Enterprise financial anomaly detection method and device

Abstract

The invention discloses an enterprise financial anomaly detection method and device, and relates to the technical field of anomaly detection. Comprising the following steps: obtaining historical financial data of a historical to-be-detected enterprise, and based on the historical financial data, calculating financial anomaly characterization parameters of the enterprise, such as a financial stability index, a profit fluctuation index and a fluidity pressure index; and in combination with the financial anomaly judgment coefficient, correcting and generating a comprehensive financial anomaly judgment coefficient, and comparing the comprehensive financial anomaly judgment coefficient with an anomaly threshold to obtain a financial anomaly detection result. The financial abnormity of the enterprise can be efficiently identified, and the financial health condition of the enterprise can be more accurately evaluated, so that the operation decision error caused by the financial abnormity can be effectively avoided.

Description

Technical Field

[0001] The present invention relates to the field of anomaly detection technology, and in particular to a method and device for detecting anomalies in corporate finance. Background Art

[0002] With the continuous development of the global economy and the changing market environment, enterprises are facing increasingly complex financial conditions and external risk factors. As an important risk management tool, financial anomaly detection can help enterprises detect abnormal fluctuations in financial conditions in a timely manner, so as to take corresponding control measures and reduce financial risks. However, traditional financial anomaly detection methods are often based only on the analysis of the company's own financial data, ignoring the impact of the market environment, competitive landscape and other external factors on the company's financial situation. Therefore, how to accurately identify the financial anomalies of an enterprise, taking into account both the internal business activities of the enterprise and the changes in the external market environment, is a key issue that needs to be solved in the current field of financial analysis.

[0003] Traditional financial anomaly detection methods generally rely on historical data in corporate financial statements, such as balance sheets, cash flow statements, and income statements, and use financial ratios, outlier detection, trend analysis, and other means to identify anomalies. The advantages of these methods are their simplicity and operability, but they also have many limitations. First, the financial data of an enterprise may be affected by various external factors, and traditional methods cannot effectively consider the changes in these factors. Secondly, relying solely on financial data for anomaly detection can easily lead to a one-sided understanding of the company's operating conditions and may not accurately reveal the root causes of the company's financial problems. For example, some companies may show good financial conditions in the short term, but due to market competition pressure, unfavorable external environment and other factors, they are in financial risk for a long time.

[0004] In recent years, with the development of big data and artificial intelligence technology, more and more companies have begun to try to incorporate factors such as external market environment and industry competition into the scope of financial anomaly detection, and use more dimensions for comprehensive analysis. This method can not only more comprehensively evaluate the financial health of the company, but also effectively predict possible financial risks. However, the complexity of such methods has also increased, and it is necessary to maintain efficiency and accuracy when processing large amounts of information. Therefore, how to accurately judge the financial anomalies of a company under the condition of huge data volume and complex variables has become a hot issue in current research.

[0005] In the prior art, the publication number CN111783829A discloses a financial anomaly detection method and device based on multi-label learning. The method first generates a feature vector sample of each enterprise based on the financial information of the existing enterprise, and then performs sample balance according to the preset sampling parameters to obtain a training sample set; then, the training sample set is labeled according to multiple labels obtained from historical accountability data, and a financial anomaly detection model is constructed based on a multi-label learning algorithm; finally, the financial information of the enterprise to be detected is obtained, and a sample input vector is constructed and input into the financial anomaly detection model to obtain the detection result. However, in this scheme, if the sampling parameters are not selected appropriately, it may cause the model to overfit or underfit on certain labels, affecting the accuracy and robustness of financial anomaly detection. In particular, in financial data, there are usually fewer abnormal samples, and oversampling may lead to an increase in false positives, while undersampling may lead to insufficient detection capabilities of abnormal samples. At the same time, multi-label learning is too complicated, which will increase the computational cost and the difficulty of model training. In addition, the predictions of multiple labels may interfere with each other, affecting the accuracy of the model, especially when the financial data has a complex nonlinear relationship. Therefore, detecting financial anomalies with only one multi-label learning algorithm reduces the accuracy and effectiveness of the detection device.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The purpose of the present invention is to provide a coal mine supervision method and system based on video images to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for detecting anomalies in corporate finance, comprising the following specific steps:

