Intelligent behavior evaluation method, device and equipment based on interval type-2 fuzzy neural network and storage medium

Abstract

The embodiment of the invention provides an intelligent behavior evaluation method and device based on an interval type-2 fuzzy neural network, equipment and a storage medium, and the method comprises the steps: introducing multivariable dynamic features, including a plurality of key variables such as access time, task execution time, system warning times and the like, comprehensively reflecting the credibility of system behaviors, and improving the reliability of the system behaviors; the interval type-2 fuzzy neural network is adopted to describe uncertainty and dynamic fluctuation in audit data, flexibility and robustness under different conditions are improved, a self-adaptive weighted compensation mechanism is designed, the weight is automatically adjusted according to time sequence changes of different behavior variables, and adaptability and evaluation precision of the system in a complex audit environment are improved. In addition, the audit terminal system behavior intelligent evaluation method based on the interval type-2 fuzzy neural network can greatly improve the intelligent level, the accuracy and the applicability of the audit system in a complex environment.

Description

Technical Field

[0001] The present application relates to the technical field of audit terminal systems, and in particular to a behavior intelligent evaluation method, device, equipment and storage medium based on interval type-2 fuzzy neural networks. Background Art

[0002] With the continuous improvement of enterprise information management level, the application of audit terminal systems in enterprise operation management is becoming more and more extensive. The audit terminal system can not only record the user's operation behavior, but also monitor the system's operating status, providing important protection for the company's information security and compliance management. The audit terminal system helps enterprises identify potential risks by collecting and analyzing data such as operation logs, task execution status, and access records. The core of the audit terminal system behavior assessment lies in the intelligent analysis of system status and behavior data. However, with the increase in system complexity, how to effectively evaluate the behavior of the audit terminal system and promptly discover potential security risks and abnormal behaviors has become a problem that needs to be solved urgently.

[0003] Most traditional assessment methods are based on fuzzy logic systems for risk assessment and behavior detection. Fuzzy logic systems, with their advantages of explicit knowledge and direct algorithms, excel in system simplicity and small data processing volume. However, fuzzy logic systems rely on preset rules and logic for behavior assessment and judgment, lack autonomous learning capabilities, and are difficult to adapt to dynamically changing system environments. In order to overcome the limitations of fuzzy logic systems, neural networks with powerful nonlinear mapping and adaptive learning capabilities have been introduced into the behavior assessment of audit terminal systems. However, neural networks are sensitive to data noise and uncertainty, making it difficult to accurately assess the behavior of audit terminal systems.

[0004] The above information disclosed in the background section is only for enhancing understanding of the background of the present application and therefore it may contain information that does not form the prior art known to a person of ordinary skill in the art. Summary of the invention

[0005] The embodiment of the present application provides a behavioral intelligent evaluation method based on interval type-2 fuzzy neural network to solve the technical problems of uncertainty and inaccurate credibility evaluation of traditional audit systems when processing complex dynamic data.

[0006] According to a first aspect of an embodiment of the present application, a behavior intelligent evaluation method based on an interval type-2 fuzzy neural network is provided, comprising:

[0007] Acquire a plurality of sample data sets of dynamic behavior characteristic variables that affect the credibility level, and obtain each normalized dynamic behavior characteristic variable by normalizing the sample data set of each dynamic behavior characteristic variable;

[0008] Constructing an interval type-2 fuzzy neural network model, and optimizing the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior characteristic variable to obtain an optimized interval type-2 fuzzy neural network model;

[0009] Compensating the output of the optimized interval type-2 fuzzy neural network model based on an adaptive weighted compensation mechanism to obtain a compensation error of the output result of the optimized interval type-2 fuzzy neural network model;

[0010] Sample data of the dynamic behavior characteristic variable to be evaluated is obtained, and the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated is obtained by using the optimized interval type-2 fuzzy neural network model and the compensation error.

[0011] Preferably, the multiple dynamic behavior characteristic variables affecting the trust level include access time, access failure time, task execution time, task stop time, access stay time, access times and system warning times.

