Establishment method of vis-nir spectral depth feature model based on sae-lssvr cadmium content in crops

Abstract

The invention discloses a method for establishing a Vis-NIR spectral depth characteristic model based on SAE-LSSVR crop cadmium content, initializes the weight matrix W and the offset b of the stacked autoencoder, and uses the spectral data set S as the input of the i layer; from the spectral The training set, prediction set and cross-validation set are respectively extracted from the data set S and the cadmium content label set V, and the coefficient of determination R corresponding to the training set, prediction set and cross-validation set is calculated using the partial least squares support vector machine regression algorithm. 2 and the root mean square error RMSE, and assign m‑1 to m. When m=0, the prediction set SP i Maximum coefficient of determination R p 2 Corresponding node number m c is the optimal number of nodes in the i+1 layer; if the i-th layer and the i+1-th layer stack the autoencoder’s optimal node number corresponding to the prediction set determination coefficient RB i and RB i+1 Satisfied conditions: ||RB i+1 -RB i ||<ε, ε is the error, and i is greater than or equal to 2. Find the optimal number of stacked autoencoder layers to complete the model. This method provides a fast and high-precision method for establishing a Vis‑NIR spectral depth model with cadmium content.

Description

technical field [0001] The invention belongs to the technical field of plant detection, in particular to a method for establishing a Vis-NIR spectral depth characteristic model based on SAE-LSSVR crop cadmium content. Background technique [0002] In the soil-crop growth and development system, the stress effect of heavy metal cadmium (Cd) on crops is increasing, causing crop toxicity, metabolic disorders and impaired growth. The seriousness of the problem also lies in the initial stage of accumulation of low-concentration heavy metals in crops, which is not easy to be noticed or perceived by people, and once the toxic effect is more obvious, it is difficult to eliminate. Moreover, heavy metals are easy to accumulate and accumulate in the stems and leaves of lettuce, which can endanger human health and life safety through the food chain. Studies have shown that eating lettuce containing cadmium (Cd) can cause nausea, vomiting, and general fatigue in mild cases, and cause os...

AI Technical Summary

Problems solved by technology

However, using the above two invention patent algorithms, there are problems such as not performing deep-level algorithm fitting, the established model still has an encoder and decoder, and the established model has a complex structure and large randomness.

Images