[0010] Obtaining a number of financial data of the enterprise to be analyzed in a number of consecutive historical months before the current month, establishing a first prediction model based on the obtained financial data, taking the financial data of the enterprise to be analyzed in a number of consecutive historical months before the previous month as input, and using the financial data of the previous month as a label, to train the first prediction model;

[0011] Inputting several financial data items of a plurality of consecutive historical months before the current month into the first prediction model that has been trained, and outputting several financial data items of the current month, the several financial data items including cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue;

[0012] Based on each item of financial data obtained in the current month, calculate and generate financial characteristic data, and calculate the enterprise's financial abnormality characterization parameters through the obtained financial characteristic data, wherein the financial characteristic data includes cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin;

[0013] The enterprise financial anomaly judgment coefficient is calculated based on the enterprise financial anomaly characterization parameters, which include the financial stability index, the profit volatility index and the liquidity pressure index. The enterprise financial anomaly judgment coefficient is corrected based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient. The comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold, and different financial anomaly detection results are generated according to different comparison results. Further, based on each item of financial data obtained in the current month, the financial characteristic data is calculated and generated, wherein the specific formula for calculating the financial characteristic data is:

[0014]

[0015]

[0016]

[0017]

[0018] In the formula, is the cash flow ratio, represents cash flow from operating activities, For current liabilities, is the debt-to-asset ratio, is the total debt, is the total assets, is the working capital ratio, For current assets, is the sales profit margin, represents the net profit from sales, Represents sales revenue.

[0019] Furthermore, the obtained financial characteristic data is used to calculate the enterprise financial abnormality characterization parameters, which include a financial stability index, a profit volatility index, and a liquidity pressure index, wherein the formula for calculating the financial stability index is:

[0020]

[0021] In the formula, is the financial stability index;

[0022] The formula for calculating the earnings volatility index is:

[0023]

[0024] In the formula, is the earnings volatility index;

[0025] The formula for calculating the liquidity pressure index is:

[0026]

[0027] In the formula, represents the liquidity pressure index, and They are accounts receivable turnover days and accounts payable turnover days respectively.

[0028] Furthermore, the enterprise financial anomaly judgment coefficient is calculated based on the enterprise financial anomaly characterization parameters, wherein the formula for calculating the enterprise financial anomaly judgment coefficient is:

[0029]

[0030] In the formula, is the enterprise financial abnormality judgment coefficient, , and are the weight coefficients of the financial stability index, profit volatility index and liquidity pressure index, respectively. and , and Both are greater than 0.

[0031] Furthermore, a prediction model is established based on the LSTM long short-term memory network model, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is:

[0032]

[0033] In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer;

[0034] At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons;

[0035] The number of network layers is set to a four-layer network structure, the number of iterations is set to 300, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 300, the batch size is set to 256, and the number of hidden layer neurons is 32.

[0036] Furthermore, the market environment change parameters in the current month are input into the trained prediction model to obtain the estimated values ​​of the sales volume and sales cost of the enterprise to be analyzed in the current month. Based on the estimated values ​​of the sales volume and sales cost, the enterprise financial anomaly judgment coefficient is corrected to obtain the comprehensive financial anomaly judgment coefficient, wherein the formula for calculating the comprehensive financial anomaly judgment coefficient is:

[0037]

[0038] In the formula, is the comprehensive financial abnormality judgment coefficient, and They are the estimated sales volume and target sales volume of the company to be analyzed in the current month. and They are respectively the estimated value and target value of sales cost of the enterprise to be analyzed in the current month;

[0039] The comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold, and different financial anomaly detection results are generated according to different comparison structures. The logic for generating different financial anomaly detection results is as follows:

[0040] when When the company is analyzed, it is determined that there is no abnormality in the financial situation of the company to be analyzed;

[0041] when If the financial situation of the enterprise to be analyzed is judged to be abnormal, it should be handled in a timely manner;

[0042] In the formula, is the abnormal threshold.