[0012] Preferably, the interval type-2 fuzzy neural network model includes, from left to right, an input layer, a membership layer, an activation layer, a consequent layer, and an output layer;

[0013] The input layer is composed of 7 neurons, and the input variables are normalized dynamic behavior characteristic variables;

[0014] The membership layer consists of 5 neurons, and the output of each neuron is:

[0015]

[0016] Among them, μ 1ij (t) is the upper bound output of the jth neuron with the ith input at time t, μ 2ij (t) is the lower bound output of the jth neuron with the ith input at time t, n ij (t) is the upper center of the i-th input to the j-th neuron at time t, m ij (t) is the lower center of the i-th input to the j-th neuron at time t, σ ij (t) is the width of the jth neuron of the ith input at time t, belonging to the neurons in the layer j = 1,…,5; u i (t) is the normalized dynamic behavior characteristic variable of the i-th input at time t.

[0017] Preferably, the activation layer is composed of 5 neurons, and the upper and lower bound outputs of the activation layer are expressed as:

[0018]

[0019] Among them, f 1l(t) is the upper bound output of the lth neuron at time t, f 2l (t) is the lower bound output of the lth neuron at time t, and the neurons in the activation layer are l=1,…,5;

[0020] The subsequent layer consists of 2 neurons, and the output of the subsequent layer is expressed as:

[0021]

[0022] Among them, y1(t) is the upper output value of the subsequent layer at time t, y2(t) is the lower output value of the subsequent layer at time t, and a ij (t) is the weight coefficient of the i-th input to the j-th neuron at time t, b ij (t) is the bias of the ith input to the jth neuron at time t;

[0023] The output layer consists of 1 neuron, and the output of the output layer is expressed as:

[0024] y(t)=(1-q(t))y1(t)+q(t)y2(t)

[0025] Among them, y(t) is the predicted output at time t, and q(t) is the proportional coefficient at time t.

[0026] Preferably, the optimized interval type-2 fuzzy neural network model includes:

[0027]

[0028] Among them, Ω s (t) is the parameter vector of the sth interval type-2 fuzzy neural network at time t, Ω s+1 (t) is the s+1th interval type-2 fuzzy neural network parameter vector at time t, is the partial derivative of the sth parameter vector; y s (t) is the predicted output result of the sth sample at time t; is the true credibility of the system behavior at time t+s-1.

[0029] Preferably, the compensation error of the output result of the optimized interval type-2 fuzzy neural network model includes:

[0030]

[0031] in, is the compensation error at time t+16; λ(t+o) is the correction factor at time t+o, is the true credibility of the system behavior at time t+o, y(t+o) is the predicted output at time t+o, and o is any time.

[0032] Preferably, the use of the optimized interval type-2 fuzzy neural network model and the compensation error to obtain the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated includes:

[0033] By inputting the sample data of the dynamic behavior characteristic variable to be evaluated into the optimized interval type-2 fuzzy neural network model, a prediction output result is obtained;

[0034] By adding the prediction output result and the compensation error, a normalized credibility level of the sample data to be evaluated is obtained;

[0035] The predicted credibility level is obtained by performing a denormalization process on the normalized credibility level of the sample data to be evaluated, and the predicted credibility level is used as the evaluation result of the dynamic behavior characteristic variable to be evaluated.

[0036] According to a second aspect of an embodiment of the present application, there is provided a behavior intelligent evaluation device based on an interval type-2 fuzzy neural network, comprising:

[0037] A processing module, used for obtaining a plurality of sample data sets of dynamic behavior characteristic variables affecting the credibility level, and performing normalization processing on the sample data sets of each dynamic behavior characteristic variable to obtain each normalized dynamic behavior characteristic variable;

[0038] A construction and optimization module, used for constructing an interval type-2 fuzzy neural network model, and optimizing the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior characteristic variable to obtain an optimized interval type-2 fuzzy neural network model;

[0039] A compensation module, used for compensating the output of the optimized interval type-2 fuzzy neural network model based on an adaptive weighted compensation mechanism to obtain a compensation error of the output result of the optimized interval type-2 fuzzy neural network model;

[0040] The behavior evaluation module is used to obtain sample data of the dynamic behavior characteristic variable to be evaluated, and obtain the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated by using the optimized interval type-II fuzzy neural network model and the compensation error.

