CN103425581A - Software testing method based on learning control model - Google Patents

Software testing method based on learning control model Download PDF

Info

Publication number
CN103425581A
CN103425581A CN2013103492111A CN201310349211A CN103425581A CN 103425581 A CN103425581 A CN 103425581A CN 2013103492111 A CN2013103492111 A CN 2013103492111A CN 201310349211 A CN201310349211 A CN 201310349211A CN 103425581 A CN103425581 A CN 103425581A
Authority
CN
China
Prior art keywords
test
control
learning
software
study
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN2013103492111A
Other languages
Chinese (zh)
Inventor
曹玲玲
张新玲
马旭军
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inspur Electronic Information Industry Co Ltd
Original Assignee
Inspur Electronic Information Industry Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Inspur Electronic Information Industry Co Ltd filed Critical Inspur Electronic Information Industry Co Ltd
Priority to CN2013103492111A priority Critical patent/CN103425581A/en
Publication of CN103425581A publication Critical patent/CN103425581A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Debugging And Monitoring (AREA)

Abstract

The invention provides a software testing method based on a learning control model. According to the software testing method based on the learning control model, a learning control strategy is that based on strategy control, when complete approximation to a test expected value is realized, the condition that a parameter identification result is converged on a testing system parameter value is shown, then, the parameter characteristics of a tested system are learned, when another output state is required to be learned, an expected output can be coincided substantially after once learning, and then, the intelligence of learning control is embodied sufficiently, so that at regression testing occasions, the significance of the learning control is that system learning is carried out firstly, learned test scripts are directly added to a system during actual operation, then, a quick test on test cases can be realized in one time, and the practicability of the learning control is embodied sufficiently.