[0043] The present invention also provides a device for detecting anomalies in corporate finances, which is used to execute the above-mentioned method for detecting anomalies in corporate finances, and comprises:

[0044] The prediction model training module is used to obtain a number of financial data of the enterprise to be analyzed in a number of consecutive historical months before the current month, establish a first prediction model based on the obtained financial data, take the financial data of the enterprise to be analyzed in a number of consecutive historical months before the previous month as input, and use the financial data of the previous month as a label to train the first prediction model;

[0045] A financial data collection module is used to input a number of financial data from a number of consecutive historical months before the current month into the trained first prediction model, and output a number of financial data for the current month, wherein the number of financial data includes cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue;

[0046] The abnormal characterization parameter analysis module is used to calculate and generate financial characteristic data based on each item of financial data obtained in the current month, and calculate the abnormal financial characterization parameters of the enterprise through the obtained financial characteristic data, wherein the financial characteristic data include cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin;

[0047] The financial anomaly detection module is used to calculate the enterprise financial anomaly judgment coefficient based on the enterprise financial anomaly characterization parameters, wherein the enterprise financial anomaly characterization parameters include the financial stability index, the profit volatility index and the liquidity pressure index. The enterprise financial anomaly judgment coefficient is corrected based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient, and the comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold. Different financial anomaly detection results are generated according to different comparison results.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] First, by combining market environment change parameters with corporate financial data, we can comprehensively analyze the impact of external market factors on corporate financial conditions. Changes in external parameters such as the number of competitors, GDP growth rate, unemployment rate, and interest rates often directly affect the company's sales volume and sales costs. By incorporating market environment change parameters into the analysis framework, we can introduce changes in the external environment into financial analysis, making the test results more accurate and comprehensive. In addition, by comprehensively calculating the company's financial anomaly judgment coefficient and combining it with the revised sales volume and sales cost estimates, we can make a more detailed judgment on the company's financial anomalies. This makes the final financial anomaly judgment result more scientific and reasonable. Based on this judgment coefficient, companies can make a more accurate assessment of their financial health, thereby effectively avoiding operational decision-making errors caused by financial anomalies. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0051] Figure 2 It is a schematic diagram of the overall structure of the device of the present invention. DETAILED DESCRIPTION

[0052] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0053] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0054] Example:

[0055] See also Figure 1 , the present invention provides a technical solution:

[0056] A method for detecting anomalies in corporate finance, comprising the following specific steps:

[0057] Step 1: Obtain several financial data of the company to be analyzed for multiple consecutive historical months before the current month, and establish a first prediction model based on the obtained financial data. Take the several financial data of the company to be analyzed for multiple consecutive historical months before the previous month as input, and use the financial data of the previous month as labels to train the first prediction model.

[0058] Step 2: Input several financial data of multiple consecutive historical months before the current month into the trained first prediction model, and output several financial data of the current month, wherein the several financial data include cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue.

[0059] A prediction model is established based on the LSTM long short-term memory network model, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is:

[0060]

[0061] In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer;

[0062] At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons;

[0063] The number of network layers is set to a four-layer network structure, the number of iterations is set to 300, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 300, the batch size is set to 256, and the number of hidden layer neurons is 32.

[0064] In the context of building a forecasting model based on LSTM (Long Short-Term Memory Network), the LSTM model has many advantages in predicting sales volume and sales cost using market environment change parameters. LSTM is a deep learning model specifically designed to process and forecast time series data (such as market environment changes and sales volume, sales cost, etc.).

[0065] The main advantage of LSTM networks is that they can process and learn time series data over long time spans. Market environment changes and sales volume and sales costs usually change over time and are affected by historical data. LSTM is specially designed to solve the "gradient vanishing" problem encountered by ordinary RNN (recurrent neural network) in long sequences, so it can effectively capture long-term dependencies. LSTM can maintain information over long time spans through its special gating mechanism (input gate, forget gate and output gate), which is crucial for modeling the impact of market environment changes on sales volume and sales costs. Changes in the market environment are often not instantaneous, but gradually affect sales performance over time. LSTM can effectively capture the long-term impact of these changes on sales volume and costs.

