CN115455824A - Process simulation method and system based on cross attribute influence - Google Patents

Abstract

The invention discloses a flow simulation method and a system based on cross attribute influence, belonging to the technical field of simulation test, wherein the simulation method comprises the following specific steps: (1) collecting each data information table in the data flow; (2) constructing a simulation neural model for simulation evaluation; (3) acquiring a cross attribute table and verifying data; (4) recovering the server storage data periodically; according to the method, the corresponding simulation model is constructed through the convolutional neural network, the data analysis efficiency in each data flow can be improved, meanwhile, the performance evaluation and feedback are carried out on the simulation model in real time, managers can conveniently check the operation performance of the simulation model, the analysis difficulty of the managers is reduced, the stored data in the server can be stabilized within a specified threshold value, the reduction of data transmission efficiency caused by excessive redundant data is avoided, the stability of data storage is improved, the steps of manually clearing the data by the managers are simplified, and the working efficiency of the managers is improved.

Description

Process simulation method and system based on cross attribute influence

Technical Field

The invention relates to the technical field of simulation tests, in particular to a process simulation method and system based on cross attribute influence.

Background

The data flow is the whole process of data acquisition, input, processing and output. The method comprises the steps that after information original data are collected, the information original data are input into a computer system to be subjected to mode or statistical operation, or a special program is compiled according to special requirements of a user to process and process the data, then result data are output, transmission of cross attributes in a data flow has large influence, so that in order to enable workers to check the influence program of the cross attributes more intuitively, simulation testing becomes one of important verification means, the data flow is specified by simulating real use environment, software configuration and testing to a real use state of software, and the workers can conveniently analyze and judge each data flow;

the existing flow simulation method and system based on cross attribute influence have low data analysis efficiency and are inconvenient for managers to analyze the operation performance of a simulation model; in addition, the existing flow simulation method and system based on cross attribute influence are easy to reduce data transmission efficiency and poor stability of data storage due to excessive redundant data; therefore, a flow simulation method and system based on cross attribute influence are provided.

Disclosure of Invention

The invention aims to solve the defects in the prior art, and provides a flow simulation method and system based on cross attribute influence.

In order to achieve the purpose, the invention adopts the following technical scheme:

a flow simulation method based on cross attribute influence comprises the following specific steps:

(1) Collecting each data information table in the data flow;

(2) Constructing a simulation neural model for simulation evaluation;

(3) Acquiring a cross attribute table and verifying data;

(4) And recovering the server storage data periodically.

As a further scheme of the invention, the specific construction process of the simulated neural model in the step (2) is as follows:

the method comprises the following steps: the method comprises the following steps that a worker converts non-binary data in each data flow into binary data through a computer encoder, and then converts a detection interval of each group of data into an interval from 0 to 1 through a Min-Max normalization method;

step two: extracting characteristic information of each group of data, performing characteristic dimension reduction processing on the characteristic information, and converting the characteristic information subjected to dimension reduction processing into a sample characteristic diagram of an image format which is default by a system or set manually;

step three: marking the sample characteristic diagram, dividing the sample characteristic diagram into a training set and a testing set, then constructing a convolutional neural network, performing learning training on the training set through input, convolution, pooling, full connection and output to obtain a simulated neural model, then testing the simulated neural model by using the testing set, stopping training if the testing accuracy meets an expected value, and continuing testing the simulated neural model if the testing accuracy does not meet the expected value.

As a further scheme of the invention, the specific calculation formula of Min-Max normalization in the first step is as follows:

in the formula, x _new Representing the normalized data; x represents characteristic information of the data; x is a radical of a fluorine atom _max A maximum value representing the characteristic information; x is the number of _min A minimum value representing the characteristic information;

step one, the specific calculation formula of the feature dimension reduction is as follows:

in the formula, CV represents a standard deviation of characteristic data; σ represents the mean of the feature data; mu represents the variance coefficient of the characteristic data, if the variance coefficient is larger, the significance is higher, otherwise, the significance is not higher, and the data are eliminated.

