Analysis of the research results of time series of chemical properties, that appears during the printed board construction

This article contains results of research of time series of industrial wastewater chemical properties, that appears during the printed board and electronic blocks construction. Printed boards are connect all electronic devices with each other, include automated communication systems and control systems. Technological process of printed board construction include sequence of chemical treatments. Realization of this operation during manufacturing is pretty hard because chromium compounds are toxic and waste disposal measures needed. Stable wastewater chemical composition on the one hand confirm the technological process was uninterrupted, but on the other hand allow to apply a specific set of purification methods. Values of periodic chemical properties measures forms the time series, that uses to technological process research. For stability control of any process, at first is necessary to research its characteristics, build descriptive models, and check their quality. According to the received characteristics, using statistic analysis methods, we can obtain the process model. As the source data used measured chemical properties of industrial wastewater from two wells (106 and 127) during several years. For calculations used statistical software “STATISTICA” and some software developed for this case. Analyzing the average values and their dispersion, we can confirm the conclusion that the wastewater composition in different wells differs significantly. Differences in the values of indicators in different wells can also be explained by changes in technological processes or wastewater filters working. Also noted medium power correlation between the indicators of acidity and phosphates. The remaining indicators of industrial wastewater chemical composition are not related to each other by linear dependence. Before building a forecast for selected chemical indicators, non-stationary time series were reduced to stationarity. Of all the models built for acidity indication ARIMA (2,0,1) selected as best. To build regression models as response (dependent variable Y) used total wastewater volume in selected well.