[0066] In actual scenarios, there is usually a nonlinear relationship between the market environment (such as economic changes, consumer demand, competition, etc.) and sales volume and sales costs. Traditional statistical models such as linear regression may not be able to capture this complex nonlinear relationship. LSTM, as a deep learning model, can automatically learn and model these nonlinear relationships.

[0067] Step 3: Based on each financial data obtained in the current month, calculate and generate financial characteristic data, and calculate the company's financial abnormality characterization parameters through the obtained financial characteristic data. The financial characteristic data include cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin.

[0068] Based on each item of financial data obtained for the current month, the financial characteristic data is calculated and generated, wherein the specific formula for calculating the financial characteristic data is as follows:

[0069]

[0070]

[0071]

[0072]

[0073] In the formula, is the cash flow ratio, represents cash flow from operating activities, For current liabilities, is the debt-to-asset ratio, is the total debt, is the total assets, is the working capital ratio, For current assets, is the sales profit margin, represents the net profit from sales, Represents sales revenue.

[0074] The financial characteristic data obtained are used to calculate the enterprise financial abnormality characterization parameters, which include the financial stability index, the profit volatility index and the liquidity pressure index. The formula for calculating the financial stability index is:

[0075]

[0076] In the formula, is the financial stability index;

[0077] The higher the debt-to-asset ratio, the higher the company's debt ratio and the greater the financial leverage. Because financial stability should decrease as the debt ratio increases, that is, the higher the debt ratio, the worse the company's financial stability. In this way, a higher debt-to-equity ratio is transformed into a smaller value, reflecting its negative impact on financial stability. , is to mitigate the impact of a larger debt-to-asset ratio (i.e. a higher debt level) on financial stability and make changes smoother.

[0078] in Reflects the impact of the cash flow ratio. The cash flow ratio (CFR) is usually used to measure the ratio of a company's cash flow to current liabilities, indicating whether the company has enough cash flow to repay its liabilities. Taking the logarithm of the cash flow ratio can make the impact of cash flow nonlinear. If the cash flow ratio is large, the logarithmic function can mitigate the impact of the higher cash flow ratio, making the improvement of financial stability smoother rather than linearly amplified. Especially when the cash flow ratio is large, direct linear amplification may not be in line with the actual situation, because an excessively high cash flow ratio may not significantly improve financial stability in some cases. For example, cash flow may only be temporary, or too much cash may be inefficient.

[0079] The formula for calculating the earnings volatility index is:

[0080]

[0081] In the formula, is the earnings volatility index;

[0082] The sales profit margin reflects the profit margin generated by the company in the sales process. It is an important indicator to measure the profitability of the company. When the sales profit margin is low, the profitability of the company is poor, and when the sales profit margin is high, the profitability of the company is strong. As the sales profit margin increases, the value of PVI gradually decreases, but after reaching a certain level, the decrease will gradually slow down. This reflects that profitability gradually approaches saturation as the sales profit margin increases.

[0083] The working capital ratio reflects the funds required for the daily operations of an enterprise and is used to measure the liquidity and efficiency of capital utilization of an enterprise. The volatility characteristic means the ratio of current assets to liabilities. The larger its value is, the larger the proportion of working capital is and the smaller the profit fluctuation is. The relationship between PVI and working capital ratio is not monotonically linear.

[0084] The formula for calculating the liquidity pressure index is:

[0085]

[0086] In the formula, represents the liquidity pressure index, and They are accounts receivable turnover days and accounts payable turnover days respectively.

[0087] When a company's working capital ratio and cash flow ratio are high, the company's liquidity pressure is small. Conversely, when these two ratios are low, the company may face greater liquidity pressure.

[0088] The combination of working capital ratio and cash flow ratio can better reflect whether a company can support its short-term liabilities and operations through cash flow, and whether the company can maintain liquidity through its working capital in the absence of sufficient cash flow. If the working capital ratio and cash flow ratio are small, the company may face greater liquidity pressure, otherwise it indicates that the company's liquidity is relatively healthy.

