CN112069661A - Parameter intelligent setting method and system for test system - Google Patents
Parameter intelligent setting method and system for test system Download PDFInfo
- Publication number
- CN112069661A CN112069661A CN202010833073.4A CN202010833073A CN112069661A CN 112069661 A CN112069661 A CN 112069661A CN 202010833073 A CN202010833073 A CN 202010833073A CN 112069661 A CN112069661 A CN 112069661A
- Authority
- CN
- China
- Prior art keywords
- parameter
- test
- mapping model
- data set
- record data
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a parameter intelligent setting method and system for a test system, and belongs to the field of intelligent test design. The method comprises the following steps: constructing an initial test record data set by using a test system parameter theoretical mapping model; constructing a test system parameter actual mapping model by using the initial test record data set, expanding the data set in the test process, and performing iterative correction on the regression model by using the expanded data set to obtain an optimal system parameter actual mapping model; and substituting the expected output parameters of the test system into the optimal system parameter actual mapping model to obtain the input parameters required by the test system. The input and output data of the test system are recorded for multiple times, then the recorded data are used as the data set to construct the mapping model, the obtained model is closer to the actual mapping model of the test system, the output of the test system can be closer to the expected value during application, and the adjusting efficiency of the input parameters to be set in the test process is effectively improved.
Description
Technical Field
The invention belongs to the field of intelligent test design, and particularly relates to a parameter intelligent setting method and system for a test system.
Background
When a hardware or software system is used for testing, input parameters need to be set in order to enable the system output to reach a preset expected value, but due to the limited understanding of the characteristics of the system itself or the influence of environmental factors, the output parameters may be different from the expected values, and even have large deviation. This requires the operator to manually try different input parameters to obtain the desired output within a certain error range. For a test system with a large operation cost, how to efficiently and quickly find out a required input parameter setting value to enable the output to be as close to an expected value as possible has important application value.
For the problem of parameter setting in a test system, an operator is often required to guide parameter setting through a theoretical parameter mapping model. However, the initial theoretical parameter mapping model and the actual parameter mapping model may have a deviation, so that when the theoretical parameter mapping model is used to guide parameter setting, a difference exists between a test actual output parameter value and an expected output result, an operator needs to adjust a test input parameter for multiple times to obtain an output result close to the test output parameter expected value, and after accumulation of long-time test work experience, the operator often can obtain a certain degree of perceptual knowledge of the test system model, but still cannot quantize and correct the model, and the parameter setting efficiency cannot be improved. When the operator changes, the former operation experience and the system model perceptual knowledge return to zero, and a new operator can often search the parameter setting rule of the system only from zero, and repeat the former error or searching test work.
Therefore, it is necessary to provide an intelligent parameter setting method for test design, in which the test records of the operator are used as learning samples to construct a more accurate parameter mapping model for guiding the parameter setting work of the subsequent test, and the work difficulty and the work load of the subsequent test are reduced.
Disclosure of Invention
Aiming at the defects or the improvement requirements of the prior art, the invention provides a parameter intelligent setting method and system for a test system, aiming at constructing a more accurate parameter mapping model for the test system, guiding the parameter setting work of the subsequent test and reducing the work difficulty and the work load of the subsequent test.
To achieve the above object, according to one aspect of the present invention, there is provided a parameter intelligent setting method for a test system, including:
s1, constructing an initial test record data set by using a test system parameter theoretical mapping model; the initial data set comprises input parameters and actual output parameters of the test system;
s2, constructing a test system parameter actual mapping model by using the initial test record data set, expanding the data set in the test process, and performing iterative correction on the regression model by using the expanded data set to obtain an optimal system parameter actual mapping model;
and S3, substituting the output parameters expected by the test system into the optimal system parameter actual mapping model to obtain the input parameters required by the test system.
Further, step S1 specifically includes,
s1.1, determining fixed input parameter x of test system according to test requirements1,x2,x3,…,xP-1Input parameter x to be setPAnd a desired output parameter y; x is the number of1,x2,x3,…,xPRepresenting different parameters;
s1.2. initializing fixed input parameter x1,x2,x3,…,xP-1And the expected value of the output parameter y, f (x) is mapped by the parameter theory mapping model of the test system1,x2,x3,…,xP) Obtaining the theoretical value x of the input parameter to be setP;
S1.3. combining prior knowledge to perform trial method on xPNear adjustment is readySetting input parameters until the error between the actual output parameter of the test system and the expected output parameter is less than a set threshold; adding the output parameters meeting the conditions and the corresponding input parameters into an initial test record data set;
s1.4, repeatedly executing the steps S1.2-S1.3 until the initial test record data set reaches the set scale.
