CN110533150B - Test generation and reuse system and method based on support vector machine regression model - Google Patents

Abstract

The invention discloses a test generation and reuse system and a test generation and reuse method based on a support vector machine regression model, wherein the test generation and reuse system comprises the following steps: the test case generation unit based on the support vector machine regression model comprises a support vector machine regression module and a genetic algorithm module, wherein the support vector machine regression module is used for training a model for simulating and calculating fitness, and the genetic algorithm module is used for referring to the model trained by the support vector machine regression module to generate a test case; the test case reuse unit integrated with the support vector machine regression model comprises a test case reuse module, when a genetic algorithm is utilized to test a program, the support vector machine regression model is trained by taking an adaptation value calculated by a population individual in evolution and a pile inserting method as a sample, test data with higher adaptation degree is queried in a test case set of the program by utilizing the support vector machine regression model, and in the process of introducing the test data into new population iteration, cross operation is carried out on each selected individual and randomly selected referenced individuals with a certain probability.

Description

Test generation and reuse system and method based on support vector machine regression model

Technical Field

The present invention relates to the technical field of software development, and in particular to a test generation and reuse system and method based on a support vector machine regression model.

Background Art

Software testing runs through the entire process of software development and is an indispensable part of software development. Usually, nearly 40% of the time and energy in the software life cycle is spent on software testing. Studies have shown that the cost of software testing is as high as 3 to 5 times the total cost of all other software engineering stages. Effectively improving testing efficiency and reducing the cost of software testing and the labor intensity of test engineers are one of the hot research topics in the software field.

Software testing reuse refers to the reuse of generated test resources in new software testing. When test engineers perform new tests or regression tests, they use or simply modify existing test cases and apply them to the test. The purpose is to use the previously existing test resources multiple times to maximize the results of one generation. It is not necessary to regenerate test resources for each test, thereby improving the efficiency of software testing and the reliability of the software. Test cases are the core resources of software testing, and the reuse of test cases is the key content of the entire software testing reuse.

As a global search method inspired by the biological evolution and genetic variation mechanism in nature, the application of genetic algorithms in software testing has achieved fruitful research results in recent years. Genetic algorithms have the steps of population initialization, individual evaluation, selection operation, crossover operation, mutation operation, and evolution termination condition judgment. The initial population of genetic algorithms is usually generated randomly, and the individual evaluation calculates the fitness value of each individual in the population through the corresponding fitness function. Traditional methods often require the individual to be input into the plug-in program to calculate the individual's fitness. The fitness determines the performance of the individual, and the survival law of the fittest is used to select the individual evolution population.

There are many repetitive tasks in software testing. When testing new programs, new test cases often need to be regenerated. Failure to fully utilize test cases will increase testing costs and reduce the labor efficiency of test engineers. When using genetic algorithms to generate new test cases, traditional methods for calculating individual fitness require a lot of running time.

Summary of the invention

In view of this, it is necessary to provide a method and system for test generation and reuse based on a support vector machine regression model, which can improve the efficiency of test case generation and reduce the workload of software testing.

A test generation and reuse system based on a support vector machine regression model, comprising:

A test case generation unit based on a support vector machine regression model includes a support vector machine regression module and a genetic algorithm module. The support vector machine regression module is used to train a model for simulating and calculating fitness values. The training samples come from the test cases output by the genetic algorithm module. The genetic algorithm module is used to generate test cases by referencing the model trained by the support vector machine regression module.

A test case reuse unit incorporating a support vector machine regression model comprises a test case reuse module, wherein the test case reuse module is used to initialize a predetermined number of individuals when a genetic algorithm is used to test a program, and to train a support vector machine regression model by using individuals of the evolving population and fitness values calculated by a plugging method as samples during population evolution, so that the support vector machine regression algorithm is integrated into the test cases generated by the genetic algorithm module, the trained model is referenced in the generation of the test cases, and individuals selected by the model with fitness values higher than a predetermined value are introduced into the genetic population evolution process, and test data are referenced in the new population iteration process, and each selected individual is cross-operated with a randomly selected reference individual with a predetermined probability, so as to realize the reuse of the test cases.