[0041] According to a third aspect of the embodiments of the present application, there is provided an electronic device, including: a memory; a processor; and a computer program;

[0042] Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the steps of the behavior intelligent evaluation method based on interval type-2 fuzzy neural network.

[0043] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored; the computer program is executed by a processor to implement the steps of a behavioral intelligent evaluation method based on an interval type-2 fuzzy neural network.

[0044] The embodiment of the present application adopts the above technical solution, which has the following technical effects:

[0045] 1. Compared with traditional methods that usually have difficulty in handling complex nonlinear features and fuzzy relationships in multi-dimensional data, the fuzzy network model of this application introduces type-2 fuzzy sets to more accurately model the uncertainty in the data, effectively improving the adaptability and evaluation accuracy of the system in different environments.

[0046] 2. Compared with the limitations of single-variable evaluation in traditional methods, this application introduces a multivariate dynamic feature integration mechanism, which enables the system to simultaneously consider multiple key variables (such as access time, operation frequency, number of abnormal warnings, etc.) when evaluating behavior, providing more reliable evaluation results.

[0047] 3. Compared with the shortcomings of fixed fuzzy rules in the prior art, this application uses adaptive rule adjustment to enable the system to optimize the inference rules of fuzzy neural networks under different operating behaviors and environmental conditions, thereby accurately reflecting data characteristics in an environment with obvious nonlinear fluctuations, significantly enhancing the dynamics and flexibility of the evaluation, and effectively improving the accuracy of the trust level evaluation.

[0048] 4. Traditional methods lack an effective correction mechanism for evaluation bias. This application introduces weighted compensation and dynamically adjusts variable weights to correct bias in the evaluation process, ensuring the stability of the evaluation under conditions of parameter fluctuations or data anomalies.

[0049] 5. Compared with the limitations of traditional technologies in complex environments, this application provides a more intelligent and practical evaluation method for the audit terminal system through multi-dimensional data integration, dynamic weighting and model self-calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0051] Figure 1 A flowchart of a behavior intelligent evaluation method based on interval type-2 fuzzy neural network provided in an embodiment of the present application;

[0052] Figure 2 A schematic diagram of a behavior intelligent evaluation device based on an interval type-2 fuzzy neural network provided in an embodiment of the present application;

[0053] Figure 3 A schematic diagram of an intelligent evaluation method for audit terminal system behavior based on an interval type-2 fuzzy neural network model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the technical solutions and advantages in the embodiments of the present application more clearly understood, the exemplary embodiments of the present application are further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than an exhaustive list of all the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0055] This application proposes a behavioral intelligent evaluation method based on interval type-2 fuzzy neural network, aiming to effectively solve the problem of uncertainty and inaccurate credibility evaluation of existing audit systems when processing complex dynamic data. This method introduces multivariate dynamic features, including multiple key variables such as access time, task execution time, and number of system warnings, to comprehensively reflect the credibility of system behavior. It uses interval type-2 fuzzy neural networks to characterize the uncertainty and dynamic fluctuations in audit data, improves the flexibility and robustness under different conditions, and designs an adaptive weighted compensation mechanism to automatically adjust the weights according to the time series changes of different behavioral variables, thereby improving the adaptability and evaluation accuracy of the system in complex audit environments. The audit terminal system behavior intelligent evaluation method based on interval type-2 fuzzy neural networks can greatly improve the intelligence level, accuracy and applicability of the audit system in complex environments, enabling the system to provide enterprises with more reliable audit risk assessment results and management decision support, representing the future technology development trend and having broad market prospects.

[0056] Embodiment 1

[0057] like Figure 1 As shown, a flowchart of a behavior intelligent evaluation method based on interval type-2 fuzzy neural network in an embodiment of the present application includes:

[0058] Step S1: obtaining a plurality of sample data sets of dynamic behavior characteristic variables that affect the credibility level, and performing normalization processing on the sample data sets of each dynamic behavior characteristic variable to obtain each normalized dynamic behavior characteristic variable;

[0059] Furthermore, the multiple dynamic behavior characteristic variables affecting the trust level include access time, access failure time, task execution time, task stop time, access stay time, access times and system warning times.