[0089] Due to the smoothness of the logarithmic function, This means that even if the ratio changes greatly, The changes will not be drastic, thus avoiding overly sensitive or unstable results. At the same time, the longer the accounts receivable turnover days, the greater the liquidity pressure of the enterprise, and the shorter the accounts payable turnover days, the greater the liquidity pressure of the enterprise. Therefore, and is proportional to and Directly proportional.

[0090] Step 4: Calculate the enterprise financial anomaly judgment coefficient based on the enterprise financial anomaly characterization parameters, where the enterprise financial anomaly characterization parameters include financial stability index, profit volatility index and liquidity pressure index. Correct the enterprise financial anomaly judgment coefficient based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient. Compare the comprehensive financial anomaly judgment coefficient with the anomaly threshold, and generate different financial anomaly detection results based on different comparison results.

[0091] The enterprise financial anomaly judgment coefficient is calculated based on the enterprise financial anomaly characterization parameters, where the formula for calculating the enterprise financial anomaly judgment coefficient is:

[0092]

[0093] In the formula, is the enterprise financial abnormality judgment coefficient, , and are the weight coefficients of the financial stability index, profit volatility index and liquidity pressure index, respectively. and , and Both are greater than 0.

[0094] Because the enterprise financial abnormality judgment coefficient The larger the value, the greater the probability of abnormal financial situation in the enterprise. Indicates financial stability. The larger the value, the more stable the financial situation and the smaller the probability of abnormal financial situation of the enterprise. Therefore, the financial stability index and the enterprise financial abnormality judgment coefficient Inversely proportional, through the fraction characterizes the inverse relationship and is expressed by The financial stability index The increase of the enterprise financial abnormality risk index decreases, and the enterprise financial abnormality judgment coefficient Also decreased.

[0095] Among them, the profit volatility index The larger the ratio, the worse the financial health of the company. Larger earnings volatility will lead to increased uncertainty in financial forecasts, which in turn affects the financial stability of the company. If the working capital ratio is high and earnings volatility is large, it may indicate that the company is facing greater liquidity pressure, which may lead to financial anomalies. Therefore, the earnings volatility index It is directly proportional to the enterprise's financial abnormality judgment coefficient. It indicates a positive correlation with the coefficient of judgment of corporate financial anomaly.

[0096] Liquidity Stress Index The larger the value, the greater the liquidity pressure that the enterprise may face, that is, the greater the probability of abnormal financial situation of the enterprise. Therefore, the liquidity pressure index and the enterprise financial abnormality judgment coefficient Proportional, through Represents a proportional relationship and uses logarithmic form Reduce the impact of liquidity pressure index, making the judgment coefficient of corporate financial abnormality The lifting effect becomes smoother instead of being amplified linearly.

[0097] Among them, the financial stability index has the greatest impact on the judgment of corporate financial anomalies, and the profit volatility index has a slightly greater impact on the judgment of corporate financial anomalies than the liquidity pressure index. Therefore, and , and Both are greater than 0.

[0098] Based on the estimated sales volume and sales cost, the enterprise financial abnormality judgment coefficient is corrected to obtain the comprehensive financial abnormality judgment coefficient, where the formula for calculating the comprehensive financial abnormality judgment coefficient is:

[0099]

[0100] In the formula, is the comprehensive financial abnormality judgment coefficient, and They are the estimated sales volume and target sales volume of the enterprise to be analyzed in the current time period. and They are respectively the estimated value and target value of sales cost of the enterprise to be analyzed in the current time period;

[0101] If the sales volume estimate is higher than the target sales volume, it may mean that the company performs well in the market and its sales performance exceeds expectations. This usually reflects that the company has strong profitability and healthy financial condition. Therefore, the value of the comprehensive financial anomaly judgment coefficient should be reduced, indicating that the risk of financial anomaly is low.

[0102] If the estimated sales cost is lower than the target cost, it may mean that the company has performed well in cost control, has high operating efficiency, and is in a relatively healthy financial situation. Therefore, the value of the comprehensive financial abnormality judgment coefficient should be reduced, indicating that the risk of financial abnormality is low.