Further, step S2 specifically includes:
s2.1, calculating average absolute error record of initial test record data setWherein k is0The number of data sets is recorded for the initial trial,denotes by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPIs determined by the estimated value of (c),denotes xPActual setting value of;
s2.2. initializing j to 1, and setting the current optimal parameter mapping model M to the theoretical parameter mapping model y to f (x)1,x2,x3,…xP) And the average absolute error e of the optimal parameter mapping modelM=e0(ii) a Wherein j is an iteration turn in the model training process;
s2.3, constructing a j-th iteration optimal parameter mapping model y-f by using the existing test record data set and adopting a regression algorithm1 j(x1,x2,x3,…xP) Generating k using the optimal parametric mapping modeljRecording data of new test and calculating average absolute error of the model
S2.4. judge ej<eTWhether the result is true or not; if so, outputting the current optimal parameter mapping model, and finishing the searching process of the optimal parameter mapping model; otherwise, go to step S2.5; wherein e isTIs a preset test error threshold;
s2.5. judge ej<eMWhether the result is true or not; if yes, updating the current optimal parameter mapping model to be y-f1 j(x1,x2,x3,…xP) (ii) a Otherwise, keeping the current optimal parameter mapping model unchanged;
s2.6, judging whether j is more than n; if yes, go to step S2.7; otherwise, outputting the current optimal parameter mapping model, and ending the searching process of the optimal parameter mapping model; wherein n is the maximum value of the iteration turns in the parameter mapping model searching process;
s2.7, adding 1 in iteration turns, j equals j +1, and generating k by using the current optimal parameter mapping model MjNew test record data is read and the mean absolute error of the new test record data set is calculated
And S2.8, taking the initial test record data set and the test record data generated in the previous j-1 iteration process as the existing test record data set, and returning to execute the step S2.3.
Further, the regression algorithm includes a linear regression algorithm, a decision tree regression algorithm and a random forest regression algorithm.
Further, k is generated in step S2.3 and step S2.7 using the optimal parametric mapping model MjThe new test record data specifically comprises:
01. initializing x according to test requirements1,x2,x3,…xp-1The value of y;
02. by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPThe estimated value of the predicted ith piece of data is recorded as
03. In thatAnd its vicinity search xPActual setting value ofInputting the test data into a test system and recording the current test output parameters
04. Judgment ofWhether the result is true or not; if yes, recordTaking the data as the ith test record data of the current round, and entering the step 05; otherwise, returning to execute the step 03; t is2(j) Is a preset error threshold value, yiThe expected value of the test output parameter of the ith test record data in the iteration of the round is obtained;
05. judging i < kjWhether the result is true or not; if yes, making i equal to i +1, returning to execute the step 01, and collecting the next test record data; otherwise, k will be generatedjAdding new test record data into the original test record data set, and finishing the new data generation process; wherein k isjThe scale of the recorded data set is tested in the current iteration round.
Further, the maximum iteration turn n of the parameter mapping model and the scale k of the test record data set in each iteration turnjAnd initial trial record data set size k0The determination method specifically comprises the following steps:
setting the maximum iteration turn n of the parameter mapping model and the scale k of the data set recorded in each iteration testjAnd initial trial record data set size k0A value range;
setting a plurality of different theoretical parameter mapping models and assumed actual parameter mapping models, and performing one-to-one combination, wherein the combination number is more than or equal to 3;
under each combination condition, n and k are listedjAnd k0All the values of (1) are combined;
for each parameter combination, the following steps are carried out in sequence; setting different inputs (x)1,x2,x3,…xP) Obtaining the actual output y of the test system, then (x)1,x2,x3,…xPY) is a piece of data, resulting in k0Strip data as an initial test record dataset; carrying out the optimal parameter mapping model searching process of the test system in a simulation mode, and satisfying the condition ej<eTUnder the premise of (1), make xpParameters n, k with minimum total search timesjAnd k0The value of (a) is an optimal value; optimal n and k obtained under various combinations of theoretical parameter mapping model and assumed actual parameter mapping modeljAnd k0The average values are obtained, and the average value of each parameter is used as the final determination value.
