CN105844077B - A kind of testability growth test method based on timely correction strategy - Google Patents

Abstract

The invention discloses a test growth test method based on a timely correction strategy, and aims to solve the problem of lack of a systematic test method when carrying out a test growth test based on a timely correction strategy. The technical solution is: first, analyze the initial level of testability of the equipment based on the previous information, then calculate the expected change process of the testability indicators of the equipment during the testability growth test, and draw the test planning curve; then calculate the improved equipment in real time during the test. The changes of the testable indicators; finally, the actual change process of the testable indicators is compared with the expected process, so as to give the control decision of the test process. Utilizing the present invention can solve the problem that the test growth test based on the timely correction strategy lacks the test planning method; it can also effectively monitor the test growth process of the system under test, and control the test process by timely adjusting the test growth test plan, reducing Small trial risks and costs.

Description

A testable growth experiment method based on a timely correction strategy

technical field

The invention relates to a test development test method in the equipment development stage; in particular, a test growth test method based on a timely correction strategy.

Background technique

Testability is the ability of the equipment to correctly detect and isolate faults in a specified time and under a specified environment. It can be measured by a variety of testability indicators, including fault detection rate (FDR), fault isolation rate (FIR), and false alarm rate (FAR). Wait. For large and complex equipment, there are often various problems in the design process of the fault diagnosis system, such as unpredictable faults, test failures, fuzzy groups, inappropriate testability tolerances and thresholds, etc. To make the testability level of equipment meet the design requirements, it is necessary to identify and correct testability design defects through various means, so as to improve the testability index of equipment and achieve testability growth.

GJB2547A "General Requirements for Equipment Testability Work" clearly gives the basic concept of testability growth: in the process of product development and use, by gradually correcting product testability defects, the product testability level is continuously improved, so as to achieve the desired goal. process. Although testability growth can be achieved through model-based design and other methods, testability growth experiment is still the most effective method, especially for various new research systems that have strong requirements for system testability. The testability growth test realizes the growth of the testability index through a series of test-correction-test processes until the testability index requirements of the system are met.

The consumption of time and resources for the test cannot be ignored, especially for the test growth test with many uncertain factors in the test cycle and test conditions. Therefore, there must be scientific test methods to reduce test costs and shorten test time. cycle. However, there are no specific provisions in the relevant published standards on how to carry out test growth experiments. According to the experience of the reliability growth test, the work of the growth test includes four aspects: test planning, test implementation, test tracking and prediction. Specifically, a reasonable test plan is given before the specific implementation of the growth test, and then the fault injection test is carried out according to the test plan, and the exposed design defects are improved, and then the test data is used to evaluate the growth effect in time during the test process, and according to Growth tracking and expected results to adjust the test program in time. Referring to the reliability growth test process, the timely correction strategy is one of the test methods used in the test growth test. In terms of testing growth experiments with timely correction strategies, only Zhao Chenxu et al. A MarkovChain-Based Testability Growth Model With a Cost-Benefit Function from the National University of Defense Technology Test growth model, published in ieee transactions on systems, man, and cybernetics: systems, doi: 10, 1109/tsmc.2015.2437837) and Li Tianmei (T.M.Li et al. The Assessment and Foundation of Bell-Shaped TestabilityGrowth Effort Functions Dependent System Testability Growth Models Based onNHPP is the modeling and application of testability growth models based on NHPP and bell-shaped testability growth performance functions, see http://dx.doi.org/10.1155/2015/613170 for details) at Studies have been carried out in related papers, but these studies have obvious deficiencies: 1) The specific implementation method of the test growth experiment based on the timely correction strategy is not clearly given; 2) The growth models proposed are all parameter models, and the relevant parameters need to be used The test data is estimated, and it is difficult to predict its value before the test is carried out, so it is difficult to use in the planning and formulation of the test growth test plan. In order to carry out test growth experiments more effectively, it is necessary to give a systematic test method for test growth experiments based on timely correction strategies.