Furthermore, in the test case generation unit based on the support vector machine regression model, when the genetic algorithm module generates test cases using the genetic algorithm, the population is evolved through operations such as population initialization, selection, crossover, and mutation. The fitness values of individuals in the population are accurate values output by the input plug-in program. The population individuals are input into the support vector machine regression model as samples of the training model for training. When the prediction model training is completed, when the population is evolved again, the trained model is used to predict the fitness values of the population individuals. Other genetic operations of the population are still performed according to the method in the traditional mode. The range of fitness values under the coverage of the target path is set according to the prediction of the training model. If the fitness value of the population is within the interval during the evolution process, the individual needs to be input into the plug-in program for accurate value calculation.

Furthermore, the test case reuse unit integrated with the support vector machine regression model integrates the support vector machine regression module and the genetic algorithm module, and uses the trained support vector machine regression model as a prediction model. After the prediction model is trained, the fitness of the individuals in the population is simulated and calculated by the prediction model; the support vector machine regression model is used to query the test data with a fitness higher than a predetermined value in the test case set of the program; when the number of citations of a referenced individual exceeds a predetermined value, the predetermined value is set to three times, and the referenced individual is removed to avoid falling into a local optimum; if the individual fitness trained by the support vector machine regression model is better, it indicates that the individuals with higher fitness selected by the support vector machine regression model have excellent genes that are beneficial to individual survival, and the referenced individuals carry excellent genes that are beneficial to the survival of the population. Combined with the individuals in the population, the introduction of excellent genes will accelerate the evolution of the population.

和所述特征对应真值

；特征x为测试用例的输入向量，d为其适应度值。Furthermore, the test case generation process requires the use of a predetermined number of samples to train the constructed support vector machine regression model, wherein the samples ( X, d ) contain input features

The true value corresponding to the feature

; Feature x is the input vector of the test case, and d is its fitness value.

，其中a为样本的个数，适应度值

为输入插桩程序得到的适应度真值，得到容量大小为a的样本

；将样本输入支持向量机回归模型，根据样本数据及适应度值拟合每个输入特征的权重

，预测值与输入特征值的关系公式为：Furthermore, the test case is expressed as

, where a is the number of samples and the fitness value

The true fitness value obtained by the input plug-in program is obtained, and a sample with a capacity of a is obtained.

; Input the sample into the support vector machine regression model, and fit the weight of each input feature according to the sample data and fitness value

, the relationship between the predicted value and the input feature value is:

。

.

Furthermore, the training model is a support vector machine regression model that uses the generated test cases and the fitness obtained by the plugging method as samples to train the model during the process of generating test cases using the genetic algorithm. The trained model can simulate and calculate the fitness of the test cases of the program to be tested.

And, a method for test generation and reuse based on a support vector machine regression model, using the test generation and reuse system based on a support vector machine regression model as described in any of the above items, using the support vector machine regression model to predict the value of fitness, and using the support vector machine regression model to find individuals with higher fitness in the test case library of the program to be tested, and reusing the found individuals with higher fitness in the test of the program, including the following steps:

Step 1: Design a training model of a support vector machine for simulating the calculation of the fitness value of the test case. During the program testing process, use the plugging method to calculate the individual fitness value, and input the population individuals and their fitness values into the training model of the support vector machine.

Step 2: Provide a method for simulating fitness calculation using a support vector machine regression model. In the process of generating test cases using a genetic algorithm, use the predetermined population individuals and their fitness as samples to train the support vector machine regression model. In the subsequent population evolution process, use the model to calculate individual fitness instead of the traditional method of calculating fitness by plugging.

Step three, propose a method for using a support vector machine regression model to find test data and reuse it in program testing, using the model trained by the simulation calculation method of individual fitness, and use the model trained by the simulation calculation method of individual fitness to find test data with higher fitness in the test case library of the corresponding program. When using a genetic algorithm to test the program to generate corresponding test cases, the test data is referenced to perform a cross operation with the evolving individuals.

Furthermore, the method for simulating and calculating fitness using a support vector machine regression model comprises the following steps:

Step a, dividing the obtained samples into training samples and prediction samples in a ratio of 8:2;

Step b, calculating the weights of each feature of the training sample, including the risk function R, the Lagrange factor α, and the fitness value of the test case predicted by the kernel mapping method in the presence of errors;

Step c, modifying the weight of each sample and verifying it using the cross method;

Step d: After all training samples are trained, the prediction model training ends;

Step e: After model training is completed, the range of fitness of individuals on the target path is determined using the predicted samples.

