Enterprise credit risk assessment method and system based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based enterprise credit risk assessment method and system. The method comprises the steps of data acquisition, data preprocessing, credit risk assessment model construction, hyper-parameter optimization and enterprise credit risk assessment. The invention relates to the technical field of enterprise credit risk assessment, in particular to an artificial intelligence-based enterprise credit risk assessment method and system, and the method comprises the steps: obtaining credit assessment original data through data collection; a data preprocessing method of data cleaning, data labeling, data coding, data normalization and data set segmentation is adopted; an improved kernel extreme learning machine model is adopted to carry out enterprise credit risk assessment, a two-way gating loop structure is utilized to efficiently learn a dependency relationship in time sequence data, and the accuracy and stability of assessment are improved; a waterfowl optimization algorithm is adopted to carry out hyper-parameter optimization, higher calculation efficiency and exploration capability are achieved, and hyper-parameters can be optimized in real time in a dynamic learning environment.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise credit risk assessment, and specifically to an enterprise credit risk assessment method and system based on artificial intelligence. Background Art

[0002] Corporate credit risk assessment is a systematic analysis of a company's debt repayment ability and credit status, usually involving financial indicator analysis, industry background research, and management quality assessment, with the goal of identifying, measuring, and managing risks that may lead to corporate defaults. It can help investors, financial institutions, and other stakeholders make more informed decisions, thereby reducing potential losses, improving capital operation efficiency, and promoting the sustainable development of companies and improving the credit environment.

[0003] However, traditional corporate credit risk assessment methods have technical problems such as difficulty in handling the large amount of nonlinear and time series data involved in credit risk assessment and inability to capture complex patterns in corporate data, resulting in low model prediction accuracy; traditional corporate credit risk assessment methods have technical problems such as low efficiency in large-scale data processing and weak generalization ability when optimizing model hyperparameters. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an enterprise credit risk assessment method and system based on artificial intelligence. In view of the technical problems that traditional enterprise credit risk assessment methods have difficulty in processing a large amount of nonlinear and time series data involved in credit risk assessment, cannot capture complex patterns in enterprise data, and result in low model prediction accuracy, this solution creatively adopts an improved kernel extreme learning machine model for enterprise credit risk assessment, and utilizes a bidirectional gated loop structure to efficiently learn the dependencies in time series data, and can effectively process nonlinear data, thereby improving the accuracy and stability of the assessment; in view of the technical problems that traditional enterprise credit risk assessment methods have low efficiency in large-scale data processing and weak generalization ability when optimizing model hyperparameters, this solution creatively adopts a water bird optimization algorithm for hyperparameter optimization, and has higher computing efficiency and exploration ability when facing large-scale data, and can optimize hyperparameters in real time in a dynamic learning environment.

[0005] The technical solution adopted by the present invention is as follows: The enterprise credit risk assessment method based on artificial intelligence provided by the present invention comprises the following steps:

[0006] Step S1: data collection;

[0007] Step S2: data preprocessing;

[0008] Step S3: Credit risk assessment model construction;

[0009] Step S4: hyperparameter optimization;

[0010] Step S5: Enterprise credit risk assessment.

[0011] Furthermore, in step S1, the data collection is used to collect the original data required for assessing the credit risk of the enterprise, specifically, to obtain the original data set for credit assessment through data collection;

[0012] The credit assessment original data set specifically includes basic enterprise information data, enterprise financial data, enterprise credit data and other information data. The basic enterprise information data specifically includes enterprise name, enterprise type, industry and business scope. The enterprise financial data specifically includes total assets, net assets, liabilities, revenue and expenditure data and cash flow data. The enterprise credit data specifically includes overdue record data and loan record data. The other information data specifically includes market share of other enterprises in the industry and average credit score of the industry.