[0060] Step S2: constructing an interval type-2 fuzzy neural network model, and optimizing the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior characteristic variable to obtain an optimized interval type-2 fuzzy neural network model;

[0061] In the embodiment of the present application, the interval type-2 fuzzy neural network model includes, from left to right, an input layer, a membership layer, an activation layer, a consequent layer, and an output layer;

[0062] The input layer is composed of 7 neurons, and the input variables are normalized dynamic behavior characteristic variables;

[0063] The membership layer consists of 5 neurons, and the output of each neuron is:

[0064]

[0065] Among them, μ 1ij (t) is the upper bound output of the jth neuron with the ith input at time t, μ 2ij (t) is the lower bound output of the jth neuron with the ith input at time t, n ij (t) is the upper center of the i-th input to the j-th neuron at time t, m ij (t) is the lower center of the i-th input to the j-th neuron at time t, σ ij (t) is the width of the jth neuron of the ith input at time t, belonging to the neurons in the layer j = 1,…,5; u i (t) is the normalized dynamic behavior characteristic variable of the i-th input at time t.

[0066] In the embodiment of the present application, the activation layer is composed of 5 neurons, and the upper and lower bound outputs of the activation layer are expressed as:

[0067]

[0068] Among them, f 1l (t) is the upper bound output of the lth neuron at time t, f 2l (t) is the lower bound output of the lth neuron at time t, and the neurons in the activation layer are l=1,…,5;

[0069] The subsequent layer consists of 2 neurons, and the output of the subsequent layer is expressed as:

[0070]

[0071] Among them, y1(t) is the upper output value of the subsequent layer at time t, y2(t) is the lower output value of the subsequent layer at time t, and a ij (t) is the weight coefficient of the i-th input to the j-th neuron at time t, b ij(t) is the bias of the ith input to the jth neuron at time t;

[0072] The output layer consists of 1 neuron, and the output of the output layer is expressed as:

[0073] y(t)=(1-q(t))y1(t)+q(t)y2(t)

[0074] Among them, y(t) is the predicted output at time t, and q(t) is the proportional coefficient at time t.

[0075] Step S3: compensating the output of the optimized interval type-2 fuzzy neural network model based on an adaptive weighted compensation mechanism to obtain a compensation error of the output result of the optimized interval type-2 fuzzy neural network model;

[0076] In the embodiment of the present application, the optimized interval type-2 fuzzy neural network model includes:

[0077]

[0078] Among them, Ω s (t) is the parameter vector of the sth interval type-2 fuzzy neural network at time t, Ω s+1 (t) is the s+1th interval type-2 fuzzy neural network parameter vector at time t, is the partial derivative of the sth parameter vector; y s (t) is the predicted output result of the sth sample at time t; is the true credibility of the system behavior at time t+s-1.

[0079] Step S4: acquiring sample data of the dynamic behavior characteristic variable to be evaluated, and using the optimized interval type-2 fuzzy neural network model and the compensation error to obtain a behavior evaluation result of the dynamic behavior characteristic variable to be evaluated.

[0080] In the embodiment of the present application, the compensation error of the output result of the optimized interval type-2 fuzzy neural network model includes:

[0081]

[0082] in, is the compensation error at time t+16; λ(t+o) is the correction factor at time t+o, is the true credibility of the system behavior at time t+o, y(t+o) is the predicted output at time t+o, and o is any time.

[0083] In the embodiment of the present application, the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated is obtained by using the optimized interval type-2 fuzzy neural network model and the compensation error, including:

[0084] By inputting the sample data of the dynamic behavior characteristic variable to be evaluated into the optimized interval type-2 fuzzy neural network model, a prediction output result is obtained;

[0085] By adding the prediction output result and the compensation error, a normalized credibility level of the sample data to be evaluated is obtained;

[0086] The predicted credibility level is obtained by performing a denormalization process on the normalized credibility level of the sample data to be evaluated, and the predicted credibility level is used as the evaluation result of the dynamic behavior characteristic variable to be evaluated.