According to another aspect of the present invention, there is provided a parameter intelligent setting system for a test system, comprising: a computer-readable storage medium storing a computer program; when the computer program is executed by a processor, the computer program controls the equipment on which the storage medium is positioned to execute the parameter intelligent setting method for the test system.
In general, the above technical solutions contemplated by the present invention can achieve the following advantageous effects compared to the prior art.
(1) According to the parameter intelligent setting method based on the test system, the input and output data of the test system are recorded for multiple times, then the recorded data are used as the data set to construct the mapping model, the obtained model is closer to the actual mapping model of the test system, and the model is iteratively applied to the guidance of test parameter setting, so that the adjustment efficiency of the input parameters to be set in the test process is effectively improved.
(2) The method can obtain a concrete actual mapping model of the test system and is used for replacing the perceptual knowledge of operators on the test system, so that even if the operators are replaced, new operators can set input parameters under the guidance of the model, the parameter setting rule of the system does not need to be searched, the previous error or searching test work is repeated, and the time cost for operating the test system is greatly reduced.
Drawings
FIG. 1 is a flowchart of an optimal parameter mapping model iterative training and optimal model search process;
FIG. 2 is a diagram of k generation using an optimal parametric mapping model MjThe data flow chart of the new test is recorded.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The embodiment of the invention provides an intelligent parameter setting method for a test system, which comprises the following steps:
s1, constructing an initial test record data set by using a test system parameter theoretical mapping model; the initial data set comprises input parameters and actual output parameters of the test system;
the step S1 specifically includes the steps of,
s1.1, determining fixed input parameter x of test system according to test requirements1,x2,x3,…,xP-1Input parameter x to be setPAnd a desired output parameter y; x is the number of1,x2,x3,…,xPRepresenting different parameters; s1.2. initializing fixed input parameter x1,x2,x3,…,xP-1And the expected value of the output parameter y, f (x) is mapped by the parameter theory mapping model of the test system1,x2,x3,…,xP) Obtaining the theoretical value x of the input parameter to be setP(ii) a S1.3. combining prior knowledge to perform trial method on xPAdjusting input parameters to be set nearby until the error between the actual output parameters of the test system and the expected output parameters is smaller than a set threshold; adding the output parameters meeting the conditions and the corresponding input parameters into an initial test record data set; s1.4, repeatedly executing the steps S1.2-S1.3 until the initial test record data set reaches the set scale.
Taking a laser attenuation system as an example, the laser attenuation system is used for attenuating laser energy to obtain specified laser-to-target power, the input parameters of the system comprise current, repetition frequency, pulse number, pulse width and attenuation coefficient, and the output parameter is power. Since there are 5 input parameters, p is 5 in step S1.1; x is the number of1,x2,x3,x4Respectively current, repetition frequency, pulse number and pulse width, and are fixed and unchanged; x is the number of5For attenuation coefficient, to be set, x1,x2,x3,x4,x5And y represents a data record.
S2, constructing a test system parameter actual mapping model by using the initial test record data set, expanding the data set in the test process, and performing iterative correction on the regression model by using the expanded data set to obtain an optimal system parameter actual mapping model;
as shown in fig. 1, step S2 specifically includes:
s2.1, calculating average absolute error record of initial test record data setWherein k is0The number of data sets is recorded for the initial trial,denotes by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPIs determined by the estimated value of (c),denotes xPActual setting value of; s2.2. initializing j to 1, and setting the current optimal parameter mapping model M to the theoretical parameter mapping model y to f (x)1,x2,x3,…xP) And the average absolute error e of the optimal parameter mapping modelM=e0(ii) a Wherein j is an iteration turn in the model training process; s2.3, constructing a j-th iteration optimal parameter mapping model y-f by using the existing test record data set and adopting a regression algorithm (such as a linear regression algorithm, a decision tree regression algorithm, a random forest regression algorithm and the like)1 j(x1,x2,x3,…xP) Generating k using the optimal parametric mapping modeljRecording data of new test and calculating average absolute error of the model S2.4. judge ej<eTWhether the result is true or not; if so, outputting the current optimal parameter mapping model, and finishing the searching process of the optimal parameter mapping model; otherwise, go to step S2.5; wherein e isTIs a preset test error threshold; s2.5. judge ej<eMWhether the result is true or not; if yes, updating the current optimal parameter mapping model to be y-f1 j(x1,x2,x3,…xP) (ii) a Otherwise, keeping the current optimal parameter mapping model unchanged; s2.6, judging whether j is more than n; if yes, go to step S2.7; otherwise, outputting the current optimal parameter mapping model, and ending the searching process of the optimal parameter mapping model; wherein n is the maximum value of the iteration turns in the parameter mapping model searching process; s2.7, adding 1 in iteration turns, j equals j +1, and generating k by using the current optimal parameter mapping model MjNew test record data is read and the mean absolute error of the new test record data set is calculatedAnd S2.8, taking the initial test record data set and the test record data generated in the previous j-1 iteration process as the existing test record data set, and returning to execute the step S2.3.
