Category:
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
0
1
2
3
4
5
6
7
8
9
Position:IcFull.com » IC Electronic information » Category: R

IC Electronic information

Reliability test data processing and engineering implementation

In Electronic Infomation Category: R | on April 20,2011

Abstract: Reliability of electronic equipment during the assessment process, obtained through various tests on the raw data reliability assessment and ADV7302AKST datasheet and reliability design of equipment is important, the effect of the reliability of test data directly affects the reliability of the estimated frequency the accuracy and ADV7302AKST price and effectiveness. Here the reliability test data processing method and ADV7302AKST suppliers and MATLAB and Visual C + + mixed programming in several ways, and considering the advantages and disadvantages of using one of the methods of its programming reliability of the data processing system, data processing to achieve reliability analysis, data processing automation, reduce data processing cycle, and illustrates the effectiveness and availability.

0 Introduction

With the increasing complexity of electronic equipment, high reliability of the equipment of modern warfare demands are rising, and therefore a higher reliability test requirements, and equipment reliability test data is reliable * assessment of the design, an important basis and strong research support, data analysis is the basis of the work of all the reliability, the reliability of test data analysis has increasingly shown its great value and role. On system reliability analysis, must be a lot of raw data collected pretreatment, using artificial means to accomplish this task is a waste of time, and prone to error. MATLAB software has the powerful numerical computing environment and statistical analysis, and VC + + and can achieve good interactive interface, using both software hybrid programming approach, to achieve both the reliability of test data to make up their own shortcomings, but also improves test efficiency, the best choice of software programming.

1 reliability test raw data processing theory

Reliability test raw data processing theory references. Reliability test raw data can not be made directly available to researchers and analysis of reliability assessment shall be for the test object, test type, select a different signal on the type of data processing standard for the treatment of the original test data include the following:

1.1 feature extraction

Feature extraction mainly for extraction of the raw data collected test subjects had significant changes in the trend of stress conditions, the reliability of assessment parameters for the extraction, as described in the literature of a type of radar equipment, signal boards for the 20 kHZ the reliability study, for example, the final output is a signal board 20 kHz sinusoidal signal, so its frequency, period and amplitude (or peak to peak) are able to reflect the provisions of the board is complete function and an important indicator of the failure parameters and therefore the raw data processing for the test work is to extract the signal frequency, amplitude and other parameters, organized into a certain format for reliability evaluation and analysis. Specific to different types of test data extraction methods vary mainly in the following areas:

1.1.1 Nonlinear data fitting

The basic principles of curve fitting is: given a set of measured data (such as N points (xi, yi) obtained from the variable x and dependent variable y, an approximate analytical expression of y = (x). If the write error i = (xi) - yi, i = 1,2, ..., N, is to make the square error and minimum requirements:


Minimum, which is commonly used least square method.

Matlab function in a nonlinear data fitting lsqcurvefit, lsqnONlin, inline three functions to lsqcurvefit example, the call format is: x = lsqcurvefit (fun, x0, xdata, ydata) in which fun M-function for the fitting function file name, x0 is the initial vector, xdata, ydata curve fitting for the participation in the independent variables and the dependent variable test data. Function call with two other similar format. Nonlinear data fitting can be used mainly for simple signals and specific parameter function model signals, such as sine wave, triangle wave, square wave.

1.1.2 Wavelet Analysis

No specific signal for a number of complex signal function model can not be used for nonlinear curve fitting parameter extraction, such as the reliability of the test pilot applied vibration stress the need for vibration signal analysis and extraction parameters, such signals wavelet analysis method parameter extraction.

Wavelet analysis to process data consists mainly of two aspects: wavelet noise reduction; extracted from the wavelet packet energy.

(1) wavelet denoising theory

Because of the low entropy of wavelet transform, and other characteristics related to better connect the signal to noise reduction, wavelet noise reduction process can be divided into three steps: Choose a wavelet, and determine the level of a wavelet decomposition N, N layer and then the signal wavelet decomposition; of the high frequency coefficients of wavelet thresholding, from layer 1 to layer N high-frequency coefficients of each layer to select a threshold value threshold of quantification; According to the first N wavelet coefficients and after low level after quantification of Layer 1 to Layer N high-frequency coefficients for signal reconstruction.

(2) wavelet packet based feature extraction

Extraction based on wavelet packet parameters of the steps are:

wavelet denoising the signal, extract the useful signal; of the signal after noise reduction using a reasonable scale wavelet decomposition, the decomposition coefficients are different frequency bands; , respectively, the decomposition coefficients of different frequency bands for signal reconstruction, get a new set of time series; on the reconstruction of the Hilbert transform signal the envelope, the envelope of wavelet noise reduction and treatment; signals in different frequency bands respectively, seeking their envelope spectrum, and calculate the energy spectrum of each envelope in order to construct the energy of the signal feature vector elements.