As a further scheme of the invention, the simulation evaluation of the simulated neural model in the step (2) comprises the following specific steps:

the first step is as follows: the simulation neural model receives each data information table, simulates data transmission among terminal devices, and simultaneously leads data in each data information table into a simulated transmission flow to generate a related cross attribute table and records the transmission efficiency of each data;

the second step: and then, performing loss calculation on the simulated neural model by adopting a focus loss function, performing performance evaluation on the simulated neural model according to the calculated loss value, namely performing accuracy rate, detection rate and false alarm rate evaluation, and feeding the evaluation result back to a manager for checking.

As a further aspect of the present invention, in the second step, the specific calculation formula of the focus loss function is as follows:

FL(pi)＝-α(1-pi) ^γ log(pi) (3)

in the formula, pi represents a predicted value, α represents a weighting factor, and γ represents a focusing parameter.

As a further scheme of the invention, the data verification in the step (3) comprises the following specific steps:

s1: extracting bridging data in the cross attribute table, distributing a group of weighted values for each group of bridging data, accumulating the weighted values of ownership to be 1, and then extracting main data and subdata related by the bridging data from the data information table;

s2: and checking each group of extracted main data and subdata with each group of data in the cross attribute table, if missing data exists, supplementing according to corresponding bridging data, and if redundant data exists, removing the redundant data from the cross data table.

As a further scheme of the present invention, the specific recovery steps of the stored data in step (3) are as follows:

p1: the management platform detects the server storage data in real time, when the server storage data reach a specified threshold value, the management platform sends a recovery instruction to the data recovery module, and then the data recovery module periodically calculates and updates the recovery rate of the server storage data according to a system default or manually set cycle time value;

p2: and feeding back the updated collection value to the management platform for the management personnel to check, then extracting the storage data from old to new by each group of servers, collecting the storage data of each group according to the calculated recovery rate, and feeding back the collection information to the management platform.

A flow simulation system based on cross attribute influence comprises a management platform, a data acquisition module, a transmission simulation module, a simulation optimizer, a data verification module, a server and a data recoverer;

the management platform is used for receiving data fed back by each submodule and feeding back the data to a manager for checking;

the data acquisition module is used for acquiring various groups of data in the data flow;

the transmission simulation module is used for constructing a simulation neural model for transmission simulation and feeding back a simulation result to a manager for checking;

the simulation optimizer is used for receiving an optimization instruction issued by the management platform and calling past data from the server to optimize parameters of the transmission simulation module;

the data verification module is used for verifying whether the data in the cross attribute table is correct or not;

the server is used for storing simulation results;

the data recoverer is used for performing data recovery on data stored in the server.

Compared with the prior art, the invention has the beneficial effects that:

1. compared with the conventional simulation method, the flow simulation method based on the cross attribute influence comprises the steps of performing symbol value conversion processing on each data through a computer encoder, then performing normalization and feature dimension reduction processing on each converted data, converting the processed data into a sample feature map and marking the sample feature map, dividing the sample feature map into a training set and a test set, constructing a convolutional neural network, performing learning training on the training set to obtain a simulated neural model, then performing testing through the test set to obtain the simulated neural model meeting an expected value, then receiving each data information table by the simulated neural model, simulating data transmission among terminal devices, recording data transmission efficiency, performing loss calculation on the simulated neural model by adopting a focus loss function, performing performance evaluation on the simulated neural model according to the calculated loss value, namely performing performance evaluation, detection rate and false alarm rate evaluation, feeding an evaluation result back to a manager for checking, constructing a corresponding simulation model through the convolutional neural network, improving the data analysis efficiency in each data flow, performing performance evaluation and feedback on the simulation model in real time, and facilitating the manager to check the simulation analysis model, and reducing the difficulty of the operation of the management manager;

2. the method comprises the steps of detecting the data stored in the server in real time through the management platform, sending a recovery instruction to the data recovery module by the management platform when the data stored in the server reaches a specified threshold, calculating and updating the recovery rate of the data stored in the server by the data recovery module periodically according to a cycle time value default by a system or manually set, feeding back a collection value updated each time to the management platform for a manager to check, extracting the stored data from old to new by each group of servers, collecting the stored data of each group according to the calculated recovery rate, and feeding back the collected information to the management platform.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.

FIG. 1 is a block diagram of a process simulation method based on cross attribute influence according to the present invention;

fig. 2 is a system block diagram of a process simulation system based on cross attribute influence according to the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.

Example 1

Referring to fig. 1, the present embodiment discloses a process simulation method based on cross attribute influence, which specifically includes the following steps:

and collecting each data information table in the data flow.