Furthermore, the method for generating test cases using a genetic algorithm comprises the following steps:

Step S1, initializing the population;

Step S2, the population individuals are input into the plugging program to obtain the fitness value of each individual, and the individuals and their fitness values are input into the support vector machine model to train the support vector machine regression model;

Step S3, the fitness of each evolving individual is approximated by the trained model;

Step S4, using the instrumentation program to find the exact value of the individuals whose fitness values are within the preset threshold;

Step S5: if the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates, otherwise, go to step S6;

Step S6, perform selection, crossover and mutation operations on individuals, and go to step S3.

Furthermore, the method of using the support vector machine regression model to find test data and reuse it in program testing includes the following steps:

Step (1), using the trained model to select individuals with higher fitness, and setting a range of fitness intervals covering the target path individuals in the support vector machine regression module;

Step (2), encapsulating the selected test data as genetic individuals into a database and introducing it into the genetic process of the genetic algorithm;

Step (3), in the process of evolving the next generation of new population, the population performs a crossover operation with individuals randomly introduced from the database with a certain probability;

Step (4), if the individual is cited more than three times, remove the individual from the database;

Step (5): if the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates; otherwise, go to step (6);

Step (6): Repeat steps (3) and (4) in each new generation of population evolution.

The present invention mainly has the following contributions:

1) Train a model that can simulate and calculate the fitness of test cases. In the process of generating test cases by genetic algorithm, the generated test cases and the fitness obtained by the instrumentation method are used as samples to train the support vector machine regression model. The trained model can simulate and calculate the fitness of the test cases of the program to be tested.

2) Apply the training model to the process of generating test cases by genetic algorithm. When the traditional instrumentation method calculates the fitness value of individuals in the population, it is necessary to run the program, which consumes a lot of time. The support vector machine regression model can simulate the fitness value of individuals according to the individual, without running the program to be tested, reducing the time consumed by the individual fitness evaluation. The value of individual fitness calculated by the model is not an accurate value. Therefore, this software formulates the fitness value interval range of excellent individuals according to the specific situation of the model simulating the individual fitness. The individuals with fitness values in this interval may be the optimal individuals covering the target path, and the fitness of the individuals needs to be accurately calculated. This method uses the instrumentation method to cover the test data of the target path to the minimum extent. The experimental results also show that this method more effectively reduces the time of generating test data and improves the test efficiency.

3) Apply the training model to the reuse of test cases. In terms of test case reuse, the trained test model is used to search for individuals with higher fitness in the test case database of the corresponding program. The individuals carry excellent genes that are beneficial to the survival of the population. In the process of generating test cases using genetic algorithms, the population individuals are combined with the introduced individuals with a certain probability. The experimental results show that this test case generation method can further reduce the time consumption of test case generation and improve the test efficiency of the program.

In the above-mentioned test generation and reuse system and method based on support vector machine regression model, a test case generation and reuse method integrating support vector machine regression model is proposed, and the support vector machine regression model is used to calculate the fitness value. Compared with the neural network model, support vector machine regression shows many unique advantages in solving small sample, nonlinear and high-dimensional pattern recognition. The support vector machine regression model is used to predict the fitness value, and the model is used to find individuals with higher fitness in the test case library of the program to be tested, and reuse them in the test of the program. The efficiency of test case generation is improved by reducing the time consumption of calculating individual fitness and accelerating population evolution.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a technical architecture flow chart of a test generation and reuse method based on a support vector machine regression model according to an embodiment of the present invention.

FIG2 is a flow chart of a test case generation framework of a test generation and reuse method based on a support vector machine regression model according to an embodiment of the present invention.

FIG3 is a flow chart of a test case reuse framework of a test generation and reuse method based on a support vector machine regression model according to an embodiment of the present invention.

FIG. 4 is a software operation interface of support vector machine regression model training and application of the test generation and reuse method based on the support vector machine regression model according to an embodiment of the present invention.

DETAILED DESCRIPTION

This embodiment takes a test generation and reuse system and method based on a support vector machine regression model as an example, and the present invention will be described in detail below in conjunction with specific embodiments and drawings.

Please refer to FIG. 1 , FIG. 2 and FIG. 3 , which illustrate a test generation and reuse system and method based on a support vector machine regression model provided by an embodiment of the present invention.