Description

A kind of method for testing software based on learning control model
Technical field
The present invention relates to the software test field, be specifically related to a kind of Software Testing Models of controlling based on study.
Background technology
In software test case research at present, be the angle research software automated testing platform from through engineering approaches on the one hand, automatically produce the problems such as test case script and test script path accessibility based on platform; Be the given test target of research on the other hand, how the devise optimum test use cases is realized the software automated testing target.Domestic and international many scholar's research the On The Choice of different Test Strategies in branch's coverage test, describe by compound Poisson model the branch that different Test Strategies produce and cover number, balanced various Test Strategy efficiency has provided corresponding stopping criterion; Use control theory that software test etc. is converted to a control problem, to software test, use the methods such as markov control method, adaptive control using tested software as controlled device, using Strategy of Software as controller, and can adjust online Strategy of Software.But these researchs only are optimized for the test script use-case, do not consider that test process is a continuous regression process, in each regression test, how the Continuous optimization test case is a good problem to study.Studied under tested software dynamic state of parameters change condition, how to regulate Test Strategy, the devise optimum software test case, fast detecting is also got rid of software defect, a kind of test macro state transition model has been proposed, Fast Learning software process defect in the regression test process, accelerated to optimize whole test regression process.
Summary of the invention
The purpose of this invention is to provide a kind of method for testing software based on learning control model.
The objective of the invention is to realize in the following manner, so-called study is controlled and is referred to for repeatable controlled device, utilize the previous control experience of control system, seek a desirable input characteristic curve according to actual output and the wanted signal of measuring system, make controlled device produce the motion of expectation, the process of seeking is exactly that quilt is feared to the process that object is done repetition training, the process that namely iterative learning is controlled, study is controlled and simply is described as with mathematical linguistics: in limited time domain
Figure 2013103492111100002DEST_PATH_IMAGE001
in, provide the controlled device Expected Response
Figure 480803DEST_PATH_IMAGE002
,
Figure 2013103492111100002DEST_PATH_IMAGE003
, seek that certain is given
Figure 765154DEST_PATH_IMAGE004
, make its response
Figure DEST_PATH_IMAGE005
in some sense than make moderate progress, wherein
Figure 877783DEST_PATH_IMAGE006
for finding number of times, if
Figure DEST_PATH_IMAGE007
the time,
Figure 246316DEST_PATH_IMAGE008
, claim iterative learning to control convergence,
If with
Figure DEST_PATH_IMAGE009
Mean software test defect constantly, at different time points software systems desired output , real output value is ,
Figure DEST_PATH_IMAGE011
Mean input parameter state variable in each test process,
Figure 105185DEST_PATH_IMAGE012
Meaning that per moment is chosen acts on control strategy on software systems, considers that software test is one and constantly returns the test process moved in circles, and the
Figure 141274DEST_PATH_IMAGE006
Each time point of inferior recurrence
Figure 954378DEST_PATH_IMAGE014
Upper output
Figure DEST_PATH_IMAGE015
And action
Figure 75918DEST_PATH_IMAGE016
The capital impact is next
Figure DEST_PATH_IMAGE017
Time point output
Figure 383403DEST_PATH_IMAGE018
, whole test process can be converted to an iteration control process, and whole test process is described by following state-transition matrix:
Figure DEST_PATH_IMAGE019
Here
Figure 539577DEST_PATH_IMAGE020
Be respectively
Figure DEST_PATH_IMAGE021
Dimension test input state vector and
Figure 526513DEST_PATH_IMAGE022
Dimension is controlled vector,
Figure DEST_PATH_IMAGE023
For Dimension external testing Uncertain Stochastic is disturbed, and considers that software test is a regression test process, and it is all identical returning each time program process, so system matrix Strict satisfied:
Figure 980945DEST_PATH_IMAGE026
Here
Figure DEST_PATH_IMAGE027
Be a learning time,
Figure 991626DEST_PATH_IMAGE028
For positive integer, . whole like this software test procedure is converted into a standard state transition matrix space problem;
Learning control strategy is on the basis of policy control, when realizing the test expectation value is approached fully, the identification result that shows parameter has converged on the test macro parameter value, at this moment it has learned the system under test (SUT) parameter characteristic, while allowing it learn another output state, through once can substantially overlapping with desired output after study, at this moment the intelligent of study control fully demonstrated out, but therefore in the regression tested occasion, the meaning that study is controlled is: first system is learnt, in actual moving process, the test script of learning well directly is added on to system afterwards, once can realize that test case tests fast, fully demonstrated the practicality that study is controlled.
The invention has the beneficial effects as follows: the method can adapt to the more and more huger software test of scale, there is the software test automation degree high, reduce the regression test number of times, can in limited test process, as often as possible find software defect, or find more software defect, accelerate two this respects of regression test process.Can choose optimization test case and test resource in the limit test use-case, take full advantage of the detecting information of collecting in the test script operational process, Test Strategy reasonable in design.
The accompanying drawing explanation
Fig. 1 is the structural drawing that software test is converted into control problem;
Fig. 2 is software test iteration control problem conversion figure.
Embodiment
With reference to Figure of description, method of the present invention is described in detail below.
Learning control strategy is on the basis of policy control, when realizing the test expectation value is approached fully, the identification result that shows parameter has converged on the test macro parameter value, at this moment it has learned the system under test (SUT) parameter characteristic, while allowing it learn another output state, through once can substantially overlapping with desired output after study, the intelligent of at this moment study control fully demonstrated out.But therefore in the regression tested occasion, the meaning that study is controlled is: first system is learnt, in actual moving process, the test script learnt well directly is added on to system afterwards, once can realizes that test case tests fast, fully demonstrated the practicality that study is controlled.
Study is controlled and is referred to for repeatable controlled device, utilize the previous control experience of control system, seek a desirable input characteristic curve according to actual output and the wanted signal of measuring system, make controlled device produce the motion of expectation, the process of seeking is exactly that quilt is feared to the process that object is done repetition training, namely a process that iterative learning is controlled.Study is controlled and can simply be described as with mathematical linguistics: in limited time domain
Figure 146533DEST_PATH_IMAGE001
In, provide the controlled device Expected Response
Figure 242665DEST_PATH_IMAGE002
,
Figure 954269DEST_PATH_IMAGE003
, seek that certain is given
Figure 22719DEST_PATH_IMAGE004
, make its response
Figure 161576DEST_PATH_IMAGE005
In some sense than Make moderate progress, wherein For finding number of times.If
Figure 166944DEST_PATH_IMAGE007
The time,
Figure 476703DEST_PATH_IMAGE008
, claim iterative learning to control convergence.
If with
Figure 281848DEST_PATH_IMAGE009
Mean software test defect constantly, at different time points software systems desired output
Figure 272938DEST_PATH_IMAGE002
, real output value is
Figure 112718DEST_PATH_IMAGE010
,
Figure 593377DEST_PATH_IMAGE011
Mean input parameter state variable in each test process,
Figure 151398DEST_PATH_IMAGE012
Meaning that per moment is chosen acts on control strategy on software systems.Consider that software test is one and constantly returns the test process moved in circles, and the
Figure 929867DEST_PATH_IMAGE006
Each time point of inferior recurrence
Figure 889733DEST_PATH_IMAGE014
Upper output
Figure 541294DEST_PATH_IMAGE015
And action
Figure 258714DEST_PATH_IMAGE016
The capital impact is next
Figure 653923DEST_PATH_IMAGE017
Time point output , whole test process can be converted to an iteration control process, and software test iteration control problem transforms, as shown in Figure 2.
Whole test process can be described by following state-transition matrix:
Figure 290758DEST_PATH_IMAGE019
Here
Figure 759390DEST_PATH_IMAGE020
Be respectively
Figure 958291DEST_PATH_IMAGE021
Dimension test input state vector and
Figure 627169DEST_PATH_IMAGE022
Dimension is controlled vector, For
Figure 312546DEST_PATH_IMAGE024
Dimension external testing Uncertain Stochastic is disturbed.Consider that software test is a regression test process, it is all identical returning each time program process, so system matrix Strict satisfied:
Figure 838522DEST_PATH_IMAGE026
Here
Figure 189738DEST_PATH_IMAGE027
Be a learning time,
Figure 431363DEST_PATH_IMAGE028
For positive integer, . whole like this software test procedure is converted into a standard state transition matrix space problem.
Except the described technical characterictic of instructions, be the known technology of those skilled in the art.

