers. The column headers contain the payment category, and the row headers contain the timestamp. The smart table extracts and confirms the collected data through the set association verification model and numerical processing window. The collected data is cleaned, denoised and normalized using the numerical processing window. The data cleaning operation is used to deal with missing values, outliers and duplicate values ​​in the data. The data denoising operation is used to eliminate random errors or irrelevant signals in the data. The data normalization operation is used to scale the data to a uniform range. The processed data is classified into smart tables and used in conjunction with the index module.

3. The intelligent financial statement analysis system based on cloud AI technology according to claim 2, characterized in that: The operations of extracting and confirming collected data by setting the associated verification model and the numerical processing window in the smart form are as follows: The numerical processing window is located on the smart table by setting multiple associated windows to move for traversal operation. The traversal track starts from the first row header of the table and moves along the column header direction until the last column header with content in the current direction is traversed. Then the associated window moves to the next row header for traversal. After multiple associated windows realize data extraction, the associated verification model is introduced to trace the data voucher based on the extracted content, and the data anomalies are analyzed and determined and fed back to the numerical processing window for abnormal situation processing operations.

4. The intelligent financial statement analysis system based on cloud AI technology according to claim 3, characterized in that: The traversal operation of the multiple associated windows on the smart table is: Multiple associated windows include a row title window (a1, b1), a column title window (a2, b2), and a numerical result window (a2, b1). The row title window (a1, b1) is only used to identify the row title content in the table, the column title window (a2, b2) is only used to identify the column title content in the table, and the numerical result window (a2, b1) is only used to identify the content in the corresponding column title and row title extended cross table; When the row title window (a1, b1) remains stationary, the column title window (a2, b2) and the numerical result window (a2, b1) keep moving at the same frequency. When the column title window (a2, b2) returns to the initial column title table, the numerical result window (a2, b1) keeps moving at the same frequency and also moves at the same frequency as the row title window (a1, b1). And the vertical width b1 of the row title window (a1, b1) and the numerical result window (a2, b1) are expanded synchronously, and the horizontal length a2 of the column title window (a2, b2) and the numerical result window (a2, b1) are expanded synchronously.

5. The intelligent financial statement analysis system based on cloud AI technology according to claim 4, characterized in that: The operation of introducing the associated verification model traceability data certificate based on the extracted content is: Match the extracted payment category results at the corresponding timestamp with the corresponding stored data in the cloud server; If the extracted content in the numerical result window is blank, identify whether there is a corresponding timestamp under the corresponding payment category in the traceability process. If no corresponding timestamp is generated, the current numerical result is blank. Otherwise, the current data column is abnormal. Verify the voucher seal and transaction voucher under the corresponding payment category and timestamp. If the voucher seal and transaction voucher are correct, extract the numerical result and fill it into the data column where the current numerical result window is located. Expand the data column according to the specific numerical result situation. If the content extracted from the numerical result window is a normal value, the voucher seal and transaction voucher under the corresponding payment type and timestamp are identified and verified. If the voucher seal and transaction voucher are correct, the numerical result of the traceability is compared with the extracted normal value. If the two are consistent, there is no abnormality. Otherwise, if the current data column is abnormal, the numerical result is extracted and replaced with the data in the data column where the current numerical result window is located, and the data column is expanded according to the specific numerical result situation. If the content extracted from the numerical result window is an error code and the current data column is abnormal, the voucher seal and transaction voucher under the corresponding payment type and timestamp will be identified and verified for tracing back. If the voucher seal and transaction voucher are correct, the numerical result of tracing back will be extracted and replaced with the error code in the data column where the current numerical result window is located. The data column will be expanded according to the specific numerical result situation. When there are abnormalities in the voucher seal and transaction voucher, they need to be fed back to the corresponding payer and payee for data verification and then data extraction.

6. The intelligent financial statement analysis system based on cloud AI technology according to claim 5, characterized in that: The operation of expanding the data column according to the specific numerical result is: Extract the numerical result and determine the expansion distance between the data column and the numerical result window (a2, b1) according to the number of characters in the numerical result; Set the number of characters to c, the single horizontal distance of a character to f, the single vertical distance of a character to g, the spacing between characters to d, and the maximum horizontal length of the data column to e. Then, when [c×f+(c+1)×d]≤a2, the current data column does not need to be expanded. When a2<[c×f+(c+1)×d]≤e, the current data column is only expanded horizontally, and the horizontal expansion length is [c×(f+d)-a2]. When [c×f+(c+1)×d]>e, the current data column is expanded horizontally and vertically simultaneously, and the horizontal expansion length is e, and the vertical expansion width is [[c×(f+d) / e]-1]×(g+d)+(g+2d). Then, the expansion operation of the data bar is implemented according to the conditions of each expansion, and after the horizontal distance of the current data bar is expanded, the horizontal lengths of all corresponding vertical data bars under the current data bar and the horizontal lengths of the numerical result window (a2, b1) are expanded synchronously. After the vertical distance of the current data bar is expanded, the vertical widths of all corresponding horizontal data bars under the current data bar and the vertical width of the numerical result window (a2, b1) are expanded synchronously.