Support Vector Machine (SVM) mainly includes the definition of the objective function, the processing of noise in the objective function, the further optimization of the objective function and the method of solving nonlinear regression problems. The support vector machine regression model for predicting the fitness of test cases is trained according to the following knowledge of support vector machines.

Support vector machine is based on the VC dimension theory and structural risk minimization principle of statistical learning theory. It is a generalized classifier that performs binary classification of data in a supervised learning manner to obtain the best generalization ability. Support vector machine regression (SVR) is a regression method based on penalty learning. The goal of SVR is to simulate the regression relationship f ( x ) between input x and result y , which is expressed as:

和

，表示超出和未超出ε惩罚区间的情况。并结合参数C来构成最终的风险函数R。SVR的问题即转化为存在误差情况下风险函数R的最小化问题，公式表示为：When using SVR for classification, there are often noise points in the data. In order to correctly classify the noise points, the hyperplane will move closer to the samples of another class, which makes the geometric spacing of the dividing hyperplane smaller and reduces the generalization performance of the model. In order to minimize the impact of noise points on the model, the relaxation factor is introduced.

and

, indicating the situation of exceeding and not exceeding the ε penalty interval. And combined with parameter C to form the final risk function R. The SVR problem is transformed into the minimization problem of the risk function R in the presence of errors, and the formula is expressed as:

This optimization problem is a constrained quadratic programming problem. It can be transformed into a quadratic problem by introducing Lagrange factors α - i and α + i . The constraints are integrated into the objective function through the Lagrange function, and the partial derivatives are traversed to obtain the relationship between the Lagrange factor and w . Therefore, the problem can be clearly expressed with only one function expression, that is, the original parameters w and b are simplified to only the Lagrange factor α . Therefore, the initial regression target can be expressed as:

，来避免在同一特征空间的内积

的计算，最后得到非线性回归表达式的最终形式为：In order to solve the nonlinear regression problem, the kernel mapping method is introduced. The variable x is mapped to a high-dimensional nonlinear space using the transformation function Φ , and the kernel function K is introduced.

, to avoid inner products in the same feature space

The final form of the nonlinear regression expression is:

The selection of kernel function determines the accuracy of the SVR model. This software selects radial basis function (RBF) as the kernel function, which is better for nonlinear data simulation and suitable for predicting the fitness value of the test case.

则称为支持向量。支持向量即为落在ε误差边界上的向量。 支持向量代表了回归模型的特征。除了支持向量以外，其他样本的增删对模型的训练效果影响很小，这使得SVR方法相对其它机器学习方法，需要建模的样本数量更少。这点在特征维度大于数据量时更为明显。Input data set with all parameters satisfying ξ － i , ξ ＋ i ＝ 0

It is called a support vector. The support vector is a vector that falls on the ε error boundary. The support vector represents the characteristics of the regression model. In addition to the support vector, the addition and deletion of other samples have little effect on the training effect of the model, which makes the SVR method require fewer samples to model than other machine learning methods. This is more obvious when the feature dimension is larger than the data volume.

Supervised learning is actually an optimization problem of empirical risk or structural risk function. The risk function measures the quality of model prediction in an average sense, and the quality of each prediction of the model is measured by the loss function. In the hypothesis space F, model f is selected as the decision function. For a given input X , the corresponding output Y is given by f ( X ). The predicted value f ( X ) of this output may have a certain difference with the true value Y. A loss function is used to measure the degree of prediction error. The loss function is recorded as L ( Y, f ( X )). The loss function selected by this software is the square loss function, and its formula is expressed as:

This software combines machine learning methods with genetic algorithms, and uses the support vector machine regression model to simulate the value of individual fitness. It does not need to input every population individual into the plug-in program for fitness calculation, which reduces the time consumed by fitness evaluation and improves test efficiency.

This software includes test case generation and test case reuse.

1. Test case generation based on support vector machine regression model

In the process of using genetic algorithms to generate test cases, the population is evolved by performing operations such as population initialization, selection, crossover, and mutation according to the traditional algorithm. The fitness value of the individual in the population is the accurate value output by the input plug-in program. The individual in the population is used as a sample of the training model and input into the support vector machine regression model for training. When the prediction model training is completed, when the population is evolved again, the trained model is used to predict the fitness value of the individual in the population. Other genetic operations of the population are still performed according to the traditional method. According to the prediction of the training model, the range of fitness values under which the target path is covered is set. If the fitness value of the population is within the range during the evolution process, the individual needs to be input into the plug-in program to calculate the accurate value.

