| Describing and analyzing the measurement process from the viewpoint and method of information theory is one of the research directions of measurement theory at present.The research method based on Shannon entropy has gradually developed and achieved some results.From the point of view of information acquisition,this paper uses entropy to study the law of information transmission in measurement system,in order to build a theoretical framework which can not only describe the characteristics of information acquisition in measurement process,but also guide the design and optimization of measurement system.The main contents of this paper are as follows:Firstly,the measurement is analyzed from the angle of uncertainty,and the law of information change in the measurement process is described by using entropy.On this basis,the basic characteristics of multi-link series measurement system are further studied,and the entropy balance characteristics of the measurement system are expounded.The entropy increasing link and the entropy decreasing link are discussed respectively,and the validity principle of the general measurement system is obtained from the angle of mutual information.It is proved that the true value of the measured entropy exists and can be obtained theoretically.Subsequently,the series measurement system is studied in depth.The processing function of the measurement link is expressed by transformation matrix and transfer function.The information entropy transfer rule of the multi-link series measurement system is analyzed in time domain and frequency domain,and the information acquisition methods of typical information processing links such as band-pass filtering,matched filtering and noise matching are deduced.Considering the problems of Shannon Entropy,this paper introduces cross-entropy into the description of measurement system,deduces the entropy balance property of measurement system in the form of cross-entropy,and proves that cross-entropy can also quantify the role of noise and error in measurement system at the information level.In addition,according to the "distance measurement" characteristic of cross-entropy,the evaluation method and design principle of measurement system based on cross-entropy are proposed.Finally,in order to solve the problem of fault diagnosis in measurement system,the cross-entropy algorithm,an optimization algorithm based on the principle of minimum cross-entropy,is studied.The optimization process is regarded as the measurement process of the optimal solution,and the process of obtaining the optimal solution information by the cross-entropy optimization algorithm is analyzed.On this basis,the multi-objective crossentropy algorithm is improved from two aspects of ranking algorithm and elite strategy.A fault diagnosis method combining cross-entropy algorithm and support vector machine are proposed.The effectiveness of this method is proved by the prediction of anode effect in aluminium electrolysis process. |
