Development And Application Of Automatic Inspection System For Bearing Factory

The target implementation of this article’s testing system is a factory that produces a series of miniature deep groove ball bearings with ultra-small apertures,which are mainly used in unmanned aerial vehicles,laser projectors,and monitoring pan-tilts.Due to the more demanding working conditions of its supporting products(relatively large working temperature range,long continuous working time,high reliability requirements),this enterprise has strict control and requirements for raw material reception,production process,and process parameters.However,in the specific production organization,quality documents are still recorded manually and in electronic spreadsheets,with no unified management.When encountering customer complaints,the process quality data of specific batches are traced back and the time required for retrieval is measured weekly.From the perspective of production quality control,it is necessary to upgrade the original testing and inspection methods and means to meet the current practical needs of automation and intelligence.Based on the production status and degree of informationization of the target implementation enterprise,a three-part plan for automated implementation of the testing and inspection system is proposed:(1)Digitization of raw material inspection documents,compatible with the original work method,and adding inspection document digital recognition and formatted storage to facilitate the traceability of raw material batch numbers.A mobile APP has been developed to upload images of raw material inspection documents to the host computer for recognition and analysis: pre-processing operations are performed on table image to improve recognition accuracy;morphology and Harris corner detection methods are used to extract table frames,and a dynamic masking method is designed to recognize areas of interest.A specific character library is trained using the j Tess Box Editor tool to recognize characters in the cells and fill them into a dedicated digital table.(2)Automated collection of production quality data,directly uploading real-time detection data from numerically controlled units on the production line that have a bus interface to realize automated reading and storage of process quality data.The machine networking system is based on self-owned communication modules and acquisition modules to achieve physical isolation between the industrial control system and the Ethernet system.Communication between modules and the industrial control system is based on the specific PLC model and uses the Fins protocol.A micro-service for data receiving and uploading based on Python has been developed,which,combined with the MQTT message publishing / subscription mode,pushes quality data to the server-side for storage,facilitating the implementation of statistical process control.(3)Statistical process control analyzes process quality data and displays it on a large screen to provide real-time feedback on the production line.Combining the basic data of the MES system previously deployed by the enterprise and the production line quality data,a data mapping method is designed to match a large number of data types with corresponding SPC calculation processes.An X-R control chart is drawn to analyze the stability of product quality,and a Pareto chart is drawn to analyze the causes of product nonconformity,thus improving the ability to control product quality.A television APP has been developed to visualize associated data and display real-time production process quality information.This research project has been formally deployed and applied in the target enterprise.The comprehensive accuracy of raw material document recognition has reached 96%,and numerically controlled equipment uploads detection data in real-time,with analysis results displayed in a dedicated client and television APP in the form of control charts and other visualizations.The networking and digitization of quality data have been realized,solving the traceability problem in the quality system,and real-time feedback on on-site quality management has been achieved based on data visualization.