和各个特征对应真值

；特征x为测试用例的输入向量，d为其适应度值。When generating test cases, a certain number of samples are needed to train the constructed support vector machine regression model. A sample ( X, d ) contains the input features

Corresponding to each feature

; Feature x is the input vector of the test case, and d is its fitness value.

，a为样本的个数，这些测试数据的适应度值

是输入插桩程序得到的适应度真值，得到容量大小为a的样本

。这些样本输入支持向量机回归模型会根据样本数据及其适应度值拟合每个输入特征的权重

，预测值与输入特征值的关系用公式表示为：First, generate a certain number of test cases

, a is the number of samples, and the fitness value of these test data is

is the true value of fitness obtained by inputting the instrumentation program, and a sample with a capacity of a is obtained

These sample input support vector machine regression models will fit the weights of each input feature based on the sample data and its fitness value.

, the relationship between the predicted value and the input feature value is expressed as:

。

.

The support vector machine regression model will train a model to predict the fitness value of a test case based on the test case input to a certain program. In order to make the sample data representative, the test cases selected for the training sample are data randomly generated by the genetic algorithm. It should be noted that the capacity of the selected sample should not be too large or too small. If the sample capacity is too small, the accuracy of the trained model is low and it cannot reflect the inherent linear law of the sample; if the capacity is too large, more computing resources will be consumed, it will take longer and reduce the test efficiency.

The test case generation integrated with the support vector machine model is divided into two modules, the support vector machine regression module and the genetic algorithm module. The support vector machine regression module is mainly used to train the model of simulation calculation of fitness size, and the training samples come from the test cases output by the genetic algorithm module. The genetic algorithm module references the model trained by the support vector machine module to generate test cases.

(1) Support vector machine regression module

The test cases generated by the genetic algorithm are used as samples to train the support vector machine regression model. The main steps are as follows:

1) Obtain a certain number of samples and divide them into training samples and prediction samples in a ratio of 8:2;

2) According to formula (2), the objective function is transformed into the minimization problem of the risk function R in the presence of errors;

3) According to formula (3), the Lagrange function is integrated into the objective function as a constraint condition to obtain the relationship between w and the Lagrange factor a ;

4) Formula (4) introduces the kernel mapping method, and uses the kernel function in formula (5) to solve nonlinear high-dimensional problems;

5) As the number of samples increases, the weights are modified according to step 2), step 3) and formula (6), and verified using the cross method. After all samples are trained, the prediction model training ends;

6) After model training is completed, the predicted samples are used to determine the range of fitness of individuals covering the target path.

In the process of generating test cases by genetic algorithm, a support vector machine regression model simulating the fitness of individuals is trained and applied to the subsequent evolution of the population. The genetic algorithm module includes the training of the support vector machine regression model and the method of using the model.

(2) Genetic Algorithm Module

Genetic algorithm is used to generate test cases. The main steps are as follows:

1) Population initialization;

2) The population individuals are input into the plug-in program to obtain the fitness value of each individual, and the individuals and their fitness values are input into the support vector machine model to train the support vector machine regression model;

3) The fitness of each evolving individual is approximated by the trained model;

4) For individuals whose fitness values are within the preset threshold, use the instrumentation program to find their exact values;

5) If the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates, otherwise go to step 6);

6) Perform selection, crossover and mutation operations on individuals and go to step 3).

When using the genetic algorithm to test the program, before the support vector machine regression model is successfully trained, the genetic population is evolved using the fitness calculated by the plugging method. When the population evolves to a certain generation, the model training is successful. In the next population evolution, the training model is used to simulate and calculate the individual fitness. Using the training model instead of the plugging method to calculate the fitness value is intended to reduce the time required for population evolution.

The test case reuse method reuses the support vector machine regression model and applies the training model to the reuse of test cases, aiming to speed up the evolution of individuals covering the target path and improve the test efficiency.

2. Test case reuse with support vector machine regression model

The support vector machine regression algorithm is integrated into the genetic algorithm to generate test cases. The use of its training model cannot speed up the evolution of the population. It is necessary to improve the test efficiency by reducing the time it takes for individuals to run the plug-in program. The reuse of the test case of this software is to refer to the trained model in the generation of test cases, and once again improve the efficiency of test case generation based on the model. That is, the individuals with higher fitness selected by the model are introduced into the process of genetic population evolution to speed up the evolution of the population, thereby further improving the efficiency of test data generation.