[0103] The comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold, and different financial anomaly detection results are generated according to different comparison structures. The logic for generating different financial anomaly detection results is as follows:

[0104] when When the company is analyzed, it is determined that there is no abnormality in the financial situation of the company to be analyzed;

[0105] when If the financial situation of the enterprise to be analyzed is judged to be abnormal, it should be handled in a timely manner;

[0106] In the formula, is the abnormal threshold.

[0107] See also Figure 2 The present invention also provides a device for detecting anomalies in corporate finance, which is used to execute the above-mentioned method for detecting anomalies in corporate finance, and includes:

[0108] The prediction model training module is used to obtain a number of financial data of the enterprise to be analyzed in a number of consecutive historical months before the current month, establish a first prediction model based on the obtained financial data, take the financial data of the enterprise to be analyzed in a number of consecutive historical months before the previous month as input, and use the financial data of the previous month as a label to train the first prediction model;

[0109] A financial data collection module is used to input a number of financial data from a number of consecutive historical months before the current month into the trained first prediction model, and output a number of financial data for the current month, wherein the number of financial data includes cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue;

[0110] The abnormal characterization parameter analysis module is used to calculate and generate financial characteristic data based on each item of financial data obtained in the current month, and calculate the abnormal financial characterization parameters of the enterprise through the obtained financial characteristic data, wherein the financial characteristic data include cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin;

[0111] The financial anomaly detection module is used to calculate the enterprise financial anomaly judgment coefficient based on the enterprise financial anomaly characterization parameters, wherein the enterprise financial anomaly characterization parameters include the financial stability index, the profit volatility index and the liquidity pressure index. The enterprise financial anomaly judgment coefficient is corrected based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient, and the comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold. Different financial anomaly detection results are generated according to different comparison results.

[0112] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0113] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0114] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0115] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for detecting anomalies in corporate finance, characterized in that: The specific steps include: Obtaining a number of financial data of the enterprise to be analyzed in a number of consecutive historical months before the current month, establishing a first prediction model based on the obtained financial data, taking the financial data of the enterprise to be analyzed in a number of consecutive historical months before the previous month as input, and using the financial data of the previous month as a label, to train the first prediction model; Inputting several financial data items of a plurality of consecutive historical months before the current month into the first prediction model that has been trained, and outputting several financial data items of the current month, the several financial data items including cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue; Based on each item of financial data obtained in the current month, calculate and generate financial characteristic data, and calculate the enterprise's financial abnormality characterization parameters through the obtained financial characteristic data, wherein the financial characteristic data includes cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin; The enterprise financial anomaly judgment coefficient is calculated based on the enterprise financial anomaly characterization parameters, which include a financial stability index, a profit volatility index and a liquidity pressure index. The enterprise financial anomaly judgment coefficient is corrected based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient. The comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold, and different financial anomaly detection results are generated according to different comparison results.

2. The method for detecting anomalies in corporate finance according to claim 1, characterized in that: Based on each item of financial data obtained for the current month, the financial characteristic data is calculated and generated, wherein the specific formula for calculating the financial characteristic data is as follows: In the formula, is the cash flow ratio, represents cash flow from operating activities, For current liabilities, is the debt-to-asset ratio, is the total debt, is the total assets, is the working capital ratio, For current assets, is the sales profit margin, represents the net profit from sales, Represents sales revenue.

3. The method for detecting anomalies in corporate finance according to claim 2, characterized in that: The financial characteristic data obtained are used to calculate the enterprise financial abnormality characterization parameters, which include the financial stability index, the profit volatility index and the liquidity pressure index. The formula for calculating the financial stability index is: In the formula, is the financial stability index; The formula for calculating the earnings volatility index is: In the formula, is the earnings volatility index; The formula for calculating the liquidity pressure index is: In the formula, represents the liquidity pressure index, and They are accounts receivable turnover days and accounts payable turnover days respectively.