As shown in fig. 2, k is generated in step S2.3 and step S2.7 using the optimal parametric mapping model MjThe new test record data specifically comprises:
01. initializing x according to test requirements1,x2,x3,…xp-1The value of y; 02. by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPThe estimated value of the predicted ith piece of data is recorded as03. In thatAnd its vicinity search xPActual setting value ofInputting the test data into a test system and recording the current test output parameters04. Judgment ofWhether the result is true or not; if yes, recordTaking the data as the ith test record data of the current round, and entering the step 05; otherwise, returning to execute the step 03; t is2(j) Is a preset error threshold value, yiThe expected value of the test output parameter of the ith test record data in the iteration of the round is obtained; 05. judging i < kjWhether the result is true or not; if yes, making i equal to i +1, returning to execute the step 01, and collecting the next test record data; otherwise, k will be generatedjTest record of new stripAdding the data into the original test record data set, and finishing the new data generation process; wherein k isjThe scale of the recorded data set is tested in the current iteration round.
Step S2 requires first determining the maximum iteration turn n in the optimal model search process, and the scale k of the data set recorded in each iteration testj(j ═ {1,2,3.., n }) and initial test record dataset scale k0The process need not be run on a laser attenuation system, but rather, the empirical values may be determined by simulation assuming several sets of actual mapping models of the laser attenuation system. It should be noted that, here, n and k are obtained by simulation methodj(j ═ {1,2,3.., n }) and k0The empirical values of the parameters are equal, so an assumed actual parameter mapping model can be set in advance, but in the actual operation process of step S2, the actual parameter mapping model is unknown, and it is expected that the optimal model searching process of the method can be obtained approximately. The specific method comprises the following steps:
setting the maximum iteration turn n of the parameter mapping model and the scale k of the data set recorded in each iteration testjAnd initial trial record data set size k0A value range; for example, n ranges from 1 to 4, kjIn the range of 0 to 200, k0The range of (1) is 0 to 200; setting a plurality of different theoretical parameter mapping models and assumed actual parameter mapping models, and performing one-to-one combination, wherein the combination number is more than or equal to 3; under each combination condition, n and k are listedjAnd k0All the values of (1) are combined; for each parameter combination, the following steps are carried out in sequence; setting different inputs (x)1,x2,x3,x4,x5) Obtaining the actual output y, then (x) of the laser attenuation system1,x2,x3,x4,x5Y) is a piece of data, resulting in k0Strip data as an initial test record dataset; carrying out the searching process of the optimal parameter mapping model of the laser attenuation system in a simulation mode, and in the searching simulation process of the optimal parameter mapping model, satisfying the condition ej<eTUnder the premise of (1), make x5Total number of searches (i.e. execution of step 03 above)Number of lines) is minimizedjAnd k0The value is the optimal value; optimal n and k obtained under various combinations of theoretical parameter mapping model and assumed actual parameter mapping modeljAnd k0The average values are obtained, and the average value of each parameter is used as the empirical value.
The simulation mode is more convenient for searching the laser attenuation system parameter mapping model than the actual operation of the laser attenuation system, manual system operation is not needed, and the loss of a hardware system is avoided, so that the embodiment of the invention determines n and kj(j ═ {1,2,3.., n }) and k0When the parameter empirical values are equal, the parameter empirical values are obtained in a simulation mode, and the difference from the optimal parameter mapping model searching process in step S2 is that the test record data set in the simulation mode is constructed in a simulation mode, that is, the data set is obtained through an actual parameter mapping model assumed in advance.
