Enterprise risk prediction method and system based on neural network

Abstract

The invention discloses an enterprise risk prediction method and system based on a neural network, and relates to the field of data processing, and the method comprises the steps: obtaining a plurality of risk evaluation indexes, and constructing a multi-dimensional index system; constructing a hierarchical game method evaluation model; determining the final weight of each index according to the multi-dimensional index system; determining an output result of the hierarchical game method evaluation model according to the actual value of each index and the corresponding final weight; constructing a neural network prediction model, and training the neural network prediction model based on the risk evaluation indexes to obtain a risk prediction model; inputting the enterprise data into the risk prediction model, and determining a risk prediction value; the result output by the hierarchical game method evaluation model is compared with the risk prediction value of the risk prediction model, the structure of the hierarchical game method evaluation model is adjusted, the enterprise risk is predicted according to the adjusted hierarchical game method evaluation model, the prediction result is obtained, and accurate prediction of the enterprise risk is achieved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method and system for predicting enterprise risks based on a neural network. Background Art

[0002] In the current complex and changing economic environment, industrial enterprises above designated size in the region are facing many development challenges such as market demand fluctuations, industry policy adjustments, and management risks. Timely and accurate identification of the risks faced by enterprises is crucial to enhancing their ability to resist risks and promoting the healthy development of regional industries.

[0003] However, most of the existing enterprise risk assessment methods are difficult to fully and objectively reflect the actual risk status of the enterprise, resulting in poor accuracy in enterprise risk prediction. Summary of the invention

[0004] The present invention provides a method and system for enterprise risk prediction based on a neural network, which is used to solve the above-mentioned problem existing in the prior art, that is, how to improve the accuracy of enterprise risk prediction in the prior art. The present invention provides a method for enterprise risk prediction based on a neural network, which comprises: Obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including the target layer, the criterion layer and the indicator layer, select different dimensional indicators of the same layer in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix; Construct a hierarchical game evaluation model; according to the judgment matrix, use the geometric mean method to obtain the initial weights of each indicator in the multi-dimensional indicator system, determine the objective function of maximizing the benefits of each stakeholder based on the initial weights, construct a game matrix, determine the Nash equilibrium of the game matrix through an iterative algorithm, obtain the final indicator weights, and determine the comprehensive risk evaluation results of the enterprise based on the actual values ​​of each risk evaluation indicator and the corresponding indicator weights; Construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value; By comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, the weights of various indicators of the hierarchical game method evaluation model are adjusted, and the risk value of each indicator is determined according to the adjusted hierarchical game method evaluation model. The risk values ​​of various indicators are weighted and summed up to predict the enterprise risk and obtain the prediction result.

[0005] Optionally, the geometric mean method is used to obtain the initial weight of each indicator in the multi-dimensional indicator system, specifically including: The following formula is used to obtain the initial weight of each indicator: , in, w i For the i The initial weight of the indicator, a ij Indicates the expert's i Relative to the indicator j The importance of judgment, n Indicates the total number of indicators.

[0006] Optionally, the training of the neural network prediction model specifically includes: The back propagation algorithm is used to train the neural network prediction model.

[0007] Optionally, determining the comprehensive risk assessment result of the enterprise according to the actual value of each risk assessment indicator and the corresponding indicator weight specifically includes: The following formula is used to obtain the comprehensive risk assessment results of the enterprise: , in, x j is the actual value of each risk assessment indicator, w j is the final weight corresponding to the actual value of each risk assessment indicator, R It is the comprehensive risk assessment result of the enterprise.

[0008] Optionally, the neural network prediction model specifically includes: , in, y is the output of the neural network, x i is the input layer node, w ij is the connection weight, b j is the bias term of the hidden layer node, f (·) is the activation function.

[0009] Also includes: After obtaining the initial weights of each indicator, the following formula is used to obtain the consistency index and consistency ratio: ， , in, CI is the consistency index, CR is the consistency ratio, λ maxis the maximum eigenvalue of the judgment matrix, RI is the average random consistency index, n Indicates the total number of indicators; Each judgment matrix is ​​tested for consistency by using consistency indicators and consistency ratios.

[0010] Also includes: After obtaining the prediction results through the adjusted hierarchical game evaluation model, a risk early warning linkage mechanism at the regional level is established.

