| Heavy metal pollution is currently one of the most important environmental problems.Contamination of farmland with heavy metals will not only affect crop yields and quality,but more importantly it will endanger human health,and enev threaten human survival and development.As the first link in the whole food chain,plants directly absorb heavy metals from the soil which would later affect their normal growth and development.Digital agriculture and agricultural Internet of Things information perception technology is one of the most advanced fields of modern agriculture.It is the core technology for the development of modern agriculture in the world today and the realization of sustainable agricultural development.At present,the effect of heavy metal pollution on crops in soil-plant production systems in China is becoming more and more serious,causing plant toxic effects,metabolic process disturbances,growth and development injuries,and even plant death,which greatly restricts crop yield,quality,efficiency,ecology,and safety in production.sustainable development.Crop physiology information,as an important indicator of adverse stress,plays an important role in monitoring the adversity of crops,and adopting comprehensive prevention and repair measures in a timely and effective manner.In this study,the rapeseed was the main reserch object,to explore the reponses to different levels of heavy metal cadmium(Cd)stress,and to establish a discriminative model for different levels of Cd stress based on spectroscopy and data fusion methods.Pyrolysis pretreatment for solid sample,as a method of signal enhancement,was also put forward to onstruct a quantitative analysis model of plant heavy metals and nutrient elements,which provided the basis for the rapid detection of nutrients/heavy metals in plants.The main conclusions are as follows:(1)Using chemical detection methods,the relevant physiological responses of rape under different levels of heavy metal Cd stress,and its changing trends of time were explored to provide basis for studying the toxic effects of heavy metals on rapeseed.The main physiological indexes included water content,ascorbic acid,glutathione,and free proline and salicylic acid(both free and bound states).The experimental results show that when the plant sample was exposed to heavy metal stress,the water content in the leaf would decrease,and the higher the stress level,the lower the water content in the sample.The content of AsA and GSH responded rapidly to heavy metal stress.After being exposed to heavy metal stress,the plant sample would rapidly reduce the content of AsA in leaves and increase the content of GSH.On the other hand,free proline has a relatively slow response to stress.When the stress level is low,the free proline content in the sample has a relatively small effect,while a greater degree of stress leads to a large accumulation of free proline in the leaf.Under stress conditions,the content of bound salicylic acid in the leaves of plants rised rapidly,and then it is converted into free salicylic acid;when the degree of stress is low,the conversion rate of salicylic acid from bond state to free state,might depend on the content of free salicylic acid in leaves,which keeped the content of free salicylic acid within a certain range.(2)Established a rapid discriminative model for the defree of Cd stress in rapeseed leaves,which provided a method and technical support for the monitoring of farmland with heavy metal Cd.In the study,fresh rape samples were used as the research object,and the samples were discriminated and analyzed with the LDA model using hyperspectral data,LIBS data and hyperspectral-LIBS data fusion.According to the analytical results,the use of hyperspectral-LIBS data fusion method could significantly improve the accuracy of discriminant results,in which the highest accuracy of discriminating whether the sample was exposed to heavy metal stress was higher than 90%,but the discriminative accuracy of samples suffered different stress levels was also high At 70%.In general,the use of spectroscopy to determine whether or not a sample was exposed to heavy metal stress was good,while the results of discriminant analysis of samples with different levels of stress needed to be improved.(3)Proposed a pyrolysis pretreatment for solid samples as a signal enhancement method,and established a optimal parameter system for pyrolysis pretreatment.The influence of pyrolysis pretreatment on the characteristic peak signals of 9 nutrient elements(i.e.K,Na,Mn,B,P,Cu,Fe,Ca,Mg)in dry leaf samples was compared using LIBS technology,and the pyrolysis treatment parameters were optimized.Experiments showed that the pretreatment of sample pyrolysis could significantly increase the peak signal intensity of the characteristic peaks of each element;however,the enhancement of the SNRs varied according to the kind of element,where the intensities of some elements were inhibited after pretreatment;RSDs were redused after pretreatment,and the heating rate had a relatively large influence on performance of RSD,RSD increased along with the increasing of heating rate.Pyrolysis pretreatment conditions(final temperature and heating rate)were also optimized,showing that the signal strength of the element itself would affect the enhancement performance.Comprehensively compare the variation of the peak-to-peak signal,SNR,and RSD of each element,the pyrolysis parameters were optimized as the final temperature of 250 ? and the heating rate of 5 ?/min.(4)Established a LIBS quantitative detection model based on signal enhancement method of pyrolysis pretreatment,used for the determination of nine nutrient elements and heavy metal Cd in dry rape leaves.The study compared the results of quantitative analysis of 9 nutrient elements and heavy metal Cd in dry leaf samples before and after pyrolysis pretreatment.The PLS loadings were used to extract the characteristic peaks of each element for univariate quantitative analysis.The experiment showed that the pyrolysis pretreatment method was suitable for elements with weak LIBS characteristic peak signals(like P,Mn,B and Cu).This method could significantly increase the performances of modeling and predicting,where Rc2 improved 1.87%-38.5%,and Rp2 2.54%-23.44%;while,those with strong LIBS signals(K,Na,Mg)are prone to the consequances of self-absorption,affecting the results of quantitative analysis.The LODs of each element were significantly reduced,decending 34.19%-75.86%.The biPLS model was used to screen the characteristic bands for multivariate quantitative analysis.Compared with univariate analysis,biPLS could significantly increase the analytical performance of P and Cu elements,indicating that biPLS-based quantitative analysis could comprehensively consider the background signal and the characteristic signals of the tested elements to improve the results of quantitative analysis.The effect of pyrolysis pretreatment on the quantitative analysis of each element was relatively small,and only K,Mn,B,and Cu elements had some improvements for modeling.On the other hand,quantitative analysis of Cd indicated that the pretreatment of pyrolysis could significantly increase the univariate and multivariate quantitative modeling and prediction results,where Rs for both uni-and multi-variate analysis were bigger than 0.97.What's more,the LOD of Cd after pre-treatment was one-tenth of that before pre-treatment,reaching only 5.86ppm.The comparative analysis of LODs before and after pyrolysis showed that the pretreatment method could not only concentrate the effect of the element content in the solid sample,but also improve the influence of the sample matrix on the measured element,thus playing a role in reducing the element detection limit. |
PDF Full Text Request