Study On Identification And Quantitative Analysis Of Chinese Medicinal Materials By Laser-induced Breakdown Spectroscopy Combined With Artificial Neural Network

Chinese medicinal materials are an indispensable raw material for Chinese medicine treatment.Different origins of Chinese medicinal materials have different composition due to many factors such as growth environment and collection time,and thus the therapeutic effect is different,especially plant herbal medicines.In the medicinal medicine market,it is difficult to manually determine the origin and grade of the medicinal materials,and if the product appearance differ slightly,it will easily lead to the mixed use of medicinal materials from different origins,and may even cause the similar appearance medicinal materials to be filled with false medicines,resulting in a low therapeutic effect and even damaging the human body.Laser-induced breakdown spectroscopy(LIBS)technology has become a new technology for the detection of Chinese medicinal materialss because of its advantages such as simultaneous detection of multiple elements,rapid and real-time detection process,and simple pretreatment.The LIBS is an atomic emission spectroscopy technique that focuses a high-energy laser beam to the surface of sample to induce it to generate a hyperthermal and high-density plasma.In this paper,LIBS experimental device was used to identify different origins and different grades of Chinese medicinal materials,and quantitative detection of Ca,Mg,Al,and K in the root of angelica pubescens.The specific research contents are as follows:Firstly,the qualitative analysis of the root of angelica pubescens was carried out.The results showed that the root of angelica pubescens contains K,Ca,Na,Mg,Al,Fe,Li,Ba,Si and C elements,and three different origins have similar spectral profile,its slight difference is reflected in the spectral intensity.The principal component analysis was used to simplity the full spectrum data into several principal components.The first three principal components represented 92% of the full spectrum data and clustering effect works.The first seven principal components were extracted for modeling analysis.The correct recognition rates of BP artificial neural network,linear discriminant analysis,and support vector machine were 99.89%,98.17%,and 98.83%,respectively.The results showed that BP artificial neural network method is better than the other two methods.In order to verify the validity of this model,the origin of codonopsis pilosula was identified.The result showed that the average recognition rate of codonopsis pilosula origin was 95.83%.Secondly,the qualitative analysis of Dendrobium sample was carried out.The results showed that Dendrobium spectra contain elements spectral lines such as Ca,Na,Al,K,Fe,Mg,C,H,O,and CN molecular bands.The BP artificial neural network model was established to distinguish the grades of Dendrobium.The results showed that the model has a good recognition effect and the average recognition rate reaches 98%.Among them,the recognition rate of Dendrobium at grade 1 and level 4 is 100%.The other three grades of Dendrobium also have higher recognition rate,and the recognition rate is higher than 93%.Finally,the quantitative analysis of Ca,Mg,Al,and K elements in the sample of the root of angelica pubescens was performed using LIBS.The quantitative analysis performance of external standard method and BP artificial neural network method was compared.The limits of detection for Ca,Mg,Al,and K were calculated to be 48.96 mg/kg,77.28 mg/kg,35.49 mg/kg,and 948.96 mg/kg,respectively.The concentration of Ca,Mg,Al,and K in the samples of the root of angelica pubescens was predicted using external standard method and BP artificial neural network respectively.Two methods were used to predict the linear fit of concentration and standard concentration,and each quantitative analysis performance was evaluated by 5 parameters.The result showed that the BP artificial neural network has higher prediction accuracy and stability than the external standard method,indicating that the artificial neural network method is an advantageous method for quantitatively analyzing Chinese medicinal materials.