[0011] The present invention provides an enterprise risk prediction system based on a neural network, comprising: The acquisition module is used to obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including a target layer, a criterion layer and an indicator layer, select different dimensional indicators of the same layer in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix; The hierarchical game method evaluation model construction module is used to construct the hierarchical game method evaluation model; according to the judgment matrix, the geometric mean method is used to obtain the initial weights of each indicator in the multi-dimensional indicator system, and according to the initial weights, the objective function of maximizing the benefits of each stakeholder is determined, and the game matrix is ​​constructed. The Nash equilibrium of the game matrix is ​​determined by an iterative algorithm to obtain the final indicator weights, and the comprehensive risk evaluation results of the enterprise are determined according to the actual values ​​of each risk evaluation indicator and the corresponding indicator weights; A neural network prediction model construction and training module is used to construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value; The prediction module is used to adjust the weights of various indicators of the hierarchical game method evaluation model by comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, determine the risk value of each indicator according to the adjusted hierarchical game method evaluation model, perform weighted summation of the risk values ​​of each indicator, predict enterprise risks, and obtain prediction results.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method for enterprise risk prediction based on a neural network. The method constructs a multi-dimensional indicator system by acquiring multiple risk evaluation indicators, which can comprehensively reflect the business conditions and development momentum of the enterprise, thereby providing a data basis for subsequent risk evaluation; according to the hierarchical game method evaluation model, the final weight of each indicator can be determined; the comprehensive risk evaluation score of the enterprise is determined by combining the actual value of each indicator and the corresponding final weight; by constructing and training the neural network prediction model, the risk prediction model is determined, and new enterprise data is input into the risk prediction model to obtain the risk prediction value, and then by comparing the results output by the hierarchical game method evaluation model with the neural network prediction results, the hierarchical structure of the hierarchical game method evaluation model is corrected according to the results, the final credibility of the algorithm is improved, and a dynamic quantitative evaluation of the enterprise development risk status can be achieved, thereby improving the accuracy of enterprise risk prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0014] Figure 1 A flowchart of a method for predicting enterprise risks based on a neural network provided by an embodiment of the present invention; Figure 2 A schematic diagram of an iterative process of a hierarchical game evaluation model for a neural network-based enterprise risk prediction method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0017] like Figure 1 and Figure 2 As shown, this embodiment shows a method for predicting enterprise risks based on a neural network, including: S1: Obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including the target layer, the criterion layer and the indicator layer, select different dimensional indicators of the same level in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix.

[0018] Optionally, risk assessment indicators include profitability indicators, debt-paying ability indicators, operational efficiency indicators, and development capability indicators.

[0019] For example, profitability indicators may include return on equity (ROE), gross profit margin and operating profit margin; debt-paying ability indicators may include debt-to-asset ratio, current ratio and quick ratio; operating efficiency indicators may include accounts receivable turnover rate, inventory turnover rate and total asset turnover rate; development capability indicators may include operating income growth rate, net profit growth rate and total asset growth rate.

[0020] For example, historical enterprise data, including multiple evaluation indicators and actual risk conditions, can be obtained, and then the acquired data can be standardized and divided into a training set and a test set.

[0021] In this step, a multi-dimensional indicator system is constructed through risk assessment indicators, which can comprehensively reflect the business conditions and development momentum of the enterprise and provide a data basis for subsequent risk assessment.

[0022] S2: Construct a hierarchical game evaluation model; based on the judgment matrix, use the geometric mean method to obtain the initial weights of each indicator in the multi-dimensional indicator system, determine the objective function of maximizing the benefits of each stakeholder based on the initial weights, construct a game matrix, determine the Nash equilibrium of the game matrix through an iterative algorithm, obtain the final indicator weights, and determine the comprehensive risk assessment results of the enterprise based on the actual values ​​of each risk assessment indicator and the corresponding indicator weights.

[0023] Optionally, the geometric mean method is used to obtain the initial weight of each indicator in the multi-dimensional indicator system. The specific calculation formula is as follows: ; in, w i For the i The initial weight of the indicator, a ij Indicates the expert's i Relative to the indicator j The importance of judgment, n Indicates the total number of indicators.

[0024] Optionally, the objective function of each subject is determined according to the initial weight, and the game matrix is ​​constructed based on the objective function of each subject.

[0025] For example, suppose there are p agents, each agent j (j=1,2,...,p) has its own objective function f j (w), we hope to maximize; based on the objective function of each subject, we can construct a p×n game matrix G=(g ij ), where g ij It indicates the preference of subject j for indicator i.

[0026] Optionally, after obtaining the initial weights of each indicator, the following formula is used to obtain the consistency indicator and consistency ratio: ； ; in, CI is the consistency index, CR is the consistency ratio, λ max is the maximum eigenvalue of the judgment matrix, RI is the average random consistency index, n Indicates the total number of indicators.

[0027] Each judgment matrix is ​​tested for consistency by using consistency indicators and consistency ratios.

