Research And Design Of Width And Height Detection Instrument Based On K-means Clustering Algorithm

With the development of the national economy, China accelerates the construction of road network. While the road transport industry develops rapidly, the problem of overloading in the vehicle also arises. It has been very urgent to detect the vehicles on the road whether is overrun and overloading, therefore, vehicle width and height detection system begins to develop, and it plays a very important role as a part of the vehicle overrun and overloading detection system. Vehicle width and height detection developed from traditional manual detection to using laser sensor detection whose measurement results are more intuitive, and the measurement accuracy have been greatly improved. However, currently vehicle width and height detection system which using laser sensor generally consists of laser sensors and PC, the PC directly deals with data returned by laser sensor, which resulted in the system’s not strong cohesion. It is difficult to be embedded into overrun and overloading detection system as a module and the measurement error is large. At the same time, it needs to individually configure a computer to the system which will increase the system cost.To solve the above problem, this paper designs a kind of width and height detection instrument based on K-means clustering algorithm. The ARM embedded industrial control board is used as the instrument master core, and the WinCE operating system is embedded into the instrument. The instrument uses MFC(Microsoft Foundation Classes) under the environment of Microsoft Visual Studio 2008 to realize modular programming which coordinates every hardware module of instrument to receive and process laser sensor data, and uploading vehicle width and height data. Process the data of vehicle width and height by using K-means clustering algorithm equivalent to current vehicle width and height detection system using laser sensors to improve the measurement accuracy.The main contents of this paper are as follows:First, the paper introduces the research background of the width and height detection instrument and the research status of the related fields at home and abroad.Second, the components, workflow, system structure and the method of calculating vehicle width and height in the system which using the laser sensors is discussed.Third, data mining and K-means clustering algorithm of the relevant theoretical knowledge is learned and introduced, including the principle of data mining process, tasks, K-means clustering algorithm, processes, advantages and Matlab realization etc..Fourth, according to the functional requirements, the hardware structure of the width and height instrument is designed, and the design of the main control board module, switch module and LED instruction module is described in detail.Fifth, research the embedded Win CE system, write instrument functional module programs and dynamic link library based on WinCE system, design parameters configuration tool of instrument.Sixth, the data collected by the laser sensors are analyzed, and the K-means clustering algorithm is applied to the data processing process. The experimental results are analyzed.Seventh, summarize the research content of this paper, and look forward to the development of the width and height detection instrument.The main innovation of this paper:First, the WinCE operating system is applied to the width and height detection instrument design, reducing the development threshold of application software, and the portability of the program and the stability of the system are enhanced.Second, the instrument has optical fiber interface so that it could upload the vehicle width and height data to PC through optical fiber, extending the distance of data transmission.Third, the K-means clustering algorithm is applied to processing width and height data, and improves the measurement accuracy compared with the current vehicle width and height detection system which using laser sensors.This instrument has been applied to practical engineering project, and runs stably and normally in a number of highway overload testing station of Shanxi Province. Using K-means clustering algorithm can effectively filter out interference points, and when the speed in the range of 0~20km/h, single laser sensor vehicle width and height detection system height measurement realizes error within ±10mm and width measurement error within ±20mm; double laser sensors vehicle width and height detection system height measurement realizes error within ±5mm and width measurement error within ±10mm.