[0013] Furthermore, in step S2, the data preprocessing is used to preprocess the collected raw data, and specifically includes the following steps:

[0014] Step S21: data cleaning, which is used to clean the original data, specifically to remove missing values ​​and duplicate values ​​in the original credit assessment data set to obtain a preliminary processed data set;

[0015] Step S22: data labeling, used to label the preliminary processed data, specifically labeling the data in the preliminary processed data set as safe, low risk, medium risk and high risk, and using them as data labels to obtain a labeled data set;

[0016] Step S23: data encoding, which is used to encode the marked data, specifically, using a one-hot encoding method to encode the marked data set to obtain an encoded data set;

[0017] Step S24: data normalization, which is used to normalize the coded data, specifically, using the minimum-maximum method to normalize the coded data set to obtain a normalized data set;

[0018] Step S25: data set segmentation, for segmenting the data set, specifically segmenting the normalized data set to obtain a credit assessment training set and a credit assessment test set;

[0019] Step S26: preprocessing, specifically obtaining a credit assessment training set and a credit assessment test set through the data cleaning, data labeling, data encoding, data normalization and data set segmentation.

[0020] Furthermore, in step S3, the credit risk assessment model is constructed to construct a model required for credit risk assessment, specifically including the following steps:

[0021] Step S31: extracting features, specifically constructing a bidirectional gated loop module and extracting features, the steps include:

[0022] Step S311: Preliminary feature extraction, the formula used is as follows:

[0023] ;

[0024] In the formula, represents the hidden state at the tth time step, represents the bidirectional gated recurrent unit function, and In represents the input of the bidirectional gated recurrent module;

[0025] Step S312: Calculate the attention score using the following formula:

[0026] ;

[0027] In the formula, represents the attention score of the hidden state at the t-th time step, represents the weight parameter used to control the degree of participation of cosine similarity, Le represents a learnable vector, Tr represents a transposition operation, represents the hyperbolic tangent function, represents the weight matrix for the linear transformation of the hidden state at the tth time step, represents the weight matrix used for the linear transformation of the average hidden state, represents the hidden state of the i-th time step, T represents the total number of time steps, Indicates calculation of L2 norm;

[0028] Step S313: Extract features using the following formula:

[0029] ;

[0030] In the formula, represents the attention weight, represents the attention score of the hidden state at the i-th time step, Fe represents the weighted feature, represents an exponential function with a natural constant as base;

[0031] Step S32: constructing an output matrix, specifically constructing an output matrix based on a Gaussian kernel function, the formula used is as follows:

[0032] ;

[0033] Where O represents the output matrix, represents the element in the ath row and bth column of the output matrix, represents the Gaussian kernel function, represents the characteristics of the ath sample, represents the characteristics of the b-th sample, represents the Gaussian kernel width;

[0034] Step S33: Calculate the output weight using the following formula:

[0035] ;

[0036] In the formula, represents the output weight, represents the output matrix obtained from the training set, represents the regularization parameter, I represents the unit matrix of the same size as the output matrix obtained from the training set, and Y represents the target true value matrix;

[0037] Step S34: Calculate the model output using the following formula:

[0038] ;

[0039] In the formula, represents the model prediction matrix, Represents the model's prediction result for the ath sample, represents the softmax function, represents the output layer weight matrix, Represents the output layer bias term;

[0040] Step S35: Design a loss function, specifically using a cross entropy loss function as the model loss function, and the formula used is as follows:

[0041] ;

[0042] In the formula, Loss represents the model loss function value, A represents the number of samples, and C represents the total number of categories. represents the true value of the a-th sample for the c-th category, Represents the model prediction result of the a-th sample for the c-th category;

[0043] Step S36: constructing and training a model, specifically, constructing a credit risk assessment model by extracting features, constructing an output matrix, calculating output weights, calculating model outputs and designing a loss function to obtain a credit risk assessment model.

[0044] Further, in step S4, the hyperparameter optimization is used to optimize the hyperparameters of the model, specifically using the waterbird optimization algorithm to optimize the hyperparameters of the credit risk assessment model to obtain an optimized credit risk assessment model;

[0045] The hyperparameter optimization specifically includes the following steps:

[0046] Step S41: Initialization, the steps include:

[0047] Step S411: Initialize the search space, specifically construct a water bird unit set and initialize the water bird unit position matrix. The water bird unit is used to represent the hyperparameter combination of the credit risk assessment model to be selected. The water bird unit position matrix is ​​expressed as follows:

[0048] ;

[0049] In the formula, represents the initial water bird unit position matrix, represents the position of the i-th water bird unit in the j-th dimension, represents the position of the I-th water bird unit in the J-th dimension, represents the location of the first water bird unit, represents the location of the second water bird unit, represents the position of the I-th water bird unit, I represents the total number of water bird units, and J represents the total number of dimensions contained in the water bird unit;