[0087] In summary, the behavioral intelligent evaluation method based on interval type-2 fuzzy neural network provided in the embodiment of the present application comprehensively reflects the credibility of system behavior by introducing multivariate dynamic characteristics, including multiple key variables such as access time, task execution time, and number of system warnings. It uses interval type-2 fuzzy neural network to characterize the uncertainty and dynamic fluctuations in audit data, improves the flexibility and robustness in dealing with different conditions, and designs an adaptive weighted compensation mechanism to automatically adjust the weights according to the temporal changes of different behavioral variables, thereby improving the adaptability and evaluation accuracy of the system in complex audit environments.

[0088] Embodiment 2

[0089] The audit terminal system behavior intelligent evaluation method based on interval type II fuzzy neural network described in this application is mainly used in risk assessment scenarios in train operation behavior audit, train maintenance management audit, and financial audit. The framework of the audit terminal system behavior intelligent evaluation method based on interval type II fuzzy neural network is as follows: Figure 3 As shown, including:

[0090] 1) Taking the audit terminal system as the research object, the access time, access failure time, task execution time, task stop time, access stay time, access times and system warning times are selected as the characteristic variables affecting the trust level, and the characteristic variables are normalized:

[0091]

[0092] Where i = 1,…,8; t = 1,2,…,Z, Z is the total number of samples, S1(t) is the access time, S2(t) is the access failure time, S3(t) is the task execution time, S4(t) is the task stop time, S5(t) is the access stay time, S6(t) is the number of accesses, S7(t) is the number of system warnings, S8(t) is the trust level, S i,min is the minimum value of all samples of the i-th variable, S i,max is the maximum value of all samples of the i-th variable; u1(t)=I1(t) is the normalized access time, u2(t)=I2(t) is the normalized access failure time, u3(t)=I3(t) is the normalized task execution time, u4(t)=I4(t) is the normalized task stop time, u5(t)=I5(t) is the normalized access stay time, u6(t)=I6(t) is the normalized access number, u7(t)=I7(t) is the normalized system warning number, is the true credibility of the normalized audit terminal system behavior; the credibility level is obtained from the audit terminal system and determined according to the collected credibility. If the credibility>0.5, the credibility level corresponding to the sample is considered high; if -0.5<credibility<0.5, the credibility level corresponding to the sample is considered medium; if credibility<-0.5, the credibility level corresponding to the sample is considered low; the credibility is the predicted output value of the interval type II fuzzy neural network, which is used to realize the intelligent evaluation of the audit terminal system behavior.

[0093] 2) The intelligent evaluation method of the audit terminal system behavior is based on the interval type-2 fuzzy neural network. The model consists of five layers, such as Figure 3 As shown:

[0094] Input layer: consists of 7 neurons, input variable u(t) = [u1(t),u2(t),…,u6(t),u7(t)] T , T is the transpose calculation;

[0095] Membership layer: It consists of 5 neurons, and the output of each neuron is:

[0096]

[0097] Among them, μ 1ij (t) is the upper bound output of the jth neuron with the ith input, μ 2ij (t) is the lower bound output of the jth neuron of the ith input, n ij (t) is the upper center of the ith input to the jth neuron, m ij (t) is the lower center of the ith input to the jth neuron, σ ij (t) is the width of the i-th input to the j-th neuron, j = 1,…,5;

[0098] Activation layer: It consists of 5 neurons. The upper and lower bound outputs of the activation layer are expressed as:

[0099]

[0100] Among them, f 1l (t) is the upper bound output of the lth neuron, f 2l (t) is the lower bound output of the lth neuron, l = 1,…,5;

[0101] The post-processing layer is composed of 2 neurons, and the output of the post-processing layer is expressed as:

[0102]

[0103] Among them, y1(t) is the upper output value, y2(t) is the lower output value, and a ij (t) is the weight coefficient of the jth neuron of the ith input, b j (t) is the bias of the jth neuron with the ith input;

