Development Of The Information Collection Terminal For Agricultural Machinery And Research On Evaluation Method Of Deep Scarification

China’s argiculture has a large size,but it is not competitive in the world.There is still a long way to go to realize argiculture modernisation in China.In this case,the modernisation and informatization are the main development for argiculture in the 13 th Five Year Plan(2016-2020)of the country.Compared with the conventional farming,the modern argiculture no more rely on the weather.It firstly uses modern technology to collect the information of the farmlands and machinery,and then propose the personalized planting and cultivating plan for each farmland.It can enhance farmland’s output per unit area with the large argiculture mechanics and promote substainable development.Information collection is the first step in the implementation of the precision agriculture.At the same time,the large agriculture machinery takes on the most work in the farming,the intelligent machinery is so important for the argiculture modernisation.This thesis firstly focuses on the information collection of machinery based on the deep scarification.Deep scarification uses the mechanical shovel to loosen soil,so it will be easy for the absorping of water and fertilizer.The travel route,the work quality and the mechinery movement are the main concerns for the deep scarification.Based on the major information collected,we propose the deep scarification evaluation methodology.First of all,the thesis introduces the design of the information collection terminal for agricultural machinery used in the deep scarification.Its main functions include data uploading and local backups except data collection.We analyze the function requirement and then design the entire modular framework.The design of module software is also described.The main work is follows: The trajectory data,depth and movement data which are collected by the sensors will be integrated into a data frame.And the data frame will be sent to he remote server through the GPRS network and the data can be stored in the SD card at the same time.Secondly,we analyze the captured data including data preprocessing and feature extraction.Data preprocessing includes coordinate transformation and trajectory segmentation for the trajectory data,filtering and acceleration modifying for movement data.After the data normalization,we will extract the the degree of the track regularity,dangerous driving index,fuel consumption and depth variation coefficient as the features.These features will be used to constract training sample set.Finally,we use ranking support vector machine which is a method of learning to rank to get the ranking model by training the dataset.And improve the model accurary by optimizing the parameters of kernel function.