[0050] Step S412: determining the iteration stop condition, specifically selecting reaching the maximum number of iterations and the fitness reaching a threshold as the iteration termination condition;

[0051] Step S413: determine the fitness, specifically, using the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit, and obtaining the accuracy of the credit risk assessment model as the fitness of the water bird unit. The fitness vector is expressed as follows:

[0052] ;

[0053] In the formula, F represents the fitness vector, represents the fitness of the first water bird unit, represents the fitness of the second water bird unit, represents the fitness of the I-th water bird unit, represents the model accuracy calculation function obtained based on the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit;

[0054] Step S42: updating the location, the steps include:

[0055] Step S421: Design the approaching prey phase, the formula used is as follows:

[0056] ;

[0057] In the formula, represents the position of the i-th water bird unit in the j-th dimension during the approaching prey stage, represents a random value in the range [0,1]. represents a random value in the range [0,1]. Indicates the location of the prey, represents the parameter used to control the speed at which water birds approach their prey. represents the fitness of the prey, represents the fitness of the i-th water bird unit;

[0058] Step S422: updating the position in the approaching prey stage, specifically, comparing the fitness of the water bird unit in the approaching prey stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the approaching prey stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the approaching prey stage. If the fitness of the water bird unit in the approaching prey stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit.

[0059] Step S423: Design the water surface predation stage, the formula used is as follows:

[0060] ;

[0061] In the formula, represents the position of the i-th water bird unit in the j-th dimension during the surface predation phase, R represents the hunting radius, d represents the current number of iterations, and D represents the total number of iterations. Represents a random value in the range [0,1];

[0062] Step S424: updating the position in the water surface predation stage, specifically, comparing the fitness of the water bird unit in the water surface predation stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the water surface predation stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the water surface predation stage. If the fitness of the water bird unit in the water surface predation stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit.

[0063] Step S43: updating the global optimal position, specifically selecting the position of the water bird unit with the largest fitness as the global optimal position;

[0064] Step S44: iterative updating, specifically, continuously iteratively updating until the iteration stop condition is reached;

[0065] Step S45: performing hyperparameter optimization, specifically, optimizing the hyperparameters of the credit risk assessment model through the initialization, the update position, the global optimal position update and the iterative update to obtain an optimized credit risk assessment model.

[0066] Furthermore, in step S5, the enterprise credit risk assessment is specifically to use the optimized credit risk assessment model to conduct enterprise credit risk assessment, obtain credit risk reference data, and comprehensively consider the enterprise credit risk level based on the credit risk reference data.

[0067] The enterprise credit risk assessment system based on artificial intelligence provided by the present invention includes a data acquisition module, a data preprocessing module, a credit risk assessment model construction module, a hyperparameter optimization module and an enterprise credit risk assessment module;

[0068] The data collection module is used for data collection, obtains the original credit evaluation data set through data collection, and sends the original credit evaluation data set to the data preprocessing module;

[0069] The data preprocessing module is used for data preprocessing, and obtains a credit assessment training set and a credit assessment test set through data preprocessing, and sends the credit assessment training set and the credit assessment test set to the credit risk assessment model building module;

[0070] The credit risk assessment model construction module is used to construct a credit risk assessment model, obtain a credit risk assessment model by constructing the credit risk assessment model, and send the credit risk assessment model to the hyperparameter optimization module;

[0071] The hyperparameter optimization module is used to perform hyperparameter optimization, obtain an optimized credit risk assessment model by using the waterbird optimization algorithm to perform hyperparameter optimization, and send the optimized credit risk assessment model to the enterprise credit risk assessment module;

[0072] The enterprise credit risk assessment module is used for enterprise credit risk assessment, and obtains credit risk reference data by using an optimized credit risk assessment model to conduct enterprise credit risk assessment.

[0073] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0074] (1) In view of the technical problems that traditional enterprise credit risk assessment methods have difficulty in processing the large amount of nonlinear and time series data involved in credit risk assessment and cannot capture the complex patterns in enterprise data, resulting in low model prediction accuracy, this solution creatively adopts an improved kernel extreme learning machine model for enterprise credit risk assessment, and uses a bidirectional gated loop structure to efficiently learn the dependencies in time series data. It can also effectively process nonlinear data, thereby improving the accuracy and stability of the assessment.

