On-line Detection Of Pre-conditioned Cocktail Ingredients Based On Raman Spectroscopy

In the process of pre-cocktail production,some products cannot be accurately detected by the existing online detection equipment,which has problems such as poor detection accuracy and slow detection speed.Based on the above problems,this paper proposes a fast online quantitative detection of pre-cocktail production based on laser Raman spectroscopy combined with partial least square（PLS）prediction model.First,the characteristic spectra curves of ethanol,fructose,glucose and sucrose were analyzed.Polynomial moving average filtering method,wavelet transform filtering method and de-baseline algorithm were used for spectral preprocessing,and the PLS quantitative analysis model was established by selecting the characteristic peak segment of the substance to be measured.Secondly,the model is analyzed and optimized in offline state,Polynomial moving average filtering method and wavelet transform filtering method are adopted to select the 72 points at the characteristic peak after the baseline is removed.Combined with the PLS prediction model,the calculation amount is significantly reduced and the detection efficiency is improved under the premise of ensuring the prediction accuracy.Finally,the wavelet transform filtering method was adopted online.After the baseline was removed,60～1790（88）^-1continuous spectrum was selected and the PLS prediction model was combined to achieve better accuracy.After abandoning the performance of qualitative analysis of the four substances,the PLS prediction model based on the characteristic spectrum of the product used the single-component detection mode to obtain the best detection accuracy.The experimental results showed that the detection accuracy of Raman spectrum combined with PLS detection model was within±0.05%and the detection time was1.53 seconds on the products with different component proportions.Compared with the existing detection technology Anton Paar,the detection accuracy of the existing online detection equipment on specific products is±0.33%and the detection time is 10seconds,so Raman spectral detection technology has obtained obvious advantages in the detection accuracy and detection speed.