Разработка и настройка алгоритма программной идентификации отклонений в пенном слое калийной флотомашины

Abstract

The article deals with the problems that arise in the practical application of the method of glare recognition of the parameters of the froth layer of a potash ore flotation machine. These include choosing the best statistical characteristic, filtering and averaging settings, checking the possibility of taking into account anti-glare, and the need for adaptive signal renormalization. Objective of the study is to develop an algorithm for identifying a flotation cell as an object of automatic control, which is possible only after solving these problems. Materials and methods of research. The study was carried out on the materials of an experimental survey of a potash flotation machine. During the shooting, a standard step signal was applied to the machine cell, which was expressed in a change in the composition of the amino-oil mixture, which caused a transient process. For different statistical characteristics (the number of bubbles, the number of red components in the frame, the average and median distances between the centers of the bubbles), various methods of filtering and averaging the data were tested. At the same time, the problem of identifying the gain and the time constant of the object was solved. The best characteristic and methods of its processing were chosen on the basis of the root-mean-square deviation of the calculated transient process from field data obtained by glare recognition of the foam layer surface. Results. Processing several frames in a row, taken at the same position of the foam, slightly improves the result, but significantly loads computing power. Blind filtering by ten points has practically no effect on the data processing time. To improve the identification, data renormalization was used, which consists in the adaptive selection of the zero and single signals in the conditions of noisy data. The localization method was used to determine the optimal delay from the point of view of the quadratic deviation before the start of the transient process. Conclusion. As a result, the best filtering and data averaging settings were obtained, providing the smallest identification error. The time constant of the cell turned out to be close to the results of previous authors, obtained, among other things, by visual observation of the flotation machine. Accounting for antiglare does not significantly affect the parameters of the object. An important conclusion is that one statistical characteristic describes well the beginning of the transient process, and the other describes its end. This must be taken into account when building a deviation signaling system.