And S3, substituting the output parameters expected by the test system into the optimal system parameter actual mapping model to obtain the input parameters required by the test system.
When the validity of the method provided by the invention is verified, the regression algorithm adopted by the embodiment of the invention is the XGboost algorithm, and the fixed test input parameters are 4 x1,x2,x3,x4The test input parameter to be set has 1 x5The expected output of the test parameter is 1 y. The verification process is as follows:
(a) calculating n and k through simulation processjAnd k0An empirical value;
(b) and (b) searching a simulation process by using the empirical value calculated in the step (a) through an optimal parameter mapping model, and verifying the validity of the empirical value.
The detailed process of the step (a) is as follows: firstly, selecting a theoretical parameter mapping model expression and an actual parameter mapping model expression required in the test process from the expressions (1) to (6), and then performing four groups of tests by using the selected expressions, wherein the expressions are combined in a manner of like
Shown in table 1.
TABLE 1 combination of theoretical parameter mapping model and actual parameter mapping model in simulation test
Then, some parameters involved in the process of training and simulating the parameter mapping model are set, and the setting mode is as shown in expression 2:
TABLE 2 parameter settings during model training simulation
Finally, n, k is calculated0,k1,k2,k3,k4Etc. of the parameters. The results of the simulation test of the search process of the optimal parameter mapping model for tests a-d are shown in table 3.
TABLE 3 test a-d optimal parameter mapping model search process simulation test results
At this time, n, k is calculated0,k1,k2,k3,k4The empirical values of (a) are:
after calculating to obtain n, k0,k1,k2,k3,k4After waiting for the empirical value of the parameter, performing step (b) to verify the validity of the empirical value, as described in detail below:
in order to verify the validity of the empirical value, the invention adds four groups of tests e-h, the combination mode of the theoretical parameter mapping model and the actual parameter mapping model of the tests e-h is shown in table 4, the test result is shown in table 6,the data sets used in the simulation process are searched for the optimum parameter mapping model for trial e through trial h respectively,test sets for trials e through h, respectively (each trial using 200 new data sets as test sets).
TABLE 4 theoretical parameter mapping model and actual parameter mapping model combination mode of experiment e-h
TABLE 5 simulation test results of test e-h optimal parameter mapping model search process using empirical values of parameters
In Table 5, the fixed and unchangeable theoretical parameter mapping model is adopted to simulate manual search, the efficiency of the method is compared with that of manual search, and x in the data set is concentrated when the fixed and unchangeable theoretical parameter mapping model is adopted5The number of searches in (2) represents x in the data set during manual search5The number of searches of (2). As can be seen from table 5: obtaining n, k Using trials a-d0,k1,k2,k3,k4After the empirical value of (2), the empirical value is comparedApplied to the New test (tests e-h), parameter x5The search efficiency is obviously improved, which shows that the invention can greatly improve the efficiency of parameter setting, and n, k0,k1,k2,k3,k4The setting of the empirical value has certain rationality, and when the empirical value is applied to a new data test, the parameter x can still be improved5The search efficiency of (1).
The test system of the present invention refers to a hardware or software system in which the actual output and the theoretical output have errors with respect to the specific input, and an operator often needs to perform multiple adjustments and test groceries of input data in order to obtain a desired test output result. In addition, the system may be required to output different test results due to application requirements. An input parameter for a laser attenuation system is for example a current x1Repetition frequency x2Number of pulses x3Pulse width x4And attenuation coefficient x5The output is the target power y, and theoretically, the input and the output have a mapping relation of y ═ f (x)1,x2,x3,x4,x5) However, due to limited understanding of the characteristics of the laser attenuation system itself, the influence of environmental factors, and the like, the mapping relationship between the actual input and output of the laser attenuation system is y '═ f' (x ═ x ″)1,x2,x3,x4,x5) The method of the invention is applicable to similar test systems.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. A parameter intelligent setting method for a test system is characterized by comprising the following steps:
s1, constructing an initial test record data set by using a test system parameter theoretical mapping model; the initial data set comprises input parameters and actual output parameters of the test system;
s2, constructing a test system parameter actual mapping model by using the initial test record data set, expanding the data set in the test process, and performing iterative correction on the parameter actual mapping model by using the expanded data set to obtain an optimal system parameter actual mapping model;
and S3, substituting the output parameters expected by the test system into the optimal system parameter actual mapping model to obtain the input parameters required by the test system.
