Soil multi-metal feature XRF collaborative extraction and artificial neural network detection method

Abstract

The invention belongs to the technical field of soil heavy metal detection, and discloses a soil multi-metal feature XRF collaborative extraction and artificial neural network (ANN) detection method, which comprises the following steps: 1, measuring a fluorescence spectrum of a soil sample by using portable XRF; 2, extracting a characteristic spectrum by adopting a competitive self-adaptive reweighted sampling algorithm based on projection variable importance; 3, performing standard normal transformation on the fluorescence intensity of the characteristic spectrum; 4, constructing a feature training set and a feature test set by taking the converted fluorescence intensity as input and the soil heavy metal mass concentration as output, training an ANN model, and optimizing hyper-parameters by using a genetic algorithm; and 5, repeating the steps 1-3 on the new soil sample, and inputting the converted fluorescence intensity into the ANN model to obtain the mass concentration of the heavy metal in the soil. The method solves the problem that the fluorescence intensity peak value is lost when the characteristic spectrum is extracted, improves ANN modeling efficiency and detection precision, and is of great significance to rapid diagnosis of soil heavy metals.

Images