The test case reuse unit of the support vector machine regression model includes a test case reuse module. The steps of completing the test case reuse by using the trained support vector machine regression model are as follows:

1) Use the trained model to select individuals with higher fitness. In the support vector machine regression module, a range of fitness that can cover the target path individuals has been established. Individuals with fitness in this range are called individuals with higher fitness or excellent individuals;

2) Encapsulate the selected test data as genetic individuals into a database and introduce it into the genetic process of the genetic algorithm;

3) In the process of evolving the next generation, the population performs a crossover operation with individuals introduced from the database (randomly selected) with a certain probability;

4) In order to avoid local optimality, if an individual is cited more than three times, the individual will be removed from the database;

5) If the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates, otherwise go to step 6);

6) Repeat steps 3) and 4) in each new generation of population evolution.

When using a genetic algorithm to test a program, a certain number of individuals are first initialized. During the population evolution process, the fitness values calculated by the evolving population individuals and the plugging method are used as samples to train the support vector machine regression model. After the prediction model of individual fitness is trained, the fitness of the population individuals is simulated and calculated by the model. In addition, the model is used to query the test data with higher fitness in the test case set of the program. These test data are referenced in the new population iteration process. The referenced individuals are called "reference individuals". Each selected individual crosses with the randomly selected reference individual with a certain probability. In order to maintain the diversity of population genes and avoid falling into the local optimal situation, if the number of references of the reference individual exceeds a certain value (here set to 3), the individual is removed. If the support vector machine regression model simulates the individual fitness well, it means that the model selects individuals with higher fitness with excellent genes that are beneficial to individual survival. The introduction of excellent genes will accelerate the evolution of the population and improve the efficiency of test case generation.

Refer to Fig. 4, which shows the operation interface of support vector machine regression model training and its application, a plug-in prototype of support vector machine regression model training and its application, which serves the main body of the software project and effectively expands and improves the function of the host software. The training of the support vector machine regression prediction model and the process of generating and reusing test cases using the model are implemented in the form of plug-ins, which further simplifies the testing process of the program. Java is selected as the editing language for the development of the plug-in, and the development environment is MyEclipse 2010. The computer configuration is Windows (Intel (R) Core (TM) CPU i5-6500, 3.20GHz, 8.00GB RAM, 64-bit operating system.

The "Program" menu option gives the instrumentation method of the program to be tested, "SVR Model" includes the selection and analysis of the parameters for training the support vector machine regression model, "TestCase" analyzes the role and requirements of the test case database, "Options" provides environment selections such as language and font color size, and "Help" contains instructions for using this plug-in.

When using this plug-in, the buttons on the interface are executed in sequence according to the numbering order. The operation process is as follows:

1) Click the button "1. Select a Program" and select the file containing the source code of the program to be tested in the file selection interface. The file contains the program to be tested after insertion.

2) Click the button “2. Train SVR Model” to train the support vector machine regression fitness prediction model. The text box displays “Successful Training of SVR Model”, indicating that the model training is successful.

3) Click the button "3. Find Test Cases" and select the file where the use case database of the program to be tested is located on the pop-up interface.

4) Click the button "4. Reuse and Generate Test Cases". When the genetic algorithm generates test cases, the individuals in the population use the training model to solve the fitness. The training model will search for test cases with higher fitness in the file and reuse them as excellent individuals in the generation of test cases. The text box under the button outputs the test results.

The present invention mainly has the following contributions:

1) Train a model that can simulate and calculate the fitness of test cases. In the process of generating test cases by genetic algorithm, the generated test cases and the fitness obtained by the instrumentation method are used as samples to train the support vector machine regression model. The trained model can simulate and calculate the fitness of the test cases of the program to be tested.

