Студопедия

Главная страница Случайная страница

Разделы сайта

АвтомобилиАстрономияБиологияГеографияДом и садДругие языкиДругоеИнформатикаИсторияКультураЛитератураЛогикаМатематикаМедицинаМеталлургияМеханикаОбразованиеОхрана трудаПедагогикаПолитикаПравоПсихологияРелигияРиторикаСоциологияСпортСтроительствоТехнологияТуризмФизикаФилософияФинансыХимияЧерчениеЭкологияЭкономикаЭлектроника






The comparative analysis of the different methods






 

From Table 9 below, we can see that t he freight volume increases with an average growth rate of 13%.

The predicted performance test results are shown below (Table 10). If the GA-BP to 2012-2015 data most tests data, you can not use the last four years of data, resulting in a large error. Since 2013 year cargo traffic growth is slowing down when compared to 2012 year but has not been apparent in the GA-BP method. Genetic testing error neural network is relatively large, especially in 2013; the error value reached 11.71%. As can be seen in the relatively small amount of data, time series are large errors, and time series inspection error for year2011 and 2012 is over 12%. The average error of the time series is -9.157%, far higher than GM (1, 1) and GA-BP’s error of the mean. Gray theory predicts with high accuracy, GM (1, 1) fitting results were within the error range of plus or minus 7%, especially for more recent years a more accurate prediction, the prediction error values for 2012 only 0.032%.

 

Year Freight volume (thous.ton) Percent (%)   Year Freight volume (thous.ton) Percent (%)  
           
           
           
           
           
           
           
           

Table 9 Rail freight transit volume (tonnes) increase

 

 

Year Freight volume (thous.ton) Time sequence error (%) GM (1, 1) error (%) GA-BP error (%)
         
         
         
         
  Average      

Table 10 The prediction of the error values for each method

 

Conclusions

 

Research on the use of gray model prediction methods for the development of the trend of rail freight transit volume on the border between Kazakhstan and China, according to 2001-2015 cargo transport volume predicted, have residual GM (1, 1) model approach than the time sequence of their neural networks and genetic algorithms method is more accurate conclusions and predict the value of annual cargo transport volume from 2017 to 2020. Specific conclusions include the following aspects.

1. When the external economic environment, there is a greater change and hence the need to analyze the relevant impact. This paper bound by the impact of the Asian financial crisis in 1997, in order to make the predicted effects become more accurate.

2. According to the results of the relevant data, the fitting can know when a small amount of data, the gray system with high accuracy, while the effect of poor fitting time series. (dont understand this sendtence at all)

3. Calculated according to the relevant data, you can get the cargo transport volume 2017-2020 to maintain growth in 2017 is expected to reach 46, 654, 330, 000 tons, in 2018 will reach 51, 399, 860, 000 tons, in 2019 will reach 56, 628, 090, 000 tons, 2020 For freight traffic forecasts will benefit departments scheduling resources, early planning, to ensure the successful completion of cargo transport.






© 2023 :: MyLektsii.ru :: Мои Лекции
Все материалы представленные на сайте исключительно с целью ознакомления читателями и не преследуют коммерческих целей или нарушение авторских прав.
Копирование текстов разрешено только с указанием индексируемой ссылки на источник.