4. The method for detecting anomalies in corporate finance according to claim 3, characterized in that: The enterprise financial anomaly judgment coefficient is calculated based on the enterprise financial anomaly characterization parameters, where the formula for calculating the enterprise financial anomaly judgment coefficient is: In the formula, is the enterprise financial abnormality judgment coefficient, , and are the weight coefficients of the financial stability index, profit volatility index and liquidity pressure index, respectively. and , and Both are greater than 0.

5. The method for detecting anomalies in corporate finance according to claim 1, characterized in that: A prediction model is established based on the LSTM long short-term memory network model, and the activation function and optimization algorithm are selected. The Tanh function is selected as the activation function, and Adam is selected as the optimization algorithm of the LSTM model; the formula of the Tanh function is: In the formula, Represents the Tanh function, independent variable Represents the weighted sum of the neuron's input, that is, the result of weighted summation of the input received by the neuron from the previous layer; At the same time, the hyperparameters of the LSTM model are set, and the hyperparameters of the LSTM model include: the number of network layers, the number of iterations, the learning rate, the batch size, the number of training times, the batch processing number, and the number of hidden layer neurons; The number of network layers is set to a four-layer network structure, the number of iterations is set to 300, the learning rate is set to 0.001, the batch size is set to 32, the number of training times is set to 300, the batch size is set to 256, and the number of hidden layer neurons is 32.

6. The method for detecting anomalies in corporate finance according to claim 5, characterized in that: The market environment change parameters in the current month are input into the trained prediction model to obtain the estimated sales volume and sales cost of the enterprise to be analyzed in the current month. Based on the estimated sales volume and sales cost, the enterprise financial anomaly judgment coefficient is corrected to obtain the comprehensive financial anomaly judgment coefficient. The formula for calculating the comprehensive financial anomaly judgment coefficient is as follows: In the formula, is the comprehensive financial abnormality judgment coefficient, and They are the estimated sales volume and target sales volume of the company to be analyzed in the current month. and They are respectively the estimated value and target value of sales cost of the enterprise to be analyzed in the current month; The comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold, and different financial anomaly detection results are generated according to different comparison structures. The logic for generating different financial anomaly detection results is as follows: when When the company is analyzed, it is determined that there is no abnormality in the financial situation of the company to be analyzed; when If the financial situation of the enterprise to be analyzed is judged to be abnormal, it should be handled in a timely manner; In the formula, is the abnormal threshold.

7. An enterprise financial anomaly detection device, characterized by: The enterprise financial anomaly detection device is used to execute the garbage collection method according to any one of claims 1 to 6, comprising: The prediction model training module is used to obtain a number of financial data of the enterprise to be analyzed in a number of consecutive historical months before the current month, establish a first prediction model based on the obtained financial data, take the financial data of the enterprise to be analyzed in a number of consecutive historical months before the previous month as input, and use the financial data of the previous month as a label to train the first prediction model; A financial data collection module is used to input a number of financial data from a number of consecutive historical months before the current month into the trained first prediction model, and output a number of financial data for the current month, wherein the number of financial data includes cash flow from operating activities, current liabilities, total liabilities, total assets, current assets, net profit from sales, accounts receivable turnover days, accounts payable turnover days, sales volume, sales cost and sales revenue; The abnormal characterization parameter analysis module is used to calculate and generate financial characteristic data based on each item of financial data obtained in the current month, and calculate the abnormal financial characterization parameters of the enterprise through the obtained financial characteristic data, wherein the financial characteristic data include cash flow ratio, debt-to-asset ratio, working capital ratio and sales profit margin; The financial anomaly detection module is used to calculate the enterprise financial anomaly judgment coefficient based on the enterprise financial anomaly characterization parameters, wherein the enterprise financial anomaly characterization parameters include the financial stability index, the profit volatility index and the liquidity pressure index. The enterprise financial anomaly judgment coefficient is corrected based on the estimated sales volume and sales cost to obtain a comprehensive financial anomaly judgment coefficient, and the comprehensive financial anomaly judgment coefficient is compared with the anomaly threshold. Different financial anomaly detection results are generated according to different comparison results.

Images