2. The intelligent parameter setting method for the test system as claimed in claim 1, wherein the step S1 specifically comprises,
s1.1, determining fixed input parameter x of test system according to test requirements1,x2,x3,…,xP-1Input parameter x to be setPAnd a desired output parameter y; x is the number of1,x2,x3,…,xPRepresenting different parameters;
s1.2. initializing fixed input parameter x1,x2,x3,…,xP-1And the expected value of the output parameter y, f (x) is mapped by the parameter theory mapping model of the test system1,x2,x3,…,xP) Obtaining the theoretical value x of the input parameter to be setP;
S1.3. combining prior knowledge to perform trial method on xPAdjusting input parameters to be set nearby until the error between the actual output parameters of the test system and the expected output parameters is smaller than a set threshold; adding the output parameters meeting the conditions and the corresponding input parameters into an initial test record data set;
s1.4, repeatedly executing the steps S1.2-S1.3 until the initial test record data set reaches the set scale.
3. The intelligent parameter setting method for the test system according to claim 1 or 2, wherein the step S2 specifically includes:
s2.1, calculating average absolute error record of initial test record data setWherein k is0The number of data sets is recorded for the initial trial,denotes by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPIs determined by the estimated value of (c),denotes xPActual setting value of;
s2.2. initializing j to 1, and setting the current optimal parameter mapping model M to the theoretical parameter mapping model y to f (x)1,x2,x3,…xP) And the average absolute error e of the optimal parameter mapping modelM=e0(ii) a Wherein j is an iteration turn in the model training process;
s2.3, constructing a j-th iteration optimal parameter mapping model y-f by using the existing test record data set and adopting a regression algorithm1 j(x1,x2,x3,…xP) Generating k using the optimal parametric mapping modeljRecording data of new test and calculating average absolute error of the model
S2.4. judge ej<eTWhether the result is true or not; if so, outputting the current optimal parameter mapping model, and finishing the searching process of the optimal parameter mapping model; otherwise, go to step S2.5; wherein e isTIs a preset test error threshold;
s2.5. judge ej<eMWhether the result is true or not; if yes, updating the current optimal parameter mapping model to be y-f1 j(x1,x2,x3,…xP) (ii) a Otherwise, keeping the current optimal parameter mapping model unchanged;
s2.6, judging whether j is more than n; if yes, go to step S2.7; otherwise, outputting the current optimal parameter mapping model, and ending the searching process of the optimal parameter mapping model; wherein n is the maximum value of the iteration turns in the parameter mapping model searching process;
s2.7, adding 1 in iteration turns, j equals j +1, and generating k by using the current optimal parameter mapping model MjNew test record data is read and the mean absolute error of the new test record data set is calculated
And S2.8, taking the initial test record data set and the test record data generated in the previous j-1 iteration process as the existing test record data set, and returning to execute the step S2.3.
4. The intelligent parameter setting method for the test system as claimed in claim 3, wherein the regression algorithm comprises a linear regression algorithm, a decision tree regression algorithm and a random forest regression algorithm.
5. The intelligent parameter setting method for test system as claimed in claim 3, wherein the optimal parameter mapping model M is used to generate k in steps S2.3 and S2.7jThe new test record data specifically comprises:
01. initializing x according to test requirements1,x2,x3,…xp-1The value of y;
02. by y ═ f1 j(x1,x2,x3,…xP) Predicting the test input parameter x to be setPThe estimated value of the predicted ith piece of data is recorded as
03. In thatAnd its vicinity search xPActual setting value ofInputting the test data into a test system and recording the current test output parameters
04. Judgment ofWhether the result is true or not; if yes, record Taking the data as the ith test record data of the current round, and entering the step 05; otherwise, returning to execute the step 03; t is2(j) Is a preset error threshold value, yiThe expected value of the test output parameter of the ith test record data in the iteration of the round is obtained;
05. judging i < kjWhether the result is true or not; if yes, making i equal to i +1, returning to execute the step 01, and collecting the next test record data; otherwise, k will be generatedjAdding new test record data into the original test record data set, and finishing the new data generation process; wherein k isjThe scale of the recorded data set is tested in the current iteration round.
