An online nondestructive detection method for blood-spotted eggs based on spectral amplitude space conversion

Abstract

The invention relates to the technical field of egg quality dynamic detection, in particular to a blood spot egg online nondestructive detection method based on spectrum amplitude value space conversion. The method mainly comprises the following steps of collecting the spectral transmission rate of each egg sample through a blood spot egg online detection system to obtain spectral data; performing preprocessing on the spectral data; in order to reduce the interference caused by the speed, the batch, the instrument and the environment in the experiment process, converting spectral signals of the preprocessed data into an artificially specified amplitude value space by using an LM algorithm; obtaining the converted processing data; building a blood spot egg BP neural network online discriminating model; inspecting the model. The invention provides the intelligent nondestructive detection method; the precision and the applicability of the model are improved; high automation and intelligence are realized in the whole detection process; the wide market prospects are realized.

Description

technical field [0001] The invention relates to the technical field of dynamic detection of egg quality, in particular to an online nondestructive detection method for blood-spotted eggs based on spectral amplitude space conversion. Background technique [0002] my country is a big egg production and consumption country, but the export rate of eggs is low and the grading accuracy is not high, which seriously affects the development of my country's egg industry. In order to ensure the quality of egg products entering the circulation field, we are urgently required to improve the technical level of egg sorting. Online detection of blood-spotted eggs is a necessary prerequisite for egg quality grading. The U.S. Department of Agriculture's quality classification for eggs states that eggs with blood spots no larger than 0.32 centimeters (0.13 inches) in diameter must be classified as Grade B. my country's grading standards for eggs stipulate that AA, A, and B eggs should not con...

AI Technical Summary

Problems solved by technology

Zhu Zhihui, Xie Dejun, etc. used the method of multi-classifier fusion to improve the accuracy of the blood-spotted egg detection model, but the experimental environment of this method is static, and the established model is more complicated, which is not suitable for online detection.

In addition, there have been few reports on real-time online detection of blood-spotted eggs

Chen Meng used the binary logic algorithm to take dual-band modeling to realize the online detection of blood-spotted eggs. This method is feasible, but the test samples are artificially injected blood-spotted eggs, and the detection effect of natural blood-spotted eggs is unknown.

The Bayesian discriminant model established by Li Wanqing using the spectral characteristic variables optimized by the CARS algorithm avoids the influence of spectral characteristic variables on the model, but this method ignores the influence of spectral signal acquisition speed and different batches of eggs on the model’s discrimination accuracy

[0004] Due to the difference in egg batches and spectral acquisition speeds, there are cases where the spectral features (trends) are the same but the spectral amplitudes are different, which affects the accuracy of the algorithm in judging the samples. In order for the algorithm to identify samples with different amplitudes but the same Therefore, it is necessary to provide a new detection method to reduce the interference of speed, batch, instrument and environment during the experiment, so as to strengthen the adaptability and promotion ability of the model

Images