2) Apply the training model to the process of generating test cases by genetic algorithm. When the traditional instrumentation method calculates the fitness value of individuals in the population, it is necessary to run the program, which consumes a lot of time. The support vector machine regression model can simulate the fitness value of individuals according to the individual, without running the program to be tested, reducing the time consumed by the individual fitness evaluation. The value of individual fitness calculated by the model is not an accurate value. Therefore, this software formulates the fitness value interval range of excellent individuals according to the specific situation of the model simulating the individual fitness. The individuals with fitness values in this interval may be the optimal individuals covering the target path, and the fitness of the individuals needs to be accurately calculated. This method uses the instrumentation method to cover the test data of the target path to the minimum extent. The experimental results also show that this method more effectively reduces the time of generating test data and improves the test efficiency.

3) Apply the training model to the reuse of test cases. In terms of test case reuse, the trained test model is used to search for individuals with higher fitness in the test case database of the corresponding program. The individuals carry excellent genes that are beneficial to the survival of the population. In the process of generating test cases using genetic algorithms, the population individuals are combined with the introduced individuals with a certain probability. The experimental results show that this test case generation method can further reduce the time consumption of test case generation and improve the test efficiency of the program.

In the above-mentioned test generation and reuse system and method based on the support vector machine regression model, a test case generation and reuse method incorporating the support vector machine regression model is proposed, and the support vector machine regression model is used to calculate the fitness value. Compared with the neural network model, the support vector machine regression shows many unique advantages in solving small sample, nonlinear and high-dimensional pattern recognition. The support vector machine regression model is used to predict the fitness value, and the model is used to find individuals with higher fitness in the test case library of the program to be tested, and reuse them in the program test. The efficiency of test case generation is improved by reducing the time consumption of calculating individual fitness and accelerating population evolution.

It should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims (10)

1. A test generation and reuse system based on a support vector machine regression model, characterized by comprising: A test case generation unit based on a support vector machine regression model includes a support vector machine regression module and a genetic algorithm module. The support vector machine regression module is used to train a model for simulating and calculating fitness values. The training samples come from the test cases output by the genetic algorithm module. The genetic algorithm module is used to generate test cases by referencing the model trained by the support vector machine regression module. A test case reuse unit incorporating a support vector machine regression model comprises a test case reuse module, wherein the test case reuse module is used to initialize a predetermined number of individuals when a genetic algorithm is used to test a program, and to train a support vector machine regression model by using individuals of the evolving population and fitness values calculated by a plugging method as samples during population evolution, so that the support vector machine regression algorithm is integrated into the test cases generated by the genetic algorithm module, the trained model is referenced in the generation of the test cases, and individuals selected by the model with fitness values higher than a predetermined value are introduced into the genetic population evolution process, and test data are referenced in the new population iteration process, and each selected individual is cross-operated with a randomly selected reference individual with a predetermined probability, so as to realize the reuse of the test cases. 2. The test generation and reuse system based on the support vector machine regression model as described in claim 1 is characterized in that in the test case generation unit based on the support vector machine regression model, when the genetic algorithm module uses the genetic algorithm to generate test cases, it evolves the population through operations such as population initialization, selection, crossover, and mutation. The fitness value of the individual in the population is the accurate value output by the input plug-in program. The population individuals are input into the support vector machine regression model as samples of the training model for training. When the prediction model training is completed, when the population is evolved again, the trained model is used to predict the fitness value of the population individuals. Other genetic operations of the population are still performed according to the method in the traditional mode. The range of fitness values under the coverage of the target path is set according to the prediction of the training model. If the fitness value of the population is within the interval during the evolution process, the individual needs to be input into the plug-in program for accurate value calculation. 3. The test generation and reuse system based on the support vector machine regression model as described in claim 1 is characterized in that the test case reuse unit integrated with the support vector machine regression model integrates the support vector machine regression module and the genetic algorithm module, and the trained support vector machine regression model is used as a prediction model. After the individual fitness of the prediction model is trained, the fitness of the population individuals is simulated and calculated by the prediction model; the support vector machine regression model is used to query the test data with fitness higher than a predetermined value in the test case set of the program; when the number of citations of a referenced individual exceeds a predetermined value, the predetermined value is set to three times, and the referenced individual is removed to avoid falling into a local optimum; if the individual fitness trained by the support vector machine regression model is better, it indicates that the individual with higher fitness selected by the support vector machine regression model has excellent genes that are beneficial to individual survival, and the referenced individual carries excellent genes that are beneficial to population survival. Combined with the population individuals, the introduction of excellent genes will accelerate the evolution of the population. 4. The test generation and reuse system based on the support vector machine regression model as claimed in claim 1 is characterized in that the test case generation process requires the use of a predetermined number of samples to train the constructed support vector machine regression model, and the samples ( X, d ) contain input features

The true value corresponding to the feature

; Feature x is the input vector of the test case, and d is its fitness value. 5. The test generation and reuse system based on the support vector machine regression model according to claim 4, characterized in that the test case is represented as

, where a is the number of samples and the fitness value

The true fitness value obtained by the input plug-in program is obtained, and a sample with a capacity of a is obtained.