[0075] (2) In view of the technical problems that traditional corporate credit risk assessment methods have inefficiencies in large-scale data processing and weak generalization capabilities when optimizing model hyperparameters, this solution creatively uses the water bird optimization algorithm for hyperparameter optimization. When faced with large-scale data, it has higher computing efficiency and exploration capabilities, and can optimize hyperparameters in real time in a dynamic learning environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 A schematic diagram of the process of the enterprise credit risk assessment method based on artificial intelligence provided by the present invention;

[0077] Figure 2 A schematic diagram of a module of an enterprise credit risk assessment system based on artificial intelligence provided by the present invention;

[0078] Figure 3 This is a schematic diagram of the process of data preprocessing in step S2;

[0079] Figure 4 A schematic diagram of the process of constructing a credit risk assessment model in step S3;

[0080] Figure 5 Schematic diagram of the process of hyperparameter optimization in step S4.

[0081] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0082] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0083] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0084] Example 1, see Figure 1 The technical solution adopted by the present invention is as follows: The enterprise credit risk assessment method based on artificial intelligence provided by the present invention comprises the following steps:

[0085] Step S1: data collection;

[0086] Step S2: data preprocessing;

[0087] Step S3: Credit risk assessment model construction;

[0088] Step S4: hyperparameter optimization;

[0089] Step S5: Enterprise credit risk assessment.

[0090] Example 2, see Figure 1 and Figure 2 In step S1, the data collection is used to collect the original data required for assessing the credit risk of the enterprise, specifically, to obtain the original data set for credit assessment through data collection;

[0091] The credit assessment original data set specifically includes basic enterprise information data, enterprise financial data, enterprise credit data and other information data. The basic enterprise information data specifically includes enterprise name, enterprise type, industry and business scope. The enterprise financial data specifically includes total assets, net assets, liabilities, revenue and expenditure data and cash flow data. The enterprise credit data specifically includes overdue record data and loan record data. The other information data specifically includes market share of other enterprises in the industry and average credit score of the industry.

[0092] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the data preprocessing is used to preprocess the collected raw data, and specifically includes the following steps:

[0093] Step S21: data cleaning, which is used to clean the original data, specifically to remove missing values ​​and duplicate values ​​in the original credit assessment data set to obtain a preliminary processed data set;

[0094] Step S22: data labeling, used to label the preliminary processed data, specifically labeling the data in the preliminary processed data set as safe, low risk, medium risk and high risk, and using them as data labels to obtain a labeled data set;

[0095] Step S23: data encoding, which is used to encode the marked data, specifically, using a one-hot encoding method to encode the marked data set to obtain an encoded data set;

[0096] Step S24: data normalization, which is used to normalize the coded data, specifically, using the minimum-maximum method to normalize the coded data set to obtain a normalized data set;

[0097] Step S25: data set segmentation, for segmenting the data set, specifically segmenting the normalized data set to obtain a credit assessment training set and a credit assessment test set;

[0098] Step S26: preprocessing, specifically obtaining a credit assessment training set and a credit assessment test set through the data cleaning, data labeling, data encoding, data normalization and data set segmentation.

[0099] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the credit risk assessment model is constructed to construct a model required for credit risk assessment, specifically including the following steps:

[0100] Step S31: extracting features, specifically constructing a bidirectional gated loop module and extracting features, the steps include:

[0101] Step S311: Preliminary feature extraction, the formula used is as follows:

[0102] ;

[0103] In the formula, represents the hidden state at the tth time step, represents the bidirectional gated recurrent unit function, and In represents the input of the bidirectional gated recurrent module;

[0104] Step S312: Calculate the attention score using the following formula:

[0105] ;

[0106] In the formula, represents the attention score of the hidden state at the t-th time step, represents the weight parameter used to control the degree of participation of cosine similarity, Le represents a learnable vector, Tr represents a transposition operation, represents the hyperbolic tangent function, represents the weight matrix for the linear transformation of the hidden state at the tth time step, represents the weight matrix used for the linear transformation of the average hidden state, represents the hidden state of the i-th time step, T represents the total number of time steps, Indicates calculation of L2 norm;

[0107] Step S313: Extract features using the following formula:

[0108] ;