Contents of the invention

The technical problem to be solved by the present invention is to provide a test growth test method for the lack of a systematic test method when carrying out a test growth test based on a timely correction strategy. Applying this method can comprehensively consider the characteristics of equipment, historical data, and the technical level of designers before the implementation of the test, and give a reasonable test growth test plan, and during the test, according to the tracking results of the growth effect The program is adjusted in time, so as to strictly control the entire test process and reduce the risk and cost of the test.

The technical scheme of the test growth test method based on the timely correction strategy proposed by the present invention is:

Step 1: Determine the total number of failure modes N of the system under test according to the results of the failure mode, impact and hazard analysis (FMECA) of the system under test (GJB1391-2006 "Guidelines for Failure Mode, Effect and Hazard Analysis");

The second step is to collect pre-test data, including test prediction results, test virtual test data, test test data, expert experience, designer design capabilities, and test index design requirements of the system under test req;

Step 3: According to the results of FMECA and the collected pre-test data, all failure modes are divided into two types according to the updateability of the testability design: updatable failure modes and non-updatable failure modes, and accordingly the system under test is divided into two categories: Subsystems: subsystems composed of updatable failure modes and subsystems composed of non-updatable failure modes, and then use the Bayesian estimation method (Wang Chao, "Testability Experiment and Comprehensive Evaluation Technology Combining Virtual and Reality", National University of Defense Technology Dissertation, 2014 , pp. 81-92) calculate the failure detection probability β possessed by the non-updatable subsystem and the initial failure detection probability α possessed by the updatable subsystem;

Step 4, use formula (1) to calculate the initial failure undetectable probability p(1) of the updateable subsystem;

p(1)=1-α(1)

Step 5, use formula (2) to calculate the testable index growth target value σ of the updateable subsystem,

Where K is the total number of updateable failure modes in the updateable subsystem, λ _i is the failure rate of the i-th failure mode, 1≤i≤N;

Step 6, use formula (3) to calculate the failure undetectable probability p _end that the updateable subsystem needs to achieve at the end of the growth test;

p _end =1-σ (3)

At the same time, based on the historical experience of similar system testability growth, the system designer and test manager determine the average update coefficient f of the updateable failure mode, 0<f≤1, and the fault injection test that allows failure before each design update (GJB2547A "General Requirements for Equipment Testing Work") times m, 1≤m≤K/2;

Step 7, let k=1;

Step 8, use formula (4) to calculate the expected value E(r _{k )} of the number of fault injection tests r k that need to be performed before the kth design improvement,

E(p(k)) is the expected value of the failure undetectable probability p(k) of the updateable subsystem before the k-th design improvement;

Step 9, use formula (5) to calculate the expected value E(q(k)) of the testable index growth coefficient q(k) of the updateable subsystem during the k-th design improvement;

Step 10, use formula (6) to update the expected value E(p(k+1)) of the fault undetectable probability of the updateable subsystem after the kth design improvement;

Step 11, using formula (7) to calculate the expected value E(R _k ) of the total number of fault injection tests R _k required when the test reaches the k-th stage;

Step 12, if E(p(k+1))≤p _req , go to step 13, otherwise set k=k+1, and go to step 8;

Step 13, take E(R _k ) as the horizontal axis and 1-E(p(k)) as the vertical axis to draw the testability growth planning curve shown in Figure 2, and formulate the testability of the updateable subsystem according to the curve growth trial protocol;

Step 14, let k=1;

Step 15: Referring to the simple random sampling method stipulated in GJB2072-94, which is the national military standard of "maintenance test and review", randomly select the failure mode from the updateable failure mode set to carry out the fault injection test until the total number of test failures in the k-th stage is equal to m ;

Step 16: After the fault injection test is over and before the testability design is updated, use the Bayes method to calculate the fault undetectability rate of the updateable subsystem based on the test success or failure data If Go to step 18, otherwise go to step 17;

Step 17, if That is, if the test process deviates from the preset test plan, go to step 2; otherwise, update the design for the exposed testable design defects. After the design update is completed, set k=k+1 and go to step 15;

Step 18, the test is over.

