Fulltext search in archive

The data taken at the electronic reception of logs are tasked with mistaken values and do not correspond to the real shape of a stem. The aim of the data filtration is to remove the incorrect data and replace them by the ones, closer to the real values.
The goal of presented work was to analyse the effect of different filtration methods on original data and to recommend the methods, which results correspond to the real shape of the logs with their defects on a surface as close as possible.
Methods of filtration are based on simple mathematical and statistical procedures, which are subsequently variously combined, because the simplest comparative methods did not fulfil the expectations.
More than 50 methods were combined and analyzed. Approximately 15 of them were selected and tested. The methods were applied to data scanned on ca 150 logs randomly selected from thousands of files in the sawmill log yard. The filtration effects of different methods were visually assessed. The most applicable three methods were selected and they will be tested practically at sawmill in the next step of research. These three methods are described in following paper.
Results refer, that all proposed methods correspond to necessities for log dimensions determination. It is not possible to define the best method generally. Different properties of these methods make them suitable for calculation of log dimensions if different type of use is requested. It means for mid diameter determination necessary for volume calculation by using Huber method; for section diameter determination for volume calculation by using section method; for top diameter determination necessary for sorting; for filtration of drive dogs of the conveyor passing through the measuring equipment.

Essential part of raw milk component measurement is indirect infraanalyse. The checking of time measurement stability is important there. The preparation of pilot samples was described. Information about deep frozen milk pilot sample stability are sporadic. Aim of this work was to verify the stability of long-term stored and deep frozen pilot samples (FPSs). Pilot samples were frozen in liquid nitrogen bath -196 °C and after that stored at -21 °C for one month (A) and more months (B) till analyse. The pilot samples were measured by CombiFoss 6000: fat F (%); protein CP (%); lactose L (%); casein C (%); solid non fat SNF (%); urea U (mg/100ml); somatic cell count SCC (ths./ml). In the short-term (A) experiment the impact of freezing on pilot milk samples was relatively small, in the framework of acceptable value of repeatability (±0.02 %) for components. The stability of values of FPSs was better according to standard deviations for important F evaluation in H (Holstein) breed samples. The repeatability of values of FPSs in the time was very good for other components. The repeatability was better for SCC in J (Jersey) breed samples (±16.5 thousands/ml). That is why the higher variability (±30.1 thousands/ml) in H sample is visible less sceptically. The apparent result trends were not observable in the repeatability development of important milk indicators in FPSs. The mild trend was only in CP, but this trend covered absolutely very small shift (variation range). There are only oscillations, which are similar between J and H samples. It is possible to attribute these oscillations rather to instrument effects than to sample effects. The oscilations were markedly higher in the long-term (B) experiment. Between J and H samples the oscilations were very similar in curves F and L. It is also possible to attribute them more to instrumental effects. The repeatability values (standard deviations sd) of FPSs were mostly lower than ±0.06 %. It is acceptable for result agreement between indirect and reference method. Relevant sd varied from ±0.025 % for L to ±0.059 % for F (in J breed) with exception CP ±0.085 % in J sample. Also B procedure was shown as applicable for instrumental sability control in half year period. In practice the one reference value for concrete FPS is valid from one to next calibration. Persistent trend in repeatability of FPSs can indicate an inadequate instrumental drift. The applied FPS procedure is usable for stability control of instrumental measurement in milk laboratories.