[0109] In the formula, represents the attention weight, represents the attention score of the hidden state at the i-th time step, Fe represents the weighted feature, represents an exponential function with a natural constant as base;

[0110] Step S32: constructing an output matrix, specifically constructing an output matrix based on a Gaussian kernel function, the formula used is as follows:

[0111] ;

[0112] Where O represents the output matrix, represents the element in the ath row and bth column of the output matrix, represents the Gaussian kernel function, represents the characteristics of the ath sample, represents the characteristics of the b-th sample, represents the Gaussian kernel width;

[0113] Step S33: Calculate the output weight using the following formula:

[0114] ;

[0115] In the formula, represents the output weight, represents the output matrix obtained from the training set, represents the regularization parameter, I represents the unit matrix of the same size as the output matrix obtained from the training set, and Y represents the target true value matrix;

[0116] Step S34: Calculate the model output using the following formula:

[0117] ;

[0118] In the formula, represents the model prediction matrix, Represents the model's prediction result for the ath sample, represents the softmax function, represents the output layer weight matrix, Represents the output layer bias term;

[0119] Step S35: Design a loss function, specifically using a cross entropy loss function as the model loss function, and the formula used is as follows:

[0120] ;

[0121] In the formula, Loss represents the model loss function value, A represents the number of samples, and C represents the total number of categories. represents the true value of the a-th sample for the c-th category, Represents the model prediction result of the a-th sample for the c-th category;

[0122] Step S36: constructing and training a model, specifically, constructing a credit risk assessment model by extracting features, constructing an output matrix, calculating output weights, calculating model outputs and designing a loss function to obtain a credit risk assessment model.

[0123] By performing the above operations, this solution creatively adopts an improved kernel extreme learning machine model for enterprise credit risk assessment, and uses a bidirectional gated loop structure to efficiently learn the dependencies in time series data, and can effectively process nonlinear data, thereby improving the accuracy and stability of the assessment.

[0124] Example 5, see Figure 1 , Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the hyperparameter optimization is used to optimize the hyperparameters of the model, specifically, the hyperparameter optimization of the credit risk assessment model is performed using the water bird optimization algorithm to obtain an optimized credit risk assessment model;

[0125] The hyperparameter optimization specifically includes the following steps:

[0126] Step S41: Initialization, the steps include:

[0127] Step S411: Initialize the search space, specifically construct a water bird unit set and initialize the water bird unit position matrix. The water bird unit is used to represent the hyperparameter combination of the credit risk assessment model to be selected. The water bird unit position matrix is ​​expressed as follows:

[0128] ;

[0129] In the formula, represents the initial water bird unit position matrix, represents the position of the i-th water bird unit in the j-th dimension, represents the position of the I-th water bird unit in the J-th dimension, represents the location of the first water bird unit, represents the location of the second water bird unit, represents the position of the I-th water bird unit, I represents the total number of water bird units, and J represents the total number of dimensions contained in the water bird unit;

[0130] Step S412: determining the iteration stop condition, specifically selecting reaching the maximum number of iterations and the fitness reaching a threshold as the iteration termination condition;

[0131] Step S413: determine the fitness, specifically, using the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit, and obtaining the accuracy of the credit risk assessment model as the fitness of the water bird unit. The fitness vector is expressed as follows:

[0132] ;

[0133] In the formula, F represents the fitness vector, represents the fitness of the first water bird unit, represents the fitness of the second water bird unit, represents the fitness of the I-th water bird unit, represents the model accuracy calculation function obtained based on the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit;

[0134] Step S42: updating the location, the steps include:

[0135] Step S421: Design the approaching prey phase, the formula used is as follows:

[0136] ;

[0137] In the formula, represents the position of the i-th water bird unit in the j-th dimension during the approaching prey stage, represents a random value in the range [0,1]. represents a random value in the range [0,1]. Indicates the location of the prey, represents the parameter used to control the speed at which water birds approach their prey. represents the fitness of the prey, represents the fitness of the i-th water bird unit;

[0138] Step S422: updating the position in the approaching prey stage, specifically, comparing the fitness of the water bird unit in the approaching prey stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the approaching prey stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the approaching prey stage. If the fitness of the water bird unit in the approaching prey stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit.