Claims (1)

1. the method for testing software based on learning control model, it is characterized in that so-called study control refers to for repeatable controlled device, utilize the previous control experience of control system, seek a desirable input characteristic curve according to actual output and the wanted signal of measuring system, make controlled device produce the motion of expectation, the process of seeking is exactly that quilt is feared to the process that object is done repetition training, the process that namely iterative learning is controlled, study is controlled and simply is described as with mathematical linguistics: in limited time domain
Figure 2013103492111100001DEST_PATH_IMAGE002
in, provide the controlled device Expected Response ,
Figure 2013103492111100001DEST_PATH_IMAGE006
, seek that certain is given , make its response
Figure 2013103492111100001DEST_PATH_IMAGE010
in some sense than
Figure 518997DEST_PATH_IMAGE004
make moderate progress, wherein for finding number of times, if
Figure 2013103492111100001DEST_PATH_IMAGE014
the time,
Figure 2013103492111100001DEST_PATH_IMAGE016
, claim iterative learning to control convergence,
If with Mean software test defect constantly, at different time points software systems desired output
Figure 545728DEST_PATH_IMAGE004
, real output value is
Figure 2013103492111100001DEST_PATH_IMAGE020
,
Figure DEST_PATH_IMAGE022
Mean input parameter state variable in each test process,
Figure DEST_PATH_IMAGE024
Meaning that per moment is chosen acts on control strategy on software systems, considers that software test is one and constantly returns the test process moved in circles, and the Each time point of inferior recurrence
Figure DEST_PATH_IMAGE026
Upper output
Figure DEST_PATH_IMAGE028
And action
Figure DEST_PATH_IMAGE030
The capital impact is next
Figure DEST_PATH_IMAGE032
Time point output
Figure DEST_PATH_IMAGE034
, whole test process can be converted to an iteration control process, and whole test process is described by following state-transition matrix:
Figure DEST_PATH_IMAGE036
Here Be respectively
Figure DEST_PATH_IMAGE040
Dimension test input state vector and
Figure DEST_PATH_IMAGE042
Dimension is controlled vector,
Figure DEST_PATH_IMAGE044
For
Figure DEST_PATH_IMAGE046
Dimension external testing Uncertain Stochastic is disturbed, and considers that software test is a regression test process, and it is all identical returning each time program process, so system matrix
Figure DEST_PATH_IMAGE048
Strict satisfied:
Figure DEST_PATH_IMAGE050
Here
Figure DEST_PATH_IMAGE052
Be a learning time,
Figure DEST_PATH_IMAGE054
For positive integer,
Figure DEST_PATH_IMAGE056
. whole like this software test procedure is converted into a standard state transition matrix space problem;
Learning control strategy is on the basis of policy control, when realizing the test expectation value is approached fully, the identification result that shows parameter has converged on the test macro parameter value, at this moment it has learned the system under test (SUT) parameter characteristic, while allowing it learn another output state, through once can substantially overlapping with desired output after study, at this moment the intelligent of study control fully demonstrated out, but therefore in the regression tested occasion, the meaning that study is controlled is: first system is learnt, in actual moving process, the test script of learning well directly is added on to system afterwards, once can realize that test case tests fast, fully demonstrated the practicality that study is controlled.
CN2013103492111A 2013-08-12 2013-08-12 Software testing method based on learning control model Pending CN103425581A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2013103492111A CN103425581A (en) 2013-08-12 2013-08-12 Software testing method based on learning control model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2013103492111A CN103425581A (en) 2013-08-12 2013-08-12 Software testing method based on learning control model

