Smart autumn harvest | Identify compound fertilizers and distinguish true from false at the speed of light

Produced by: Science Popularization China

Author: Xu Zhuopin (Institute of Intelligent Machinery, Hefei Institutes of Physical Science, Chinese Academy of Sciences)

Producer: China Science Expo

Editor's note: To showcase the latest trends in intelligent technology, the China Science Popularization Frontier Technology Project has launched a series of articles on "Artificial Intelligence" to provide a glimpse into the latest progress in artificial intelligence and respond to various concerns and curiosities. Let us explore together and welcome the intelligent era.

AI Farmer Assistant Business Card

Assistant: New LIBS&NIRS data fusion model

Three magic weapons: less damage to test samples, fast testing process, and high testing accuracy

Two key technologies: Laser Induced Breakdown Spectroscopy (LIBS) and Near Infrared Spectroscopy (NIRS)

Energy value: 5 stars

ID photo:

(Image source: Reference [1])

AI Farming Assistant Self-introduction

Nice to meet you. I am a new LIBS&NIRS data fusion model and a member of the Institute of Intelligent Machinery, Hefei Institutes of Physical Science, Chinese Academy of Sciences. I am suitable for quality inspection in the production and processing of compound fertilizers. The inspection content includes the nitrogen, phosphorus and potassium content of compound fertilizers. It has fast detection speed, high accuracy, environmental protection and no pollution. It is a new tool to ensure the quality of fertilizer production and processing.

Compound fertilizer and its traditional detection method

Compound fertilizer is a mixed fertilizer containing two or more main nutrients. As the world's largest fertilizer producer and consumer, compound fertilizer is favored by Chinese farmers because of its convenient fertilization process and comprehensive nutrition. The proportion of compound fertilizer in my country's total fertilizer product structure has increased year by year, and since 2019, it has accounted for more than 50% of all fertilizer use.

Fertilizer in the field

(Photo source: veer photo gallery)

The content of the main elements (nitrogen, phosphorus, potassium) is the key to determining the efficiency and price of compound fertilizers. However, the traditional detection method for elemental components in compound fertilizers is generally wet chemical method, that is, the method of preparing materials in liquid phase through chemical reactions. These methods are slow, complicated to operate, and cannot detect multiple components at the same time. In addition, due to the high detection cost and low timeliness of traditional methods, the number of samples is very limited, making it difficult to achieve comprehensive monitoring of the entire production batch. Therefore, in order to ensure the production and processing quality of compound fertilizers in a timely, effective and economical manner, the introduction of new rapid detection methods is imperative.

Inject fertilizer into the soil before planting

(Photo source: veer photo gallery)

My Birth

Laser induced breakdown spectroscopy (LIBS) and near-infrared spectroscopy (NIRS) are two promising analytical techniques for rapid quality detection of compound fertilizers.

LIBS is based on the principle of analyzing the plasma emission spectra generated by laser pulse ablation of the sample surface. When a short-wave pulse laser is focused on the sample surface, plasma is generated, and the plasma emits spectra during this change process. LIBS technology predicts the element content of the sample by analyzing the intensity of these spectra at various wavelengths. Since almost all elements emit characteristic spectra when excited, LIBS can analyze most elements.

NIRS analyzes the absorption signal of products in the near-infrared spectral region when exposed to light. Different molecular structures and chemical bonds have different absorption characteristics under different wavelengths of near-infrared light, so the chemical composition information of a substance can be obtained by measuring its transmission and reflection spectra in the near-infrared region.

The main elements in compound fertilizer can be analyzed by both technologies, but both technologies still have their own limitations. Among them, NIRS is only good at analyzing organic matter, and can only perform indirect measurement of inorganic matter in products such as fertilizers, which will affect its accuracy; and LIBS also needs to overcome the interference of some elements in the ambient air, such as nitrogen, during measurement. Therefore, these problems need to be improved through appropriate steps.

Since both LIBS and NIRS are rapid detection technologies, through reasonable algorithms and steps, the LIBS spectrum and NIR spectrum of compound fertilizer are specifically processed and data fused to achieve the complementary advantages of the two technologies. I can effectively improve the analysis accuracy without affecting the collection efficiency. Compared with general spectral detection technology, it is more advanced and reliable.

There are many specific ways to implement this technology, but generally it includes four steps: spectral acquisition, feature extraction, data fusion, and model construction.

First, it is necessary to use two spectrometers to obtain the LIBS spectrum and NIR spectrum of the fertilizer; second, the LIBS spectrum and NIR spectrum of the compound fertilizer are processed by using a suitable feature extraction algorithm, so that the spectral fragments containing key information in the compound fertilizer are retained, and the rest are eliminated, thereby reducing the complexity of subsequent operations and eliminating the interference of some irrelevant signals; third, the LIBS spectrum and NIR spectrum with extracted features are merged into one piece of data using a suitable data fusion method. These fused spectra contain the beneficial parts of both LIBS and NIRS technologies; finally, the fused compound fertilizer LIBS&NIR spectra are modeled, and some linear or nonlinear algorithms are used to construct the regression equation of the spectrum and the components to be measured.

The model determined by the above series of operations will eventually be integrated into the software and presented to the user. When testing the quality of compound fertilizer, the quality inspector only needs to call the corresponding model in the software and collect the spectrum in production. The software will calculate the compound fertilizer quality index required by the quality inspector and realize rapid testing.

Message from the Assistant

As a new technology, it has the advantages of little damage to test samples, fast testing process, high testing accuracy, and environmental protection and no pollution. It plays a good supporting role in green agriculture, intelligent agricultural production, and especially in improving the quality and efficiency of agricultural production.

But at present, I can only be used in the quality inspection of a limited number of products such as fertilizers and vegetables. The main challenge is that the development and connection of related spectral detection instruments have not yet been standardized, which is not enough for large-scale promotion in the industry. However, as this technology attracts more and more attention, my market is bound to become more mature.

References:

[1] Zhuopin Xu, Xiaohong Li, Weimin Cheng, Guangxia Zhao, Liwen Tang, Yang Yang, Yuejin Wu, Pengfei Zhang, Qi Wang, Rapid and accurate determination methods based on data fusion of laser-induced breakdown spectra and near-infrared spectra for main elemental contents in compound fertilizers, Talanta, Volume 266, Part 1, 2024.