A Uniformity Evaluation Method Based on Standard Deviation Quantile Percentage

Abstract

The invention provides a uniformity evaluation method based on the standard deviation quantile percentage. The method comprises steps as follows: S1, tobacco leaf sampling; S2, acquisition of chemical values x of a tobacco leaf evaluation index; S3, the mean value xm of x is solved; upper-lower limits [Up, Lo] of confidence probability are calculated; S4, chemical values beyond [Up, Lo] are calculated, and the standard deviation of the chemical values within [Up, Lo] is calculated; S5, the chemical values x are subjected to sort descending, and y is obtained; S6, standard deviations of all data are calculated; S7, the serial number of the last data of y is set as n, the first data are rejected, the standard deviation sl is calculated, n data are rejected, the standard deviation s2 is obtained, the process is repeated k times, and PSD vector is obtained; S8, a numerical value interval is selected, p is taken as a data interval, and preset standard deviation is set; S9, the number of least rejection data of data with PSD smaller than the numerical value of a given interval and the percentage of the number accounting for the total sample number are calculated. Compared with an existing evaluation method, the evaluation method is more objective.

Description

technical field [0001] The invention relates to a uniformity evaluation method based on standard deviation quantile percentage, which belongs to the field of tobacco quality control. Background technique [0002] With the gradual improvement of people's requirements for the quality of tobacco leaves in recent years, homogenization production is more and more used in the field of leaf threshing and redrying. The index is generally measured by the ratio of the standard deviation of the chemical value to the mean value of the chemical value as the coefficient of variation. At present, the homogenization control level of most leaf threshing and re-baking is generally controlled at a coefficient of variation of 5%. For a special agricultural product, the average chemical value, batch size, and sample size are all difficult to be consistent. Generally, the size of the standard deviation is used, but in the process of comparing the standard deviation of different batches in differ...

AI Technical Summary

Problems solved by technology

[0002] With the gradual improvement of people's requirements for the quality of tobacco leaves in recent years, homogenization production is more and more used in the field of leaf threshing and redrying. The index is generally measured by the ratio of the standard deviation of the chemical value to the mean value of the chemical value as the coefficient of variation. At present, the homogenization control level of most leaf threshing and re-baking is generally controlled at a coefficient of variation of 5%. For a special agricultural product, the average chemical value, batch size, and sample size are all difficult to be consistent. Generally, the size of the standard deviation is used, but in the process of comparing the standard deviation of different batches in different rebaking factories, there will be little difference between the standard deviation of 0.07 and 0.08, and the coefficient of variation is 3.5% and 4% at the same mean level There are also great differences; some of them have very low standard deviation due to the small amount of sampling and concentrated sampling, which makes the evaluation unfair; The patent application number published by the company Shanghai Tobacco Group Co., Ltd. is 201610280750.8, and the name of the invention is a method for testing the homogenization of threshing and redrying based on the effect of homogenization control in threshing and redrying, based on the process and Whole, global and local The homogenization is evaluated in three parts, the overall coefficient of variation, the local variation coefficient and the evaluation system of the covariate coefficient of variation. The above method is a more detailed attempt to explore the homogenization evaluation system, which can quickly The analysis of the overall stability level and whether there is any abnormality in the process, and the source of the role of regulation, but in the above process of evaluating the homogenization of the chemical value of tobacco leaves, the influence of abnormal data on the entire homogenization evaluation is ignored. A batch of very stable chemical value data, but occasionally there are some outlier chemical values. This situation will easily increase the standard deviation and coefficient of variation. In the actual implementation process, it is difficult to distinguish which are indeed chemical components. Abnormal, the criteria for judging abnormal data is not yet clear

Obviously, the data that is very stable for the above data is just abnormal, not unstable

Images