Research On Description And Applications Of Entropy-based Measurement System

The viewpoint that information acquisition is the essence of measurement has gradually become a consensus in the field of measurement theory research.The direct application of the basic viewpoints and methods of Shannon's information theory to describe the measurement system has also attracted the interest of some researchers.In this thesis,the measurement system realizes the transformation of the work connotation for the purpose of information acquisition,and studies and discusses the modeling method,performance analysis and error evaluation of the measurement system in the sense of information entropy.The main contents of the paper are as follows.A new noun that represents amount of information is proposed and its related properties are proved,which plays a key role in the characterization of the series information measurement system.Several characteristics are obtained to facilitate the analysis and calculation of the system.The obtained entropy balance theorem reveals the process of processing the information of different random variables in the process of information acquisition.The measurement is based on the objective process of the measured source uncertainty from large to small,clarifying the measurement process.The entropy reduction mechanism of information acquisition is clarified.The principle of measurement effectiveness is summarized.These studies laid the foundation for the establishment of a unified measurement system theory.The entropy description of linear networks is an important application.The third chapter is based on the Topelitz distribution theorem to obtain the frequency domain representation of continuous entropy,which proves the linear network entropy theorem and derives the cascaded linear network expression.In this paper,the relationship between the frequency domain and the entropy domain in the application of the theorem does not correspond to the problem,and the interpretation of the formal method is given to provide an idea for analyzing and designing the linear measurement system in the sense of information entropy.This paper first explores the essential difference between entropy increment and noise entropy in the amplification process,and the band-pass filter network are described by entropy,in order to apply the information entropy-based measurement system model to the method design and optimization of specific measurement process.Secondly,the information characteristics of the quantization process are analyzed,and the influence of the input source on the information acquisition of the quantization process under different probability distribution conditions is studied.In addition,the entropy model of information preprocessing is analyzed,and the main factors affecting the performance of the algorithm under the meaning of information acquisition are discussed.Finally,the error evaluation of measurement results based on relative error entropy is given.These methods provide theoretical support and technical guidance for the design and optimization of measurement systems.Using the description method of measurement system based on entropy,this paper studies the maximization of information acquisition in NMR logging system.Firstly,by establishing the entropy model of NMR relaxation process,the information proportion of the first echo under different relaxation conditions is quantitatively determined,and an improved ring noise suppression method is designed to complete the acquisition of formation bound fluid information.Secondly,the problem of serious information loss in the compression process of original NMR logging echo data is studied,and the compression method based on optimal quantization is designed to ensure that the information is maximized under the same compression ratio.Finally,in view of the lack of theoretical basis for regularization inversion methods in selecting prior information such as regularization terms,this paper designs a two-dimensional NMR logging inversion algorithm based on maximum entropy regularization from the perspective of maximum entropy of prior information to maximize information acquisition of inversion operations.