And (4) constructing a simulation neural model for simulation evaluation.

Specifically, a worker converts non-binary data in each data flow into binary data through a computer encoder, then converts a detection interval of each group of data into an interval from 0 to 1 through a Min-Max normalization method, then extracts characteristic information of each group of data, performs characteristic dimension reduction processing on the characteristic information, converts the characteristic information after the dimension reduction processing into a sample characteristic diagram of an image format which is default to a system or set manually, marks the sample characteristic diagram, divides the sample characteristic diagram into a training set and a test set, constructs a convolutional neural network, performs learning training on the training set through input, convolution, pooling, full connection and output to obtain a simulated neural model, then uses the test set to test the simulated neural model, stops training if the test accuracy meets an expected value, and continues to test the simulated neural model if the test accuracy does not meet the expected value.

Specifically, the simulation neural model receives each data information table, simulates data transmission among terminal devices, meanwhile, data in each data information table is imported into a simulated transmission flow to generate a relevant cross attribute table, data transmission efficiency is recorded, loss calculation is carried out on the simulation neural model by adopting a focus loss function, performance evaluation is carried out on the simulation neural model according to a calculated loss value, namely accuracy, detection rate and false alarm rate evaluation are carried out, and an evaluation result is fed back to a manager for checking.

It should be further noted that the specific calculation formula of Min-Max normalization is as follows:

in the formula, x _new Represents the normalized data; x represents characteristic information of the data; x is a radical of a fluorine atom _max A maximum value representing the characteristic information; x is the number of _min A minimum value representing the characteristic information;

the specific calculation formula of the feature dimensionality reduction is as follows:

in the formula, CV represents a standard deviation of characteristic data; σ represents the mean of the feature data; mu represents a variance coefficient of the characteristic data, if the variance coefficient is larger, the variance coefficient is more important, otherwise, the variance coefficient is unimportant, and the data are removed;

the specific calculation formula of the focus loss function is as follows:

FL(pi)＝-α(1-pi) ^γ log(pi) (3)

in the formula, pi represents a predicted value, α represents a weighting factor, and γ represents a focusing parameter.

And acquiring a cross attribute table and performing data verification.

Specifically, the computer extracts bridging data in the cross attribute table, assigns a group of weight values to each group of bridging data, and the ownership weight values are accumulated to be 1, then extracts main data and subdata associated with the bridging data from the data information table, then checks each extracted group of main data and subdata with each group of data in the cross attribute table, if missing data exists, supplements the bridging data according to the corresponding bridging data, and if redundant data exists, removes the bridging data from the cross data table.

And recovering the server storage data periodically.

Specifically, the management platform detects server storage data in real time, when the server storage data reach a specified threshold value, the management platform sends a recovery instruction to the data recovery module, the data recovery module calculates and updates the recovery rate of the storage data in the server periodically according to a system default or manually set cycle time value, the updated collection value at each time is fed back to the management platform for a manager to check, then each group of servers automatically extracts the storage data from old to new, and then each group of storage data is collected according to the calculated recovery rate and feeds back the collection information to the management platform.

Example 2

Referring to fig. 2, the embodiment discloses a flow simulation system based on cross attribute influence, which includes a management platform, a data acquisition module, a transmission simulation module, a simulation optimizer, a data verification module, a server and a data recoverer;

the management platform is used for receiving data fed back by each submodule and feeding back the data to a manager for checking; the data acquisition module is used for acquiring various groups of data in the data flow.

And the transmission simulation module is used for constructing a simulation neural model to carry out transmission simulation, and feeding back a simulation result to a manager for viewing.

The simulation optimizer is used for receiving an optimization instruction issued by the management platform and calling past data from the server to optimize parameters of the transmission simulation module; the data verification module is used for verifying whether the data in the cross attribute table is correct.

The server is used for storing the simulation result; the data recoverer is used for performing data recovery on data stored in the server.

Claims (8)

1. A flow simulation method based on cross attribute influence is characterized by comprising the following specific steps:

(1) Collecting each data information table in the data flow;

(2) Constructing a simulation neural model for simulation evaluation;

(3) Acquiring a cross attribute table and verifying data;

(4) And recovering the server storage data periodically.