Publications (1)

Publication Number Publication Date
CN103425581A true CN103425581A (en) 2013-12-04

Family

ID=49650364

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2013103492111A Pending CN103425581A (en) 2013-08-12 2013-08-12 Software testing method based on learning control model

Country Status (1)

Country Link
CN (1) CN103425581A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104899141A (en) * 2015-06-05 2015-09-09 北京航空航天大学 Test case selecting and expanding method facing network application system
CN112395205A (en) * 2020-12-03 2021-02-23 中国兵器工业信息中心 Software testing system and method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908017A (en) * 2010-06-01 2010-12-08 南京大学 Regression test case screening method based on partial multi-coverage
CN101916222A (en) * 2010-08-09 2010-12-15 哈尔滨工程大学 Software testing method based on combination of control flow graph traversal and slice forward traversal
US20130151906A1 (en) * 2011-12-08 2013-06-13 International Business Machines Corporation Analysis of Tests of Software Programs Based on Classification of Failed Test Cases
CN103176895A (en) * 2011-12-22 2013-06-26 阿里巴巴集团控股有限公司 Method and system of regression testing

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101908017A (en) * 2010-06-01 2010-12-08 南京大学 Regression test case screening method based on partial multi-coverage
CN101916222A (en) * 2010-08-09 2010-12-15 哈尔滨工程大学 Software testing method based on combination of control flow graph traversal and slice forward traversal
US20130151906A1 (en) * 2011-12-08 2013-06-13 International Business Machines Corporation Analysis of Tests of Software Programs Based on Classification of Failed Test Cases
CN103176895A (en) * 2011-12-22 2013-06-26 阿里巴巴集团控股有限公司 Method and system of regression testing

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
曾敏，黄颖: "基于自学习控制算法优化软件测试用例的研究", 《计算机工程与应用》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104899141A (en) * 2015-06-05 2015-09-09 北京航空航天大学 Test case selecting and expanding method facing network application system
CN104899141B (en) * 2015-06-05 2017-08-04 北京航空航天大学 A kind of test cases selection and extending method of network-oriented application system
CN112395205A (en) * 2020-12-03 2021-02-23 中国兵器工业信息中心 Software testing system and method
CN112395205B (en) * 2020-12-03 2024-04-26 中国兵器工业信息中心 Software testing system and method

Similar Documents

Publication Publication Date Title
CN112131786B (en) Target detection and distribution method and device based on multi-agent reinforcement learning
Can et al. A new model for software defect prediction using particle swarm optimization and support vector machine
US20130218335A1 (en) Procedural memory learning and robot control
Liang et al. Search-based task planning with learned skill effect models for lifelong robotic manipulation
Zhou et al. RBF-ARX model-based MPC strategies with application to a water tank system
Li et al. The application of immune genetic algorithm in main steam temperature of PID control of BP network
CN103399488B (en) Multiple Model Control Method based on self study
CN103425581A (en) Software testing method based on learning control model
CN107498554A (en) A kind of mechanical arm intelligent independent control method based on dynamic mode
Lishuang et al. Sensor fault diagnosis based on least squares support vector machine online prediction
Qian et al. A learning strategy based partial feedback linearization control method for an offshore boom crane
Arshad et al. Deep Deterministic Policy Gradient to Regulate Feedback Control Systems Using Reinforcement Learning.
CN104571086A (en) Temperature controller simulation testing method based on transfer function
Raimalwala et al. An upper bound on the error of alignment-based transfer learning between two linear, time-invariant, scalar systems
CN105824239A (en) Smooth tracking eye movement adaptive tracking controller with fixation difference constraint and control method thereof
Li et al. Accurate position control of a servo-hydraulic test cylinder by iterative learning control technique
Noueili et al. Optimal iterative learning control for a class of non-minimum phase systems
CN112363398B (en) Finite-time sliding-mode control system and method for bridge crane system under control input limitation
CN104102131A (en) Infinite time-domain optimization batch process linear secondary fault-tolerant control method
Willems et al. An input-mapping initialization approach to accelerate iterative learning of similar tasks
Zhang et al. Development of a novel wave-force prediction model based on deep machine learning algorithms
Hsu et al. Minimum-phase criterion on sampling time for sampled-data interval systems using genetic algorithms
Rastogi Deep reinforcement learning for bipedal robots
CN109799765A (en) Steering engine electrohydraulic servo system intelligent control method
Su et al. Sensor fault detection with online sparse least squares support vector machine

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20131204

WD01 Invention patent application deemed withdrawn after publication