Adopt the present invention can obtain following beneficial effect:

1. Before carrying out the test growth test based on the timely correction strategy, the method proposed by the present invention can be used to formulate a reasonable test growth test plan, obtain the test growth test planning curve, and determine the test growth test plan, thereby solving the current problem based on Timely correction of the strategy's test growth trial lack of a trial planning methodology;

2. During the test development process, the method proposed by the present invention can effectively monitor the test growth process of the system under test, and control the test process by timely adjusting the test growth test plan to reduce test risks and costs.

To sum up, the testing growth method provided by the present invention comprehensively considers the characteristics of the failure mode of the test object, the technical level of the system designer, historical data and experience and other information, and the method provided by the present invention can provide scientific and reasonable guidance based on timely The development of test-growth experiments to correct strategies will help to improve the lack of systematic test methods faced by test-growth experiments in current engineering practice.

Description of drawings

Fig. 1 is the overall flow chart of the present invention;

Fig. 2 is a schematic diagram of the test growth planning curve drawn in the 14th step of the present invention.

Detailed ways

Fig. 1 is an overall flow chart of the present invention, and concrete steps are as follows:

Step 1: Determine the total number of failure modes N of the system under test according to the FMECA results of the system under test;

The second step is to collect pre-test data, including test prediction results, test virtual test data, test test data, expert experience, designer design capabilities, and test index design requirements of the system under test req;

Step 3: Using the testability information obtained in Step 1 and Step 2, the overall failure modes of the system under test are divided into two categories: updateable failure modes and non-updatable failure modes according to the updateability of testability design. The total number of modes is K, the total number of non-updatable failure modes is N-K, and accordingly the system under test is divided into two subsystems: the subsystem composed of updatable failure modes and the subsystem composed of non-updatable failure modes, and then the Bayes estimation method is used to calculate The failure detection probability β of the non-updatable subsystem and the initial failure detection probability α of the updateable subsystem;

Step 4, use formula (1) to calculate the initial failure undetectable probability of the updateable subsystem;

Step 5, use formula (2) to calculate the testable index growth target value σ of the updateable subsystem;

Step 6: use formula (3) to calculate the failure undetectable probability p _end that the updateable subsystem needs to achieve at the end of the growth test; at the same time, collect the testable growth history information of similar systems, and the system designer and test manager can determine that The average update factor for updating failure modes and the number of fault injection trials allowed to fail before each design update;

Step 7, let k=1;

Step 8, use formula (4) to calculate the expected value E(r _k ) of the number of fault injection tests that need to be performed before the k-th design improvement;

Step 9, use formula (5) to calculate the expected value E(q(k)) of the testable index growth coefficient of the updateable subsystem during the k-th design improvement;

Step 10, use formula (6) to update the expected value E(p(k+1)) of the fault undetectable probability of the updateable subsystem after the kth design improvement;

Step 11, use (7) formula to calculate the total number of times R _k of fault injection tests required to carry out the test to the k-th stage;

Step 12, if E(p(k+1))≤p _req , go to step 13, otherwise set k=k+1, and go to step 8;

Step 13, draw a testable growth planning curve, and formulate a testable growth test plan for an updatable subsystem;

Step 14, let k=1;

Step 15, carry out the fault injection test until the total number of test failures in the k-th stage is equal to m;

Step 16: Calculate the failure undetectability rate of the updateable subsystem based on the success or failure data of the test using the Bayes method like Go to step 18, otherwise go to step 17;

Step 17, if Go to step 2, otherwise, update the design for the exposed testable design defects, after the design update is completed, set k=k+1, go to step 15.

Step 18, the test is over.