2. The process simulation method based on cross attribute influence according to claim 1, wherein the process of specifically constructing the simulated neural model in the step (2) is as follows:

the method comprises the following steps: the method comprises the following steps that a worker converts non-binary data in each data flow into binary data through a computer encoder, and then converts a detection interval of each group of data into an interval from 0 to 1 through a Min-Max normalization method;

step two: extracting characteristic information of each group of data, performing characteristic dimension reduction processing on the characteristic information, and converting the characteristic information after dimension reduction processing into a sample characteristic diagram of an image format which is default by a system or set manually;

step three: marking the sample characteristic diagram, dividing the sample characteristic diagram into a training set and a testing set, then constructing a convolutional neural network, performing learning training on the training set through input, convolution, pooling, full connection and output to obtain a simulated neural model, then testing the simulated neural model by using the testing set, stopping training if the testing accuracy meets an expected value, and continuing testing the simulated neural model if the testing accuracy does not meet the expected value.

3. The process simulation method based on cross attribute influence according to claim 2, wherein the Min-Max normalization specific calculation formula in step one is as follows:

in the formula, x _new Represents the normalized data; x representsCharacteristic information of the data;

x _max a maximum value representing the characteristic information;

x _min a minimum value representing the characteristic information;

step one, the specific calculation formula of the feature dimension reduction is as follows:

in the formula, CV represents a standard deviation of characteristic data; σ represents the mean of the feature data; mu represents the variance coefficient of the characteristic data, if the variance coefficient is larger, the importance is shown, otherwise, the importance is shown, and the elimination is carried out.

4. The process simulation method based on cross attribute influence as claimed in claim 2, wherein the simulation neural model simulation evaluation in step (2) specifically comprises the following steps:

the first step is as follows: the simulation neural model receives each data information table, simulates data transmission among terminal devices, and simultaneously leads data in each data information table into a simulated transmission flow to generate a related cross attribute table and records the transmission efficiency of each data;

the second step is that: and then, performing loss calculation on the simulated neural model by adopting a focus loss function, performing performance evaluation on the simulated neural model according to the calculated loss value, namely performing accuracy rate, detection rate and false alarm rate evaluation, and feeding the evaluation result back to a manager for checking.

5. The process simulation method based on cross attribute influence according to claim 4, wherein the focus loss function in the second step is specifically calculated by the following formula:

FL(pi)＝-α(1-pi) ^γ log(pi) (3)

in the formula, pi represents a prediction value, α represents a weighting factor, and γ represents a focus parameter.

6. The cross attribute influence-based process simulation method of claim 4, wherein the data verification in step (3) comprises the following specific steps:

s1: extracting bridging data in the cross attribute table, distributing a group of weighted values for each group of bridging data, accumulating the weighted values of ownership to obtain a value of 1, and then extracting main data and subdata related by the bridging data from the data information table;

s2: and checking each extracted group of main data and sub data with each group of data in the cross attribute table, if missing data exists, supplementing according to corresponding bridging data, and if redundant data exists, removing the redundant data from the cross data table.

7. The process simulation method based on cross attribute influence according to claim 4, wherein the specific recovery step of the stored data in the step (3) is as follows:

p1: the management platform detects the server storage data in real time, sends a recovery instruction to the data recovery module when the server storage data reach a specified threshold value, and then the data recovery module calculates and updates the recovery rate of the storage data in the server periodically according to a system default or manually set cycle time value;

p2: and feeding back the updated collection value to the management platform for the management personnel to check, then extracting the storage data from old to new by each group of servers, collecting the storage data of each group according to the calculated recovery rate, and feeding back the collected information to the management platform.

8. A flow simulation system based on cross attribute influence is characterized by comprising a management platform, a data acquisition module, a transmission simulation module, a simulation optimizer, a data verification module, a server and a data recoverer;

the management platform is used for receiving data fed back by each submodule and feeding back the data to a manager for checking;

the data acquisition module is used for acquiring various groups of data in the data flow;

the transmission simulation module is used for constructing a simulation neural model to carry out transmission simulation, and feeding a simulation result back to a manager for viewing;

the simulation optimizer is used for receiving an optimization instruction issued by the management platform and calling past data from the server to optimize parameters of the transmission simulation module;

the data verification module is used for verifying whether the data in the cross attribute table is correct or not;

the server is used for storing simulation results;

the data recoverer is used for performing data recovery on data stored in the server.

Images