Detection Method Of Component Concentration Based On Image Content Retrieval

In recent years, China rare earth extraction separation industry has been developedrapidly, especially solvent extraction separation process in rare earth has reached theinternational advanced level under the guidance of the cascade extraction theory. But rareearth extraction separation process, which is made up of mixer settler, has much separationprocess series、complex mechanism, as control objects it also has same characteristics, such asmultivariable、strong coupling、nonlinear、time varying、delay and so on. Thus, it is difficultto realize elements components content online testing of the rare earth extraction separationprocess, which is a difficult problem of all the rare earth industry extraction separationprocess’s automatic control. To realize the rare earth extraction separation process ofautomatic control, the online detection of the cascade extraction slot rare earth elementscomponent content must be sloved firstly.According to the color characteristic of P507-RECL3rare earth extraction separationsystem, by researching the visual information feature retrieval, a detection method of rareearth component content is proposed based on image content retrieval. It firstly studies anddiscusses the main content of the content-based image retrieval technology, on the basis ofthis, we combine the rare earth ion feature color, color characteristics of rare earth ions withchanges of the contents of components are analyzed, a relationship model between colorfeatures H component and rare earth element component content is established, and the moselis verified, error can satisfy the requirements of online detection. In the relation model,starting from the visual information retrieval technology, using the internal relationshipbetween H component and rare earth element component content,two retrieval algorithm areput forward based on H component color histogram and H component color coherence vector,rare earth element components ‘soft-sensor is realized, and then, in order to improve thesoft-sensor precision, the relevance feedback algorithm based on PSO is presented, and thehuman-computer interaction is also realized.In order to prove the effectiveness of the proposed soft-sensor method, by using VC++and SQL Server, a rare earth element components soft-sensor experiment platform isdeveloped based on imgae content retrieval. Through the collection and shooting laboratoryimages which contain different rare earth components, the soft-sensor method is validatedwith them, the experiment results show that this method can satisfy online detectionrequirements in precision and time.