[0104] Output layer: It consists of 1 neuron. The output of the output layer is expressed as:

[0105] y(t)=(1-q(t))y1(t)+q(t)y2(t) (8)

[0106] Among them, y(t) is the predicted output at time t, and q(t) is the proportional coefficient at time t;

[0107] 3) Optimize interval type-2 fuzzy neural network parameters:

[0108] ① Set the sliding window length to 16, the leftmost starting element of the sliding window is the first group of samples, that is, t = 1; the initial number of iterations g = 1, the maximum number of iterations is 50; the upper center n of the first iteration at the initial moment ij (1) Randomly select a value in [0,1], the lower center m ij (1) = 0.8n ij (1), width σ ij (1) = 1, weight coefficient a ij (1) = 0.5, deviation b ij (1) = 0, proportionality coefficient q(1) = 0.3;

[0109] ②Set the input matrix D of the sliding window at time t t =[d(t),d(t+1),…,d(t+15)], d(t)=[u1(t),u2(t),…,u7(t)] T ; Set the output matrix of the sliding window at time t

[0110] ③ Initially select the input matrix Q at time t t The first column, that is, s = 1;

[0111] ④ Set D at time t t The sth column is used as the input of the detection model, and the output y of the corresponding model is obtained using formulas (2)-(8) s (t);

[0112] ⑤The parameter update formula is:

[0113]

[0114] in, is the parameter vector of the sth interval type-2 fuzzy neural network, is the parameter vector of the s+1th interval type-2 fuzzy neural network, is the partial derivative of the sth parameter vector;

[0115] ⑥If s<16, increase s by 1 and go to step ④; if the number of iterations s≥16, go to step ⑦;

[0116] ⑦ If the number of iterations g < 50, g increases by 1 and goes to step ③; if the number of iterations g ≥ 50, go to step 4);

[0117] 4) Optimizing the output of interval type-2 fuzzy neural network based on adaptive weighted compensation mechanism:

[0118] Using the updated interval type-2 fuzzy neural network, d(t), d(t+1), …, d(t+15) are used as network inputs in sequence, and the model outputs y(t), y(t+1), …, y(t+15) are obtained using formulas (2)-(8); the compensation error is:

[0119]

[0120] Among them, λ(t+o) is the correction factor, and λ=[λ(t),…,λ(t+15)] is updated as

[0121]

[0122] in,

[0123] Take d(t + 16) as the input of the detection model, and use formulas (2)-(8) to obtain the output y(t + 16) of the model; optimize the output of the interval type-2 fuzzy neural network based on the adaptive weighted compensation mechanism h(t + 16) = y(t + 16) + ê(t + 16), which is the normalized credibility level; perform anti-normalization processing on the normalized credibility level h(t + 16) to obtain the predicted credibility level.

[0124] If h(t + 16) > 0.5, it is considered that the credibility level corresponding to the sample is high; if -0.5 < h(t + 16) < 0.5, it is considered that the credibility level corresponding to the sample is medium; if h(t + 16) < -0.5, it is considered that the credibility level corresponding to the sample is low.

[0125] If t ≤ Z - 16, then t is incremented by 1, and it turns to step ② in 3); otherwise, the intelligent evaluation of the audit terminal system behavior ends.

[0126] In summary, the intelligent evaluation method for the audit terminal system behavior based on the interval type-2 fuzzy neural network in this application introduces a multi-variable dynamic feature integration mechanism. The system simultaneously analyzes multiple key variables (such as access time, operation frequency, number of abnormal warnings, etc.) during behavior evaluation and conducts a comprehensive evaluation according to the variable weights. In addition, it also has a fuzzy rule self-adaptive adjustment function, which can optimize the inference rules of the fuzzy neural network in real time according to different operation behaviors and environmental conditions, so as to more accurately describe the non-linear fluctuations in complex audit data.

[0127] Embodiment III

[0128] As Figure 2 shown, a schematic diagram of a behavior intelligent evaluation device based on the interval type-2 fuzzy neural network in an embodiment of this application includes:

[0129] A processing module, configured to obtain a sample data set of multiple dynamic behavior feature variables that affect the credibility level, and obtain each normalized dynamic behavior feature variable by performing normalization processing on the sample data set of each dynamic behavior feature variable.