; Input the sample into the support vector machine regression model, and fit the weight of each input feature according to the sample data and fitness value

, the relationship between the predicted value and the input feature value is:

. 6. The test generation and reuse system based on the support vector machine regression model as described in claim 2 is characterized in that the training model is a support vector machine regression model trained by referencing the generated test cases and the fitness obtained by the plug-in method in the process of generating test cases by the genetic algorithm, and the trained model can simulate the calculation of the fitness of the test cases of the program to be tested. 7. A method for test generation and reuse based on a support vector machine regression model, characterized in that the test generation and reuse system based on a support vector machine regression model as described in any one of claims 1 to 6 is used, the support vector machine regression model is used to predict the value of fitness, and the support vector machine regression model is used to search for individuals with higher fitness in the test case library of the program to be tested, and the individuals with higher fitness found are reused in the test of the program, comprising the following steps: Step 1: Design a training model of a support vector machine for simulating the calculation of the fitness value of the test case. During the program testing process, use the plugging method to calculate the individual fitness value, and input the population individuals and their fitness values into the training model of the support vector machine. Step 2: Provide a method for simulating fitness calculation using a support vector machine regression model. In the process of generating test cases using a genetic algorithm, use the predetermined population individuals and their fitness as samples to train the support vector machine regression model. In the subsequent population evolution process, use the model to calculate individual fitness instead of the traditional method of calculating fitness by plugging. Step three, propose a method for using a support vector machine regression model to find test data and reuse it in program testing, using the model trained by the simulation calculation method of individual fitness, and use the model trained by the simulation calculation method of individual fitness to find test data with higher fitness in the test case library of the corresponding program. When using a genetic algorithm to test the program to generate corresponding test cases, reference the test data to perform a cross operation with the evolving individuals. 8. The method for test generation and reuse based on support vector machine regression model according to claim 7, characterized in that the method of simulating fitness calculation using support vector machine regression model comprises the following steps: Step a, dividing the obtained samples into training samples and prediction samples in a ratio of 8:2; Step b, calculating the weights of each feature of the training sample, including the risk function R, the Lagrange factor α, and the fitness value of the test case predicted by the kernel mapping method in the presence of errors; Step c, modifying the weight of each sample and verifying it using the cross method; Step d: After all training samples are trained, the prediction model training ends; Step e: After model training is completed, the range of fitness of individuals on the target path is determined using the predicted samples. 9. The method for test generation and reuse based on support vector machine regression model according to claim 7, characterized in that the method for generating test cases using genetic algorithm comprises the following steps: Step S1, initializing the population; Step S2, the population individuals are input into the plugging program to obtain the fitness value of each individual, and the individuals and their fitness values are input into the support vector machine model to train the support vector machine regression model; Step S3, the fitness of each evolving individual is approximated by the trained model; Step S4, using the instrumentation program to find the exact value of the individuals whose fitness values are within the preset threshold; Step S5: if the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates, otherwise, go to step S6; Step S6, perform selection, crossover and mutation operations on individuals, and go to step S3. 10. The method for test generation and reuse based on support vector machine regression model according to claim 7, characterized in that the method of using support vector machine regression model to find test data and reuse it in program testing comprises the following steps: Step (1), using the trained model to select individuals with higher fitness, and setting a range of fitness intervals covering the target path individuals in the support vector machine regression module; Step (2), encapsulating the selected test data as genetic individuals into a database and introducing it into the genetic process of the genetic algorithm; Step (3), in the process of evolving the next generation of new population, the population performs a crossover operation with individuals randomly introduced from the database with a certain probability; Step (4), if the individual is cited more than three times, remove the individual from the database; Step (5): if the algorithm termination condition is met or the maximum number of iterations is reached, the algorithm terminates; otherwise, go to step (6); Step (6): Repeat steps (3) and (4) in each new generation of population evolution.

Images