[0028] Optionally, based on the actual value of each risk assessment indicator and the corresponding final weight, the following formula is used to obtain the comprehensive risk assessment result of the enterprise: ; in, x j is the actual value of each risk assessment indicator, w j is the final weight corresponding to the actual value of each risk assessment indicator, R It is the comprehensive risk assessment result of the enterprise.

[0029] Exemplarily, a hierarchical model is established, including a target layer (the comprehensive risk assessment results of the enterprise), a criterion layer (such as profitability, debt-paying ability, etc.) and an indicator layer (specific multiple risk assessment indicators), assuming that the target layer is A, the criterion layer is B = {B1, B2, ..., Bm}, and the indicator layer is C = {C1, C2, ..., Cn}; wherein, multiple risk evaluation indicators in the indicator layer construct an indicator in the criterion layer based on certain weights, and the risk assessment indicators in the criterion layer then construct the final target layer based on certain weights.

[0030] Select different dimensional indicators in the multi-dimensional indicator system for pairwise comparison, and use the following formula to obtain the judgment matrix: ; in, Indicators i Relative to the indicator j The importance of is the judgment matrix, k is the number of judgment matrices.

[0031] S3: Construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value.

[0032] Exemplarily, by acquiring historical enterprise data, which may include the above-mentioned multiple evaluation indicators and actual risk conditions, the data is standardized and divided into a training set and a test set.

[0033] Exemplarily, when constructing a neural network prediction model, for example, a feedforward neural network model can be constructed, including an input layer of 12 nodes, a hidden layer, and an output layer of 1 node, and the number of nodes in the hidden layer can be adjusted according to the training effect.

[0034] Optional, neural network prediction model, including: ; in, y is the output of the neural network, x i is the input layer node, w ij is the connection weight, b j is the bias term of the hidden layer node, f (·) is the activation function.

[0035] The back propagation algorithm is used to train the neural network model until the model converges. During the training process, the network structure and parameters need to be continuously adjusted to improve the prediction accuracy of the model. After the training, the trained model is verified using the test set data.

[0036] Exemplarily, the trained neural network model is applied to new enterprise data to output risk prediction values.

[0037] S4: By comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, the weights of various indicators of the hierarchical game method evaluation model are adjusted, and the risk value of each indicator is determined according to the adjusted hierarchical game method evaluation model. The risk values ​​of various indicators are weighted and summed up to predict the enterprise risk and obtain the prediction result.

[0038] In this step, by comparing the output results of the hierarchical game method evaluation model with the neural network prediction results, the hierarchical structure of the hierarchical game method evaluation model is reversely modified according to the results, which effectively improves the credibility of the prediction results and can realize a dynamic quantitative assessment of the enterprise development risk status.

[0039] Optionally, after obtaining the prediction results through the adjusted hierarchical game method evaluation model, a risk warning linkage mechanism at the regional level is established.

[0040] For example, the prediction results can be obtained according to the adjusted hierarchical game evaluation model, and the risk warning levels can be divided. When the risk level of a certain enterprise exceeds the warning threshold, warning information is sent to relevant regulatory departments and surrounding enterprises. The warning information may include, for example, the name of the enterprise, risk score, and major risk factors. After receiving the warning information, the relevant parties can log in to the warning platform to view the enterprise risk status and the overall risk situation of the region, and based on this, formulate targeted response measures, such as strengthening supervision and providing support, to promote risk information sharing and emergency coordination among enterprises in the region. The warning platform manager needs to regularly analyze the warning results and actual conditions of each enterprise, continuously optimize the warning rules and prediction model parameters, improve the accuracy and reliability of the prediction, and provide continuous and effective risk prevention and control support for the healthy development of regional industries.

[0041] The above is a method for predicting enterprise risk based on a neural network provided by one or more embodiments of this specification. Based on the same idea, this specification also provides a corresponding enterprise risk prediction system based on a neural network, including: The acquisition module is used to obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including a target layer, a criterion layer and an indicator layer, select different dimensional indicators of the same layer in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix; The hierarchical game method evaluation model construction module is used to construct the hierarchical game method evaluation model; according to the judgment matrix, the geometric mean method is used to obtain the initial weights of each indicator in the multi-dimensional indicator system, and according to the initial weights, the objective function of maximizing the benefits of each stakeholder is determined, and the game matrix is ​​constructed. The Nash equilibrium of the game matrix is ​​determined by an iterative algorithm to obtain the final indicator weights, and the comprehensive risk evaluation results of the enterprise are determined according to the actual values ​​of each risk evaluation indicator and the corresponding indicator weights; A neural network prediction model construction and training module is used to construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value; The prediction module is used to adjust the weights of various indicators of the hierarchical game method evaluation model by comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, determine the risk value of each indicator according to the adjusted hierarchical game method evaluation model, perform weighted summation of the risk values ​​of each indicator, predict enterprise risks, and obtain prediction results.