Fig. 2 is a schematic diagram of a testing growth planning curve drawn by the present invention, wherein the ordinate of the first horizontal line represents the testing index initial value 1-E(p(1)) of the equipment before the testing growth test is carried out, The starting point of the abscissa is 0, and the end point is E(R ₁ ); the ordinate of the other kth, k≥2 horizontal lines is the testing index 1-E(p(k)) of the equipment before the k-th design update, The starting point of the abscissa is E(R _k-1 )+1, and the end point is E(R _k ); the step between two adjacent horizontal lines represents that after the testability design is improved, the testability of the equipment has been improved.

Claims (1)

1. A kind of 1. testability growth test method based on timely correction strategy, it is characterised in that include the following steps：

   1st step determines that system under test (SUT) is had according to system under test (SUT) fault mode, influence and HAZAN, that is, FMECA results Fault mode sum N；

   2nd step collects testability data early period, knows the real situation including testability prediction result, testability virtual test data, testability Test data, expertise, designer's designed capacity and system under test (SUT) testability index design requirement value req；

   3rd step, can according to testability design by all fault modes according to FMECA results and testability data early period collected Update property is divided into two class of renewable fault mode and non-renewable fault mode, and system under test (SUT) is divided into two subsystems accordingly System：The subsystem of renewable fault mode composition and the subsystem of non-renewable fault mode composition, are then estimated using Bayes Method calculate non-renewable subsystem with fault detect probability β and renewable subsystem with primary fault detection probability α；

   4th step utilizes (1) formula to calculate the primary fault undetected probability p (1) that renewable subsystem has；

   P (1)=1- α (1)

   5th step utilizes (2) formula to calculate the testability index growth goal value σ of renewable subsystem,

   Wherein K is that fault mode sum, λ may be updated in renewable subsystem_iFor i-th kind of fault mode failure rate, 1≤i≤N；

   6th step, (3) formula is utilized to calculate renewable subsystem needs failure undetected probability to be achieved at the end of growth test p_end；

   p_end=1- σ (3)

   Meanwhile historical experience is increased according to similar similar system testability, determine the average update coefficient of renewable fault mode Allow direct fault location test number (TN) m, 1≤m≤K/2 of failure before f, 0 ＜ f≤1 and each design update；

   7th step, enables k=1；

   8th step, being calculated before kth secondary design is improved using (4) formula needs the direct fault location test number (TN) r carried out_kDesired value E (r_k),

   E (p (k)) is the desired value for the failure undetected probability p (k) that subsystem may be updated before kth secondary design is improved；

   9th step utilizes (5) formula to may be updated the testability index growth factor q's (k) of subsystem when calculating the improvement of kth secondary design Desired value E (q (k))；

   10th step utilizes (6) formula update kth secondary design that the failure undetected probability desired value E of subsystem may be updated after improving (p(k+1))；

   11st step utilizes (7) formula to calculate experiment and proceeds to direct fault location experiment total degree R required during the kth stage_kDesired value E (R_k)；

   12nd step, if E (p (k+1))≤p_req, turn the 13rd step, otherwise enable k=k+1, and go to the 8th step；

   13rd step, with E (R_k) it is horizontal axis, drawing testability for the longitudinal axis with 1-E (p (k)) increases planning curve, and according to the curve Formulate renewable subsystem testing growth test scheme；

   14th step, enables k=1；

   15th step, the simple random sampling method with reference to as defined in GJB2072-94 is " maintainability test and evaluation " national military standard, from Renewable fault mode is concentrated selects fault mode development direct fault location experiment at random, until kth step-by-step test failure sum etc. In m；

   16th step, in direct fault location after the test, before testability design update, using Bayes methods, according to experiment Success-failure Type Data calculate the undetectable rate of failure of renewable subsystemIfTurn the 18th step, otherwise turn the 17th step；

   17th step, ifThat is experiment process deviates preset testing program, turns the 2nd step, otherwise, right The testability design defect being exposed is designed update, after design update terminates, enables k=k+1, turns the 15th step；

   18th step, off-test.