7. The intelligent financial statement analysis system based on cloud AI technology according to claim 1, characterized in that: The operations of extracting and processing data from the cloud storage and AI analysis module to associate various evaluation indicators and establish a machine learning prediction model are as follows: Extract the processed data and evenly divide the data into training set and test set in a ratio of 8:2; The historical data within the training set and the corresponding period are used to calculate the solvency index, profitability index and operating capacity index. The calculated index results are compared with the threshold range of various market indicators to determine the financial situation within the current period. The solvency index, profitability index and operating capacity index are linked to form a machine learning prediction model. The test set is introduced into the machine learning prediction model for verification and optimization.

8. The intelligent financial statement analysis system based on cloud AI technology according to claim 7, characterized in that: The calculation method of the debt-paying ability index, profitability index and operating ability index is as follows: The solvency indicator is calculated and evaluated by calculating the current ratio, quick ratio and debt-to-asset ratio: current ratio = total current assets / total current liabilities; quick ratio = (current assets - inventory) / total current liabilities; debt-to-asset ratio = total liabilities / total assets × 100%; Profitability indicators are calculated and evaluated by calculating net profit margin, return on net assets and cost-to-profit ratio: net profit margin = net profit / sales revenue × 100%, return on net assets = net profit / average net assets × 100%, cost-to-profit ratio = total profit / total cost-to-profit ratio × 100%. The operating capacity indicators are calculated and evaluated by inventory turnover rate, accounts receivable turnover rate and total asset turnover rate. Inventory turnover rate = sales cost / average inventory balance, accounts receivable turnover rate = net credit sales / average accounts receivable balance, and total asset turnover rate = sales revenue / average total assets.

9. The intelligent financial statement analysis system based on cloud AI technology according to claim 7, characterized in that: The calculated indicator results are compared with the threshold ranges of various market indicators to determine the financial situation in the current cycle. The operation is as follows: Compare the results of various indicators under the debt-paying ability indicator, profitability indicator and operating ability indicator with the indicator threshold range in sequence; Compare the calculation results of the current ratio, quick ratio, and debt-to-asset ratio with the set indicator thresholds. Take the current ratio as an example. If the value of the current ratio within the period is within the indicator threshold range, the current indicator value is normal and the score is m1. If the value of the current ratio within the period is lower than the minimum threshold, the current indicator value has a low risk and is scored as m2. If the value of the current ratio within the period is greater than the maximum threshold and less than the excess threshold, and the excess threshold is greater than the maximum threshold, the current indicator value has a medium risk and is scored as m3. If the value of the current ratio within the period is greater than the excess threshold, the current indicator value has a high risk and is scored as m4, and m1<m2<m3<m4; According to the degree of risk impact, the current ratio, quick ratio and debt-to-asset ratio are weighted to obtain the debt-paying ability index value: P1=x×m u +y×n v +z×l w ; P1 is the solvency index value, m u is the score after the current ratio comparison, x is the weight value of the score after the current ratio comparison, n v is the score after the quick ratio comparison, y is the weight value of the score after the quick ratio comparison, l w is the score after the asset-liability ratio comparison, z is the weight value of the score after the asset-liability ratio comparison, and similarly, the profitability index value and operating capacity index value are P2 and P3; The warning threshold and exceeding threshold are determined based on historical data. If the sum of P1, P2 and P3 is less than the warning threshold, the current financial situation has a low risk and is classified as a level three warning. If the sum of P1, P2 and P3 is greater than the warning threshold and less than the exceeding threshold, and the exceeding threshold is greater than the maximum threshold of the indicator, the current financial situation has a medium risk and is classified as a level two warning. If the sum of P1, P2 and P3 is greater than the exceeding threshold, the current financial situation has a high risk and is classified as a level one warning. The risk levels from level one warning to level three warning decrease in sequence. If a high risk score is generated during the indicator evaluation process, it is directly classified as a level one warning. In combination with the results of each indicator, a corresponding warning mapping table is generated to match the corresponding processing strategy.

10. The intelligent financial statement analysis system based on cloud AI technology according to claim 7, characterized in that: The operations of introducing real-time data from the cloud storage and AI analysis module into the machine learning prediction model to make a forecast assessment of the financial situation are as follows: Extract real-time data and introduce it into the machine learning prediction model, perform calculations and evaluations based on the data values, and match the obtained indicator results with the early warning mapping table; After determining the risk situation, the abnormal data points are traced back to the corresponding warning levels, and the text and color are combined to inform the user, and the processing strategy is matched for the user's reference.

Images