Envelope energy spectrum:


Equation (2) ik x (i = 1,2; k = 1,2, ..., n), said reconstruction of the signal envelope amplitude spectrum of the discrete points. When the energy is large, Ei is usually a large number, may be its normalized so that:


The T = e 1, e 2 ... ... ei as the feature vector.

1.2 reliability data to draw the frequency distribution

First raw data grouped by size, by walking in the number of each group made distribution of data, this graph is called frequency distribution. For randomly distributed data, statistical frequency bar chart can be used to describe the image.

The reliability of data read into the MATLAB workspace, you can draw the frequency distribution reliability data. HiST MATLAB Statistics Toolbox provides functions for the distribution for the MATLAB command, the command format is as follows:

HIST (data, k) which, data for the raw data; k is divided by the number of residential rooms.

1.3 Parameter Estimation

Drawn out according to the shape of distribution, assuming a distribution of the reliability of the data subject, in general, the reliability of the data is negative exponential distribution. Exponential distribution parameters can be estimated by the command expfit, the order with its maximum likelihood method gives the probability distribution commonly used point estimates of parameters and regional estimates, the command format is as follows:

[Muhat, muci] = expfit (data),

Where, muhat the estimated value for the parameter ; muci estimate for the parameter , Confidence interval. Normal distribution parameters can be estimated by the command normfit, the order with the maximum likelihood method gives the probability distribution commonly used point estimates of parameters and regional estimates, the command format is as follows:

[Muhat, sigmahat, muci, sigmaci] = normfit (data),

Where, muhat the estimated value for the parameter ; muci estimates for the parameter , confidence interval; sigmahat confidence interval for the parameter .

2 design and realization

Test for the reliability of the test object, test type, the choice of subjects reflects the characteristics of failure are not the same amount of parameters. The two methods used to process the user based on the different types of raw test data, choose a different method of handling them by the software and its related parameters obtained by processing.

2.1 Scheme Selection An initial implementation using MATLAB program

data processing, human-computer interface design is done with the VC, the main problem is the VC and the MATLAB interface. MATLAB programming will be mixed with the VC has the following four methods:

call MATLAB engine. Advantage of the method is able to support all of the MATLAB function, the disadvantage is mixed from the executable program after program can not run MATLAB environment;

own using MATLAB mcc compiler;

Using Matcom compilation. Matcom conversion is very simple to use, convenient, and the generated code is quite readable, and in the C compiler to compile the code after the execution of its high efficiency. But this method can not support all the MATLAB toolbox function;

Using MATLAB COMBuilder. Provided by COM Builder MATLAB (COMBuilder), standalone applications to achieve an increase of MATLAB and a new way.

Comparing the advantages and disadvantages on the basis of several methods to determine the methods used to achieve the first reliability test raw data processing.

2.2 MATLAB to achieve specific functionality

To 20 kHZ signal of a radar target board for the test carried out under temperature stress accelerated life testing, the scope of data collection is excel file, call the Matlab software from the non-linear fitting of treatment is to achieve the following characteristics extraction of the source code:

Sampt = xlsread (F: \ 2-1, B3: B1002);% reading sampling time

V = xlsread (F: \ output195, B2: AS1001);% read in data collection

For i = 1:11 smv (:, i) = smooth (v (:, i)); end% for data smoothing

F = @ (x, xdata) x (1) * sin (x (2) * xdata + x (3)) + x (4)), x, xdata;

Xdata = sampt;

For j = 1:44

Ydata = v (:, j); x0 = [9 1 * 10 ^ 5 0 0];% initial component

X = lsqcurvefit (F, x0, xdata, ydata);

Amp (j) = x (1); fre (j) = x (2) / (2 * pi); inip (j) = x (3); inic (j) = x (4); < / P>

End

Xlswrite (F: \ processing results \ amplitude.xls, amp, B2: AS2)% the rate parameter to write Excel file

Xlswrite (F: \ processing results \ amplitude.xls, amp, B2: AS2)% the rate parameter to write Excel files 2.3 MATLAB and VC + + Mixed Programming:

As mentioned above, VC + + and MATLAB with the advantages and disadvantages, call the MATLAB engine using VC + + mixed programming with MATLAB [6-7] to achieve the reliability test method of the original data extraction. VC + + calling MATLAB engine [3-4], follow these steps:

Data processing results shown in Figure 1 with the original data waveform comparison chart.


Figure 1, the results of data fitting with the original data comparison chart

3 Conclusion

Call MATLAB engine using the methods and VC + + mixed programming method for reliability testing of the original data extraction and fitting operation, drawing histogram processing. This method not only realized the VCs powerful visual interface and the MATLAB numerical analysis and graphical display capabilities of the effective combination not only improves the reliability of the test data processing efficiency, but also saves system resources effectively and shorten the software development cycle.

ADV7302AKST datasheetADV7302AKST suppliersADV7302AKST Price

All right reserved:icfull.com © 2010-2016 Certificate