[0042] For the specific definition of the enterprise risk prediction system based on neural network, please refer to the definition of the enterprise risk prediction method based on neural network mentioned above, which will not be repeated here. Each module in the above-mentioned enterprise risk prediction system based on neural network can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0043] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present invention.

Claims

1. A method for predicting enterprise risk based on neural network, characterized in that: include: Obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including the target layer, the criterion layer and the indicator layer, select different dimensional indicators of the same layer in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix; Construct a hierarchical game theory evaluation model; According to the judgment matrix, the geometric mean method is used to obtain the initial weights of each indicator in the multi-dimensional indicator system. According to the initial weights, the objective function of maximizing the benefits of each stakeholder is determined, and the game matrix is ​​constructed. The Nash equilibrium of the game matrix is ​​determined by an iterative algorithm to obtain the final indicator weights. According to the actual values ​​of each risk assessment indicator and the corresponding indicator weights, the comprehensive risk assessment results of the enterprise are determined; Construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value; By comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, the weights of various indicators of the hierarchical game method evaluation model are adjusted, and the risk value of each indicator is determined according to the adjusted hierarchical game method evaluation model. The risk values ​​of various indicators are weighted and summed up to predict the enterprise risk and obtain the prediction result.

2. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: The geometric mean method is used to obtain the initial weight of each indicator in the multi-dimensional indicator system, specifically including: The following formula is used to obtain the initial weight of each indicator: ， in, w i For the i The initial weight of the indicator, a ij Indicates the expert's i Relative to the indicator j The importance of judgment, n Indicates the total number of indicators.

3. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: The training of the neural network prediction model specifically includes: The back propagation algorithm is used to train the neural network prediction model.

4. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: Determining the comprehensive risk assessment result of the enterprise according to the actual value of each risk assessment indicator and the corresponding indicator weight specifically includes: The following formula is used to obtain the comprehensive risk assessment results of the enterprise: ， in, x j is the actual value of each risk assessment indicator, w j is the final weight corresponding to the actual value of each risk assessment indicator, R It is the comprehensive risk assessment result of the enterprise.

5. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: The neural network prediction model specifically includes: ， in, y is the output of the neural network, x i is the input layer node, w ij is the connection weight, b j is the bias term of the hidden layer node, f (·) is the activation function.

6. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: Also includes: After obtaining the initial weights of each indicator, the following formula is used to obtain the consistency index and consistency ratio: ， ， in, CI is the consistency index, CR is the consistency ratio, λ max is the maximum eigenvalue of the judgment matrix, RI is the average random consistency index, n Indicates the total number of indicators; Each judgment matrix is ​​tested for consistency by using consistency indicators and consistency ratios.

7. The enterprise risk prediction method based on neural network as claimed in claim 1, characterized in that: Also includes: After obtaining the prediction results through the adjusted hierarchical game evaluation model, a risk early warning linkage mechanism at the regional level is established.

8. An enterprise risk prediction system based on neural network, characterized in that: include: The acquisition module is used to obtain multiple risk assessment indicators of the enterprise to be tested, generate a hierarchical structure based on the multiple risk assessment indicators, build a multi-dimensional indicator system including a target layer, a criterion layer and an indicator layer, select different dimensional indicators of the same layer in the multi-dimensional indicator system for pairwise comparison, and determine the judgment matrix; A hierarchical game method evaluation model construction module is used to construct a hierarchical game method evaluation model; According to the judgment matrix, the geometric mean method is used to obtain the initial weights of each indicator in the multi-dimensional indicator system. According to the initial weights, the objective function of maximizing the benefits of each stakeholder is determined, and the game matrix is ​​constructed. The Nash equilibrium of the game matrix is ​​determined by an iterative algorithm to obtain the final indicator weights. According to the actual values ​​of each risk assessment indicator and the corresponding indicator weights, the comprehensive risk assessment results of the enterprise are determined; A neural network prediction model construction and training module is used to construct a neural network prediction model, train the neural network prediction model based on the historical data of enterprise operations, obtain a risk prediction model for enterprise risk prediction, input enterprise data into the risk prediction model, and determine the risk prediction value; The prediction module is used to adjust the weights of various indicators of the hierarchical game method evaluation model by comparing the output of the hierarchical game method evaluation model with the output of the risk prediction model, determine the risk value of each indicator according to the adjusted hierarchical game method evaluation model, perform weighted summation of the risk values ​​of each indicator, predict enterprise risks, and obtain prediction results.

Images