[0139] Step S423: Design the water surface predation stage, the formula used is as follows:

[0140] ;

[0141] In the formula, represents the position of the i-th water bird unit in the j-th dimension during the surface predation phase, R represents the hunting radius, d represents the current number of iterations, and D represents the total number of iterations. Represents a random value in the range [0,1];

[0142] Step S424: updating the position in the water surface predation stage, specifically, comparing the fitness of the water bird unit in the water surface predation stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the water surface predation stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the water surface predation stage. If the fitness of the water bird unit in the water surface predation stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit.

[0143] Step S43: updating the global optimal position, specifically selecting the position of the water bird unit with the largest fitness as the global optimal position;

[0144] Step S44: iterative updating, specifically, continuously iteratively updating until the iteration stop condition is reached;

[0145] Step S45: performing hyperparameter optimization, specifically, optimizing the hyperparameters of the credit risk assessment model through the initialization, the update position, the global optimal position update and the iterative update to obtain an optimized credit risk assessment model.

[0146] By performing the above operations, this solution creatively uses the water bird optimization algorithm to optimize hyperparameters in order to address the technical problems of traditional corporate credit risk assessment methods, such as low efficiency in large-scale data processing and weak generalization ability when optimizing model hyperparameters. It has higher computing efficiency and exploration capabilities when facing large-scale data, and can optimize hyperparameters in real time in a dynamic learning environment.

[0147] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the enterprise credit risk assessment is specifically to use the optimized credit risk assessment model to perform enterprise credit risk assessment, obtain credit risk reference data, and comprehensively consider the enterprise credit risk level based on the credit risk reference data.

[0148] Embodiment 7, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment, and the enterprise credit risk assessment system based on artificial intelligence provided by the present invention includes a data acquisition module, a data preprocessing module, a credit risk assessment model construction module, a hyperparameter optimization module and an enterprise credit risk assessment module;

[0149] The data collection module is used for data collection, obtains the original credit evaluation data set through data collection, and sends the original credit evaluation data set to the data preprocessing module;

[0150] The data preprocessing module is used for data preprocessing, and obtains a credit assessment training set and a credit assessment test set through data preprocessing, and sends the credit assessment training set and the credit assessment test set to the credit risk assessment model building module;

[0151] The credit risk assessment model construction module is used to construct a credit risk assessment model, obtain a credit risk assessment model by constructing the credit risk assessment model, and send the credit risk assessment model to the hyperparameter optimization module;

[0152] The hyperparameter optimization module is used to perform hyperparameter optimization, obtain an optimized credit risk assessment model by using the waterbird optimization algorithm to perform hyperparameter optimization, and send the optimized credit risk assessment model to the enterprise credit risk assessment module;

[0153] The enterprise credit risk assessment module is used for enterprise credit risk assessment, and obtains credit risk reference data by using an optimized credit risk assessment model to conduct enterprise credit risk assessment.

[0154] It should be noted that, in this article, relational terms such as first and second, etc. 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 these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0155] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0156] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. An enterprise credit risk assessment method based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: data collection; Step S2: Data preprocessing: preprocessing the collected credit assessment raw data to obtain a credit assessment training set and a credit assessment test set; Step S3: constructing a credit risk assessment model, specifically constructing an improved kernel extreme learning machine model to obtain a credit risk assessment model; Step S4: Hyperparameter optimization, specifically using the waterbird optimization algorithm to optimize the hyperparameters to obtain an optimized credit risk assessment model; Step S5: Enterprise credit risk assessment, specifically using the optimized credit risk assessment model to conduct enterprise credit risk assessment and obtain credit risk reference data.