[0130] A construction and optimization module, configured to construct an interval type-2 fuzzy neural network model, and optimize the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior feature variable to obtain an optimized interval type-2 fuzzy neural network model.

[0131] A compensation module, configured to compensate the output of the optimized interval type-2 fuzzy neural network model based on the adaptive weighted compensation mechanism to obtain the compensation error of the output result of the optimized interval type-2 fuzzy neural network model.

[0132] The behavior evaluation module is used to obtain sample data of the dynamic behavior characteristic variable to be evaluated, and obtain the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated by using the optimized interval type-II fuzzy neural network model and the compensation error.

[0133] In summary, the present application introduces a multivariable dynamic feature integration mechanism to the intelligent evaluation device for the audit terminal system behavior based on the interval type-2 fuzzy neural network, analyzes multiple key variables (such as access time, operation frequency, number of abnormal warnings, etc.) simultaneously in the behavior evaluation, and conducts a comprehensive evaluation based on the variable weights. The interval type-2 fuzzy neural network is used to characterize the nonlinear relationship of the audit data, and an adaptive weighted compensation mechanism is designed to improve the accuracy of the audit terminal system behavior evaluation. In addition, a fuzzy rule adaptive adjustment function is built in, which can optimize the inference rules of the fuzzy neural network in real time according to different operating behaviors and environmental conditions, so as to more accurately describe the nonlinear fluctuations in complex audit data. It also has an adaptive weighted compensation function, which dynamically adjusts parameters to correct deviations and ensure the stability of the evaluation results. If the accuracy of the evaluation results cannot be restored through parameter adjustment, the system will further enable the model self-calibration function to recalculate the influence weights of each feature to ensure the accuracy of the evaluation.

[0134] An electronic device according to an embodiment of the present application includes: a memory; a processor; and a computer program;

[0135] Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the steps of the behavior intelligent evaluation method based on interval type-2 fuzzy neural network.

[0136] A computer-readable storage medium according to an embodiment of the present application is characterized in that a computer program is stored thereon; the computer program is executed by a processor to implement the steps of the behavioral intelligent evaluation method based on interval type-2 fuzzy neural network.

[0137] This application significantly improves the intelligence level, accuracy and applicability of the audit system in complex environments. The method can automatically adapt to different scenarios and conditions, providing an efficient, accurate and reliable solution for the behavioral evaluation of the audit terminal system.

[0138] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, devices, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0139] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (apparatus), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0140] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0141] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0142] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0143] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A behavior intelligent evaluation method based on interval type-2 fuzzy neural network, characterized in that: include: Acquire a plurality of sample data sets of dynamic behavior characteristic variables that affect the credibility level, and obtain each normalized dynamic behavior characteristic variable by normalizing the sample data set of each dynamic behavior characteristic variable; Constructing an interval type-2 fuzzy neural network model, and optimizing the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior characteristic variable to obtain an optimized interval type-2 fuzzy neural network model; Compensating the output of the optimized interval type-2 fuzzy neural network model based on an adaptive weighted compensation mechanism to obtain a compensation error of the output result of the optimized interval type-2 fuzzy neural network model; Sample data of the dynamic behavior characteristic variable to be evaluated is obtained, and the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated is obtained by using the optimized interval type-2 fuzzy neural network model and the compensation error.

2. The method according to claim 1, characterized in that The multiple dynamic behavior characteristic variables affecting the trust level include access time, access failure time, task execution time, task stop time, access stay time, access times and system warning times.

3. The method according to claim 2, characterized in that The interval type-2 fuzzy neural network model includes an input layer, a membership layer, an activation layer, a consequent layer and an output layer from left to right; The input layer is composed of 7 neurons, and the input variables are normalized dynamic behavior characteristic variables; The membership layer consists of 5 neurons, and the output of each neuron is: Among them, μ 1ij (t) is the upper bound output of the jth neuron with the ith input at time t, μ 2ij (t) is the lower bound output of the jth neuron with the ith input at time t, n ij (t) is the upper center of the i-th input to the j-th neuron at time t, m ij (t) is the lower center of the i-th input to the j-th neuron at time t, σ ij (t) is the width of the jth neuron of the ith input at time t, belonging to the neurons in the layer j = 1,…,5; u i (t) is the normalized dynamic behavior characteristic variable of the i-th input at time t.

