Research On Quality Management System Of Tractor Manufacturing Process Based On Responsibility Card Identification And Network Technology

The production and use of tractors is an important indication of the level of agricultural mechanization.At present,domestic tractor manufacturers lack competitiveness in the high-end tractor market due to low product quality.The insufficient quality management in the whole machine manufacturing process restricts the development of domestic tractor products to high quality.Therefore,a scientific and effective quality management system for the whole machine manufacturing process plays an important role in improving tractor quality and enhancing product competitiveness.The quality management system of tractor manufacturing process is based on complete quality database and quality management tools.The construction of quality database needs to collect the quality data generated by each quality control point.Some quality data of quality control points are recorded on the quality responsibility card,which does not realize the data electronization.Meanwhile,some quality control points which realize data electronization store the quality data separately in different forms,which makes the data cannot be shared.The traditional quality management tools are based on the principle of mathematical statistics,which can only analyze the quality data simply,but can’t obtain the deep quality law.The specific research contents and results are as follows:(1)In order to solve the problem of recording tractor whole machine quality data with paper documents in assembly stage,a quality data electronization method based on responsibility card identification is proposed.Firstly,the quality responsibility card is redesigned to standardize the filling of quality data;Secondly,the Paddle OCR text detection and recognition model is used to obtain the coordinates of quality data attributes keywords,and realizes the location of quality data recording area;Thirdly,a numerical character segmentation method based on stroke restoration and connected domain marking is proposed to obtain a single image of number from the quality data image;Fourthly,a CNN recognition model is trained to recognize handwritten numbers,and the recognition accuracy on the test set reaches99.22%;Finally,the software for quality data electronization is developed and the manual correction function is provided to ensure the accuracy of quality data in the database.(2)In order to solve the problem of integration of heterogeneous electronic quality data of testing stations in the whole tractor debugging stage,an automatic data reading and processing program is written to extract the structured quality data stored in database and unstructured quality data stored in text file on different industrial control computers in real time.A data transmission interface based on JSON format is designed,and the quality data is sent to the database server through HTTP protocol.The E-R model is used to build a quality database in the database server to store quality data from different sources in a unified structure.(3)In order to solve the problem of lack of high-level quality management tools in tractor manufacturers,according to the characteristics of quality data,a quality management tool based on case-based reasoning and machine learning technology is developed,which provides the functions of fault diagnosis,product quality classification,quality comprehensive index extraction and fault prediction.It is integrated into the quality management system together with traditional quality management tools and completes the development of tractor manufacturing process quality management system.The system has been applied to the actual production of enterprises successfully and improved the level of quality management informatization.