6. The intelligent parameter setting method for test system as claimed in claim 3 or 5, wherein the parameter mapping model has a maximum iteration number n and a test record data set size k in each iterationjAnd initial trial record data set size k0The determination method specifically comprises the following steps:
setting the maximum iteration turn n of the parameter mapping model and recording data of each iteration testSet size kjAnd initial trial record data set size k0A value range;
setting a plurality of different theoretical parameter mapping models and assumed actual parameter mapping models, and performing one-to-one combination, wherein the combination number is more than or equal to 3;
under each combination condition, n and k are listedjAnd k0All the values of (1) are combined;
for each parameter combination, the following steps are carried out in sequence; setting different inputs (x)1,x2,x3,…xP) Obtaining the actual output y of the test system, then (x)1,x2,x3,…xPY) is a piece of data, resulting in k0Strip data as an initial test record dataset; carrying out the optimal parameter mapping model searching process of the test system in a simulation mode, and satisfying the condition ej<eTUnder the premise of (1), make xpParameters n, k with minimum total search timesjAnd k0The value of (a) is an optimal value; optimal n and k obtained under various combinations of theoretical parameter mapping model and assumed actual parameter mapping modeljAnd k0The average values are obtained, and the average value of each parameter is used as the final determination value.
7. An intelligent parameter setting system for a test system, comprising: a computer-readable storage medium storing a computer program; controlling an apparatus in which the storage medium is located to perform the method of any one of claims 1 to 6 when the computer program is executed by a processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010833073.4A CN112069661A (en) | 2020-08-18 | 2020-08-18 | Parameter intelligent setting method and system for test system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010833073.4A CN112069661A (en) | 2020-08-18 | 2020-08-18 | Parameter intelligent setting method and system for test system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112069661A true CN112069661A (en) | 2020-12-11 |
Family
ID=73661359
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010833073.4A Pending CN112069661A (en) | 2020-08-18 | 2020-08-18 | Parameter intelligent setting method and system for test system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112069661A (en) |
-
2020
- 2020-08-18 CN CN202010833073.4A patent/CN112069661A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR102206324B1 (en) | Methods for ascertaining a model of a starting variable of a technical system | |
CN114026510B (en) | Method and system for optimizing process parameters in an additive manufacturing process | |
CN111382906A (en) | Power load prediction method, system, equipment and computer readable storage medium | |
CN113298257A (en) | Method and device for creating a model of a technical system from measurements | |
WO2019200626A1 (en) | Artificially intelligent body training system, passive circuit optimization design system, and method | |
CN115809624B (en) | Automatic analysis design method for integrated circuit microstrip line transmission line | |
CN114880806A (en) | New energy automobile sales prediction model parameter optimization method based on particle swarm optimization | |
CN114841106A (en) | Integrated circuit optimization method and system based on rule-guided genetic algorithm | |
CN113919221A (en) | Fan load prediction and analysis method and device based on BP neural network and storage medium | |
US7617166B2 (en) | Neural network for aeroelastic analysis | |
CN112069661A (en) | Parameter intelligent setting method and system for test system | |
CN113706116A (en) | Ship welding man-hour determining method and device, computer equipment and storage medium | |
CN113868765A (en) | Ship main scale parameter optimization method based on approximate model | |
CN115345106B (en) | Verilog-A model construction method, system and equipment for electronic device | |
KR20210060147A (en) | Structural Performance Based Inverse Material Design Method | |
CN115906303A (en) | Planar microwave filter design method and device based on machine learning | |
KR20200021768A (en) | An apparatus for optimizing fluid dynamics analysis and a method therefor | |
CN112257215B (en) | Maximum likelihood estimation solving method and system for product life distribution parameters | |
CN112712180A (en) | Model construction optimization method, device, equipment and medium based on reliability verification | |
van der Herten et al. | Fuzzy local linear approximation-based sequential design | |
CN114880792A (en) | Deformation prediction-based omnibearing multi-angle optimization method | |
CN112632860B (en) | Power transmission system model parameter identification method based on reinforcement learning | |
CN115544815B (en) | Method and device for generating fan model | |
Zhu et al. | 3.3 Engine Calibration Using Global Optimization Methods | |
KR102101792B1 (en) | An apparatus for analyzing fluid dynamics and a method therefor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |