Method Development And Supporting Program Design Of Paper Product Performance Detection And Quality Analysis Based On Multi-dimensional Data Processing Technology

Paper product parameters are the basis for product quality evaluation and control in the paper industry.However,the existing detection methods in the industry have not kept pace with the continuously updated scientific development.Since most paper product parameter evaluation problems involve complex multi-variables,traditional methods have been unable to meet the requirements for comprehensive evaluation of paper quality.Machine learning is a modern advanced computer data analysis method.The biggest feature is that it has self-learning ability.It can improve the adaptability of the algorithm by simulating the learning behavior of people,and continuously learn during the calculation process to enhance accuracy.So far,a large number of algorithms adapted to different subject areas and different problems have been proposed,such as neural networks,support vector machines,etc.,which are very typical data analysis and processing methods.Therefore,it is of great significance to apply machine learning algorithms to the evaluation of paper product parameters,and use machine learning algorithms to be good at processing complex data and establishing correlations.The main research contents of this article are: Use Matlab commercial mathematics software to build models and Lab VIEW graphical programming software to write programs and compile interfaces that are easy to understand and operate;A method for detecting the content of migratable fluorescent whitening agents in paper based on automatic correction technology was established: the effect of interference spectrum was subtracted by mathematical calculations to calculate the content of fluorescent whitening agents in samples.The relative standard deviation of this method is 3.7%,and the recovery rate is between 99.1% and 107%.It can be used to quickly determine the content of migratable fluorescent brightener in unknown interference paper products with high sensitivity and reliable results;Established a diaper "greenness" evaluation model based on the life cycle assessment: Aiming at the needs of the establishment of a multi-indicator quantitative analysis model,a number of evaluation parameter indicators in the diaper product,raw materials and process were selected.The model provides a fast and reliable evaluation method for the quality and safety of diapers;A fiber quality recognition model for tissue paper based on PCA and BP neural network technology was established: This model comprehensively evaluates the fiber quality of tissue paper by combining traditional detection methods and modern advanced PCA-BP neural network pattern recognition methods.The results show that the improved recognition model has faster recognition speed and better recognition ability for large quantities and complex tissue paper;A paper similarity clustering model based on multivariate analysis was established: mainly using the PCA and classic K-means clustering algorithms to cluster papers of the same quality and repeated corrections using different evaluation indicators.The development of this method is of great significance for the detection and classification of unknown sample quality levels.