Research On The Method Of Organic Acid Radical Mixed Pattern Recognition Based On Deep Learning Approach

With the continuous development of society and the rapid development of science and technology,human life is becoming more and more abundant.In addition,there are more and more synthetic things,and there are countless garbage produced.The variety of substances will make it more and more difficult for human beings to identify substances.Organic salts are a kind of salts produced by neutralization reaction of organic acid and alkali solutions.Under certain conditions,these salts can be used as catalysts to catalyze the degradation of heavy oil in oil fields.They are also a new corrosion and scale inhibitor with application value and development prospects.They also have the advantages of fresh-keeping for cold products.The increase of organic salts has brought many difficulties to recognition.However,only by identifying the types of organic salts,can we further analyze their content,concentration and other environmental and human impact,so the identification of mixed organic salts has become a crucial issue.At present,the research direction of organic salts is mainly focused on two aspects.On the one hand,the determination and extraction of organic salts.On the other hand,the effects of organic salts on animals,plants and the environment.However,there are few studies on the recognition of organic salts,and only several types of organic acid radicals with typical characteristics can be identified.The recognition range is limited.We have designed a novel,simple method that can make thousands of experiments at one time,and constructed a reaction chip and network model that can recognize the characteristics of mixed organic acid radicals in different modes.The chip has good sensitivity and discrimination to different mixed organic acid radicals.The network model can output its corresponding eigenvalues from the results of the input mixed solution containing some unknown organic acid radicals,and identify the types of mixed organic acid radicals from the eigenvalues,so as to achieve the purpose of identifying organic acid radicals.After a series of explorations,we finally chose a method based on high-throughput technology and inkjet printing technology to construct a reaction chip.Different organic acid radical solutions were printed on the reaction chip in different ways to form different combinations,and their changes were recorded by scanner.Finally,digital processing and analysis and in-depth learning are carried out.The main conclusion:(1)Selection of chromogenic agent and disturbing agent.Several kinds of chromogenic reagents and disturbing reagents were selected by hole plate and Jiugongge method.Finally,two acid-base chromogenic reagents,methyl red and bromothymol blue,and two metal chromogenic reagents,chrome black T and copper reagent,were selected through the richness of color change.The perturbators are citric acid,copper nitrate,potassium carbonate and potassium aluminium sulfate.(2)Construction of reaction chip.Based on the principle of combinatorial chemistry of high-throughput technology,we designed the printing templates of four kinds of developer solutions and four kinds of disturbant solutions in Photoshop.By chemical printing,the developer and disturbant were attached to the reaction substrate to form the reaction chip in this experiment.(3)Color imaging of mixed organic acid radicals.Four typical organic acid radical solutions with good solubility were selected.They are sodium tartrate,sodium taurine,lactic acid,ascorbic acid and sodium hydroxide,which are formed by the reaction of tartaric acid,taurine,sodium lactate and sodium ascorbate.According to different design templates(mosaic form),they were printed on the reaction substrate to form different patterns of mixed organic acid radicals.The final experimental results were obtained by scanning with a scanner.(4)Processing and analysis of experimental pictures.We standardized and digitized the scanned experimental pictures,converted the pattern information into digital signals,and obtained isograms of mixed organic acid radicals in different modes on the reaction chip,which verified the sensitivity and feasibility of the reaction chip.(5)The construction of deep learning network model.The label values and experimental results of the design template are studied in depth,and the results are analyzed in Origin.Their linear simulation correlation coefficients are between 0.8 and 0.9,which shows the overall feasibility of the method and accuracy of network model.