2. The enterprise credit risk assessment method based on artificial intelligence according to claim 1 is characterized by: In step S3, the credit risk assessment model is constructed to construct a model required for credit risk assessment, specifically including the following steps: Step S31: extracting features, specifically constructing a bidirectional gated loop module and extracting features, the steps include: Step S311: Preliminary feature extraction, the formula used is as follows: ； In the formula, represents the hidden state at the tth time step, represents the bidirectional gated recurrent unit function, and In represents the input of the bidirectional gated recurrent module; Step S312: Calculate the attention score using the following formula: ； In the formula, represents the attention score of the hidden state at the t-th time step, represents the weight parameter used to control the degree of participation of cosine similarity, Le represents a learnable vector, Tr represents a transposition operation, represents the hyperbolic tangent function, represents the weight matrix for the linear transformation of the hidden state at the tth time step, represents the weight matrix used for the linear transformation of the average hidden state, represents the hidden state of the i-th time step, T represents the total number of time steps, Indicates calculation of L2 norm; Step S313: Extract features using the following formula: ； In the formula, represents the attention weight, represents the attention score of the hidden state at the i-th time step, Fe represents the weighted feature, represents an exponential function with a natural constant as base; Step S32: constructing an output matrix, specifically constructing an output matrix based on a Gaussian kernel function, the formula used is as follows: ； Where O represents the output matrix, represents the element in the ath row and bth column of the output matrix, represents the Gaussian kernel function, represents the characteristics of the ath sample, represents the characteristics of the b-th sample, represents the Gaussian kernel width; Step S33: Calculate the output weight using the following formula: ； In the formula, represents the output weight, represents the output matrix obtained from the training set, represents the regularization parameter, I represents the unit matrix of the same size as the output matrix obtained from the training set, and Y represents the target true value matrix; Step S34: Calculate the model output using the following formula: ； In the formula, represents the model prediction matrix, Represents the model's prediction result for the ath sample, represents the softmax function, represents the output layer weight matrix, Represents the output layer bias term; Step S35: Design a loss function, specifically using a cross entropy loss function as the model loss function, and the formula used is as follows: ； In the formula, Loss represents the model loss function value, A represents the number of samples, and C represents the total number of categories. represents the true value of the a-th sample for the c-th category, Represents the model prediction result of the a-th sample for the c-th category; Step S36: constructing and training a model, specifically, constructing a credit risk assessment model by extracting features, constructing an output matrix, calculating output weights, calculating model outputs and designing a loss function to obtain a credit risk assessment model.

3. The enterprise credit risk assessment method based on artificial intelligence according to claim 1 is characterized by: In step S4, the hyperparameter optimization is used to optimize the hyperparameters of the model, specifically using the waterbird optimization algorithm to optimize the hyperparameters of the credit risk assessment model to obtain an optimized credit risk assessment model; The hyperparameter optimization specifically includes the following steps: Step S41: Initialization, the steps include: Step S411: Initialize the search space, specifically construct a water bird unit set and initialize the water bird unit position matrix. The water bird unit is used to represent the hyperparameter combination of the credit risk assessment model to be selected. The water bird unit position matrix is ​​expressed as follows: ； In the formula, represents the initial water bird unit position matrix, represents the position of the i-th water bird unit in the j-th dimension, represents the position of the I-th water bird unit in the J-th dimension, represents the location of the first water bird unit, represents the location of the second water bird unit, represents the position of the I-th water bird unit, I represents the total number of water bird units, and J represents the total number of dimensions contained in the water bird unit; Step S412: determining the iteration stop condition, specifically selecting reaching the maximum number of iterations and the fitness reaching a threshold as the iteration termination condition; Step S413: determine the fitness, specifically, using the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit, and obtaining the accuracy of the credit risk assessment model as the fitness of the water bird unit. The fitness vector is expressed as follows: ； In the formula, F represents the fitness vector, represents the fitness of the first water bird unit, represents the fitness of the second water bird unit, represents the fitness of the I-th water bird unit, represents the model accuracy calculation function obtained based on the corresponding candidate credit risk assessment model hyperparameter combination represented by the water bird unit; Step S42: updating the location, the steps include: Step S421: Design the approaching prey phase, the formula used is as follows: ； In the formula, represents the position of the i-th water bird unit in the j-th dimension during the approaching prey stage, represents a random value in the range [0,1]. represents a random value in the range [0,1]. Indicates the location of the prey, represents the parameter used to control the speed at which water birds approach their prey. represents the fitness of the prey, represents the fitness of the i-th water bird unit; Step S422: updating the position in the approaching prey stage, specifically, comparing the fitness of the water bird unit in the approaching prey stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the approaching prey stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the approaching prey stage. If the fitness of the water bird unit in the approaching prey stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit. Step S423: Design the water surface predation stage, the formula used is as follows: ； In the formula, represents the position of the i-th water bird unit in the j-th dimension during the surface predation phase, R represents the hunting radius, d represents the current number of iterations, and D represents the total number of iterations. Represents a random value in the range [0,1]; Step S424: updating the position in the water surface predation stage, specifically, comparing the fitness of the water bird unit in the water surface predation stage with the fitness of the original water bird unit. If the fitness of the water bird unit in the water surface predation stage is greater than the fitness of the original water bird unit, the updated position is the position of the water bird unit in the water surface predation stage. If the fitness of the water bird unit in the water surface predation stage is less than or equal to the fitness of the original water bird unit, the updated position is the position of the original water bird unit. Step S43: updating the global optimal position, specifically selecting the position of the water bird unit with the largest fitness as the global optimal position; Step S44: iterative updating, specifically, continuously iteratively updating until the iteration stop condition is reached; Step S45: performing hyperparameter optimization, specifically, optimizing the hyperparameters of the credit risk assessment model through the initialization, the update position, the global optimal position update and the iterative update to obtain an optimized credit risk assessment model.