4. The method according to claim 3, characterized in that The activation layer consists of 5 neurons, and the upper and lower bound outputs of the activation layer are expressed as: Among them, f 1l (t) is the upper bound output of the lth neuron at time t, f 2l (t) is the lower bound output of the lth neuron at time t, and the neurons in the activation layer are l=1,…,5; The subsequent layer consists of 2 neurons, and the output of the subsequent layer is expressed as: Among them, y1(t) is the upper output value of the subsequent layer at time t, y2(t) is the lower output value of the subsequent layer at time t, and a ij (t) is the weight coefficient of the i-th input to the j-th neuron at time t, b ij (t) is the bias of the ith input to the jth neuron at time t; The output layer consists of 1 neuron, and the output of the output layer is expressed as: y(t)=(1-q(t))y1(t)+q(t)y2(t) Among them, y(t) is the predicted output at time t, and q(t) is the proportional coefficient at time t.

5. The method according to claim 4, characterized in that The optimized interval type-2 fuzzy neural network model includes: Among them, Ω s (t) is the parameter vector of the sth interval type-2 fuzzy neural network at time t, Ω s+1 (t) is the s+1th interval type-2 fuzzy neural network parameter vector at time t, is the partial derivative of the sth parameter vector; y s (t) is the predicted output result of the sth sample at time t; is the true credibility of the system behavior at time t+s-1.

6. The method according to claim 5, characterized in that The compensation error of the output result of the optimized interval type-2 fuzzy neural network model includes: in, is the compensation error at time t+16; λ(t+o) is the correction factor at time t+o, is the true credibility of the system behavior at time t+o, y(t+o) is the predicted output at time t+o, and o is any time.

7. The method according to claim 6, characterized in that The method of obtaining the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated by using the optimized interval type-2 fuzzy neural network model and the compensation error includes: By inputting the sample data of the dynamic behavior characteristic variable to be evaluated into the optimized interval type-2 fuzzy neural network model, a prediction output result is obtained; By adding the prediction output result and the compensation error, a normalized credibility level of the sample data to be evaluated is obtained; The predicted credibility level is obtained by performing a denormalization process on the normalized credibility level of the sample data to be evaluated, and the predicted credibility level is used as the evaluation result of the dynamic behavior characteristic variable to be evaluated.

8. A behavior intelligent evaluation device based on interval type-2 fuzzy neural network, characterized in that: include: A processing module, used for obtaining a plurality of sample data sets of dynamic behavior characteristic variables affecting the credibility level, and performing normalization processing on the sample data sets of each dynamic behavior characteristic variable to obtain each normalized dynamic behavior characteristic variable; A construction and optimization module, used for constructing an interval type-2 fuzzy neural network model, and optimizing the parameters in the interval type-2 fuzzy neural network model by using each normalized dynamic behavior characteristic variable to obtain an optimized interval type-2 fuzzy neural network model; A compensation module, used for compensating the output of the optimized interval type-2 fuzzy neural network model based on an adaptive weighted compensation mechanism to obtain a compensation error of the output result of the optimized interval type-2 fuzzy neural network model; The behavior evaluation module is used to obtain sample data of the dynamic behavior characteristic variable to be evaluated, and obtain the behavior evaluation result of the dynamic behavior characteristic variable to be evaluated by using the optimized interval type-II fuzzy neural network model and the compensation error.

9. An electronic device, characterized in that: include: Memory; processor; and computer programs; Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the steps of a behavior intelligent evaluation method based on interval type-2 fuzzy neural network as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon; the computer program is executed by a processor to implement the steps of a behavior intelligent evaluation method based on an interval type-2 fuzzy neural network as described in any one of claims 1 to 7.

Images