Research on electromagnetic compatibility performance parameter modeling based on measurement data

Publisher:清新天空Latest update time:2014-12-09 Source: eccn Reading articles on mobile phones Scan QR code
Read articles on your mobile phone anytime, anywhere

0 Preface

With the continuous improvement of the level of social informatization, all walks of life have a deeper understanding of the electromagnetic compatibility performance of electronic systems and pay more and more attention to it. However, in reality, many electronic systems have serious electromagnetic compatibility problems, which have seriously affected their performance. An important reason for the problem is that during the use of the system, as the service life continues to increase, some equipment ages, resulting in excessive harmful electromagnetic radiation and reduced anti-interference thresholds; some equipment components have been repaired many times, resulting in reduced electromagnetic compatibility; some due to environmental changes, especially the increasing complexity of the electromagnetic environment, the imperfections of the electromagnetic compatibility performance design are prominent. Therefore, it is particularly important to summarize the electromagnetic parameters and electromagnetic compatibility changes of electronic systems.

Data mining technology can extract implicit, unknown but potentially useful information and knowledge from a large amount of incomplete, noisy, fuzzy and random practical application data. Data mining of electromagnetic compatibility measurement data can effectively extract the variation patterns of certain performance parameters and summarize their data models. This article describes the model establishment process and specific implementation.

1. Parametric model building process

Electromagnetic compatibility test data mainly refers to the original data generated by the electromagnetic compatibility test of the system and the data obtained after sorting, such as the peak power of the transmitting equipment, the spurious level, the number of harmonics with power greater than the set threshold, etc. In the process of data modeling of the measured data, different conclusions can be obtained from different research perspectives. For example, when studying parameter characteristics, a correlation analysis can be performed between a certain parameter and another parameter to study the correlation law between the changes in the parameters; a multivariate regression model can be established to study the change law between a certain parameter and several parameters.

The process of modeling electromagnetic compatibility performance parameters is to obtain raw data through electromagnetic compatibility measurement, preprocess the data, organize the data type and structure, perform data mining on historical data and measured data to complete data law analysis, describe the change curve of equipment parameters, and then complete the data model of parameter change law after multiple data corrections. The most common data mining methods are statistical analysis methods, neural network methods, and methods studied in machine learning. The specific modeling process is shown in Figure 1.

2 Statistical analysis methods for electromagnetic parameter modeling

The above briefly introduces the process of electromagnetic parameter modeling. Different models are established for different analysis problems. In this section, we study the change of amplification factor of a power amplifier over time based on the hypothesis. The application of data mining in electromagnetic parameter modeling is introduced through statistical analysis methods. Among them, regression analysis is the specific data statistical analysis method used in this experiment.

2.1 Regression analysis

Through regression analysis, the uncertain and irregular quantitative relationships between related variables can be generalized and standardized, so that the possible value (or estimated value) of the dependent variable can be inferred based on a given value of the independent variable. Regression analysis includes many types. According to the number of variables involved, it can be divided into univariate regression and multivariate regression; according to the different forms of variable changes, it can be divided into linear regression and curve regression.

The task of linear regression analysis is to find the linear regression equation that describes the relationship between two variables x and y based on a number of observations (xi, yi) (i=1, 2, ..., n):

Regression analysis that predicts dependent variables by establishing a regression equation based on the optimal combination of multiple independent variables is called multiple regression analysis, and its model is:

The significance test of the regression equation is to test whether β is almost all approximately 0. If it is established, it means that the linear model description is inappropriate.

The general steps are as follows:

(1) Use the input variable as the horizontal axis and the output value, i.e. the test value, as the vertical axis to draw a test curve.

(2) Analyze the curve being drawn and determine the basic form of the formula. If the data points are basically a straight line, the coordinates of the line can be determined using a univariate linear regression method. If the data points draw a curve, it is necessary to determine the type of function the curve belongs to based on the characteristics of the curve. It can be compared and distinguished with known mathematical function curves. If it is difficult to determine the type of the test curve, it can be processed using polynomial regression.

(3) Determine the constants in the fitting equation. The constants in the equation can be determined based on a series of test data.

(4) Test the stability and significance of the determined equation, substitute the independent variables in the test data into the fitting equation to calculate the function value, and see if it is consistent with the actual test value. The size of the difference is usually expressed by standard deviation, and variance analysis, F test, etc. are performed. If the basic form of the determined formula is wrong, another form of the formula should be established.

When conducting research and analysis, consider the change in the gain of a certain power amplifier. Assume that the gain of this power amplifier is obtained through data statistics as shown in Table 1.

2.2 Modeling and Simulation

According to the data given in Table 1, the amplifier gain parameters are modeled using regression analysis. Under the condition of a confidence level of 95%, the curve of its change over working time is obtained according to the steps of regression analysis, as shown in Figure 2. [page]

It can be seen from Figure 2 that when the cubic polynomial is used as its data model, it can fit the given data relatively well. The basic information of each parameter obtained by parameter estimation and model summary is shown in Table 2.

According to the data in Table 2 and the judgment criteria of the F test method, it can be seen that the regression effect of this equation is significant. Therefore, the predicted value and confidence interval of the dependent variable gain can be obtained based on this equation. The specific data are shown in Table 3.

3 Conclusion

The regression model is a very important tool for analyzing test data. It can derive the changing relationship between parameters. Perhaps the changing relationship between a pair of parameters alone is not enough to provide meaningful information. A multivariate regression model can also be established to study the changing relationship between multiple parameters and a certain parameter. Due to the limitations of regression analysis in nonlinear analysis, future work will focus on the application of artificial neural networks in this regard. The method of establishing a parameter model is not fixed. The choice of model varies with different research points. Through the established model, it is possible to predict the future electromagnetic compatibility status of the electronic system without actual measurement, effectively guiding the electromagnetic compatibility analysis and assurance of the system.

Reference address:Research on electromagnetic compatibility performance parameter modeling based on measurement data

Previous article:MEMS improves flow measurement accuracy in medical devices
Next article:Implementation of Phase Difference Measurement Algorithm of Same Frequency Signal Based on FPGA

Latest Test Measurement Articles
Change More Related Popular Components

EEWorld
subscription
account

EEWorld
service
account

Automotive
development
circle

About Us Customer Service Contact Information Datasheet Sitemap LatestNews


Room 1530, 15th Floor, Building B, No.18 Zhongguancun Street, Haidian District, Beijing, Postal Code: 100190 China Telephone: 008610 8235 0740

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号