4. The enterprise credit risk assessment method based on artificial intelligence according to claim 1 is characterized by: In step S1, the data collection is used to collect the original data required for assessing the credit risk of the enterprise, specifically, to obtain the original data set for credit assessment through data collection; The credit assessment original data set specifically includes basic enterprise information data, enterprise financial data, enterprise credit data and other information data. The basic enterprise information data specifically includes enterprise name, enterprise type, industry and business scope. The enterprise financial data specifically includes total assets, net assets, liabilities, revenue and expenditure data and cash flow data. The enterprise credit data specifically includes overdue record data and loan record data. The other information data specifically includes market share of other enterprises in the industry and average credit score of the industry.

5. The enterprise credit risk assessment method based on artificial intelligence according to claim 1 is characterized by: In step S2, the data preprocessing is used to preprocess the collected raw data, and specifically includes the following steps: Step S21: data cleaning, which is used to clean the original data, specifically to remove missing values ​​and duplicate values ​​in the original credit assessment data set to obtain a preliminary processed data set; Step S22: data labeling, used to label the preliminary processed data, specifically labeling the data in the preliminary processed data set as safe, low risk, medium risk and high risk, and using them as data labels to obtain a labeled data set; Step S23: data encoding, which is used to encode the marked data, specifically, using a one-hot encoding method to encode the marked data set to obtain an encoded data set; Step S24: data normalization, which is used to normalize the coded data, specifically, using the minimum-maximum method to normalize the coded data set to obtain a normalized data set; Step S25: data set segmentation, for segmenting the data set, specifically segmenting the normalized data set to obtain a credit assessment training set and a credit assessment test set; Step S26: preprocessing, specifically obtaining a credit assessment training set and a credit assessment test set through the data cleaning, data labeling, data encoding, data normalization and data set segmentation.

6. The enterprise credit risk assessment method based on artificial intelligence according to claim 1 is characterized by: In step S5, the enterprise credit risk assessment is specifically to use the optimized credit risk assessment model to conduct enterprise credit risk assessment, obtain credit risk reference data, and comprehensively consider the enterprise credit risk level based on the credit risk reference data.

7. An enterprise credit risk assessment system based on artificial intelligence, used to implement the enterprise credit risk assessment method based on artificial intelligence as described in any one of claims 1 to 6, characterized in that: It includes data collection module, data preprocessing module, credit risk assessment model building module, hyperparameter optimization module and enterprise credit risk assessment module.

8. The enterprise credit risk assessment system based on artificial intelligence according to claim 7 is characterized by: The data collection module is used for data collection, obtains the original credit evaluation data set through data collection, and sends the original credit evaluation data set to the data preprocessing module; The data preprocessing module is used for data preprocessing, and obtains a credit assessment training set and a credit assessment test set through data preprocessing, and sends the credit assessment training set and the credit assessment test set to the credit risk assessment model building module; The credit risk assessment model construction module is used to construct a credit risk assessment model, obtain a credit risk assessment model by constructing the credit risk assessment model, and send the credit risk assessment model to the hyperparameter optimization module; The hyperparameter optimization module is used to perform hyperparameter optimization, obtain an optimized credit risk assessment model by using the waterbird optimization algorithm to perform hyperparameter optimization, and send the optimized credit risk assessment model to the enterprise credit risk assessment module; The enterprise credit risk assessment module is used for enterprise credit risk assessment, and obtains credit risk reference data by performing enterprise credit risk assessment using an optimized credit risk assessment model.

Images