Data Mining Analysis Based On High Power Millimeter Wave Test System

High-power millimeter-wave devices have a complex structure and many characteristic parameters.In the design and manufacturing stage,a large number of performance tests need to be repeated.Through analysis of the test data,the actual performance of the device is understood.With the development of automatic test technology,the existing high-power millimeter-wave automatic test system has become more and more perfect.It has done a lot of work in accurate data acquisition,massive data storage,real-time data visualization and device basic control protection,and accumulated a large amount of historical test data how to automatically process these data in batches,and use these data to obtain more information to guide the production of high-power millimeter wave devices has become an urgent problem to be solved.In order to solve the above problems,this paper uses data mining technology to study the batch automatic processing of high-power millimeter-wave test data and the application of machine learning algorithm modeling,and expands on the basis of the existing automatic test system.The overall design scheme of the test system is described as follows:1.An overall design scheme of intelligent test system is proposed.The reason why the existing automatic test system is difficult to automatically process data in batches is that the Labwindows / CVI platform lacks corresponding data processing tools.Using data mining technology,Python and related third-party libraries are selected as the main tools to process the data.The processed data is modeled and analyzed by machine learning algorithms to obtain more information.The entire intelligent test system is divided into three layers,the existing automatic test system constitutes the "data collection and protection control layer",the Python data automatic processing part constitutes the "data management analysis layer",and the machine learning algorithm application part constitutes the "multi-parameter optimization strategy Layer",the last two layers are modularized from the overall framework,and the feasibility analysis is done from the hardware and software components.2.Introduced the automatic data processing part of Python in detail.First,the common data analysis methods are introduced with examples,and then the demand points for automatic data processing are analyzed in detail according to the test data characteristics of this subject.The functions of data cleaning,real-time processing,historical data management and other functions are implemented with modular design ideas Improve the efficiency of test data analysis and processing,and finally display the processed data visually.3.The application of the algorithm has studied the prediction based on BP neural network in detail.First,the basic principles of BP neural network algorithm are introduced.Taking a specific experimental data as an example,a BP neural network model is established,and the important attribute "efficiency" value in the test data is predicted.The results show that using BP neural network modeling can get good prediction results.Then the principal component analysis method is used to optimize the BP neural network.By reducing the input data dimension,the amount of network training calculation when the training sample data is large is reduced.Finally,based on the BP neural network,a "fire" prediction model is established,and an active control and protection scheme for improving the safety of high-power millimeter wave devices is proposed.