اثر زمان برداشت بر تلفات ذرت و پیش‌بینی آن با منطق فازی در منطقه مغان

در این تحقیق تلفات برداشت ذرت با کمباین جاندیر مدل 1165 در منطقه مغان در شرایط آب و هوایی مختلف در قالب طرح کرت‌های خرد شده بر پایه بلوک‌های کامل تصادفی و در سه تکرار مورد مطالعه قرار گرفت. تلفات واحد فرآوری و واحد جمع‌آوری کمباین در تاریخ‌های 14، 17 و 20 آبان ماه 1396 و در سه زمان برداشت در طول روز ساعت 108، 13-11 و 16-14 با سه تکرار اندازه‌گیری شدند. نتایج تجزیه واریانس اختلاف معنی‌داری را بین تاریخ‌ها و زمان‌های برداشت در سطح احتمال 5% نشان داد. با توجه به مقایسه میانگین‌ها توسط آزمون دانکن، کمترین تلفات در تاریخ 17 آبان ماه ساعت 16-14 به مقدار 10.05% بود و بیشترین آن مربوط به 20 آبان ساعت 108 به مقدار 12.78% اندازه‌گیری شد. رطوبت بالای هوا و پایین بودن دمای هوا از دلایل افزایش تلفات بود. در ادامه تحقیق با توجه به تاثیر زمان برداشت در میزان تلفات، برای پیش‌بینی مناسب‌ترین زمان برداشت اقدام به طراحی یک سیستم خبره فازی گردید که در آن تلفات واحد فرآوری، واحد جمع‌آوری و رطوبت هوا به‌عنوان ورودی‌های سیستم و زمان برداشت به‌عنوان خروجی آن در نظر گرفته شد. برای ایجاد سیستم خبره فازی از موتور استنتاج ممدانی با فازی‌ساز منفرد و غیرفازی‌ساز میانگین مراکز استفاده شد. ضریب تبیین 0.98 نشان‌دهنده همبستگی زیاد نتایج پیش‌بینی بهترین زمان برداشت با مقادیر اندازه‌گیری شده بود. بنابراین سیستم فازی طراحی شده با دقت بالایی زمان برداشت ذرت را پیش‌بینی می‌کند.

The Effect of Harvesting Time in Moghan on Corn Losses and its Prediction using Fuzzy Logic

Introduction;Corn harvest losses are imposed by several factors, the most important of which is harvestingtime. Since the harvestingtime is coincident with the rainy season, it is necessary to appropriately estimate the corn harvest time to avoid harvesting losses and losing the next cultivation. Accordingly, in the current research, the effect of harvestingtime on corn losses during the month and the day has been into consideration. An expert fuzzy system was designed to predict the best harvest time as it operates based on the losses amounts which are measured in processing and collection units into the combine, and losses due to the humidity percentage.;Materials and Methods;In this paper, corn harvest losses in a John Deere Combine, Model 1165, was studied in a different climatic circumstance in Moghan region. Moreover, a split plot experiment in a completely randomized block design was conducted with three replications. The losses data were collected from the processing and collection units of the combine harvester on the November 5th, 8th and 11th, 2017, in three different daily times of 810, 1113 and 1416 with three replications. The Mamdani fuzzy inference system with singleton fuzzifire and center average defuzzifire‎ was used to develop a fuzzy expert system. In the designed expert system, the losses percentage in the processing and collection units and ‎the humidity percentage were considered as system inputs and optimal harvesting time was used as the system output. Low, Very low, high and very ‎high and ‏Best, Suitable, Unfit, and Worst were four groups of linguistic variables for input and output parameters, respectively. These variables follow the triangular and trapezoidal membership functions. The number of 64 fuzzy rules were considered and introduced into the fuzzy system by experts, experienced farmers, and combiners. Furthermore, the same field data (measured data) were applied to evaluate the designed system, so that the predicted value was accounted as the system output.;Results and Discussion;Analysis of variance showed that there was a significant difference between the harvesting dates at the 0.05 probability level and significant difference between the harvesting times of a day at the 0.01 probability level. It can be concluded that the harvest dates and harvest times of a day were very effective in the number of corn losses, but the interaction effects were not significant. The results appeared that the lowest losses were 10.05% on November 8th, 2017, at 1416 p.m., and the highest losses were 12.88% on November 11th, 2017, at 810 a.m.‎ The amount of losses was increased due to the higher air humidity and lower temperature. In the fuzzy simulation model, the suitable harvestingtime can be predicted based on the losses quantities in the processing and collection units and the humidity percentage. The results showed that the predicted values for harvestingtimes, by a designed fuzzy system, were completely matched with measured values in this study. The coefficient of determination (R2) was 0.980 between measured and predicted harvesting times. This coefficient demonstrated that the developed fuzzy logic system was suitable for prediction of harvesting time in the studied area.;Conclusions;The experimental observations in the field and data analysis showed that in the corn harvesting in the Moghan region, the humidity level, date, and harvestingtime were the most effective factors in the harvesting losses. In this paper, based on measured data from a small farm and implementation of the expert fuzzy system, the most suitable harvest date was set on November 8th at 1416 p.m, at 2124°C and relative humidity of 44%53% to have 10.5% losses which has been confirmed by the lowest losses observed in the corn plan (10%). Moreover, the high value of the determination coefficient demonstrates a high correlation between measured and predicted data.