Research On Platform Of Assembly Data Mining In Complex Mechanical System

Nowdays most of the assembly process of complex mechanical system is designed by existent simplified model. The working mechanism between the key ingredients and assembly performance is still unknown and some assembly procedures rely heavily on the experienced technician. In sake of the wide application of manufacturing information systems where huge amount of assembly process data saves, useful assembly experiential function may hide in the data. Assembly data integration based on XML is proposed and Data Mining(DM) technology can be used to find the relation between assembly parameters and performance. Algorithm of data mining can be integrated by Python and assembly data mining platform can be developed. The main research contents and results are as follows:1. The existing possible reasons and ingredients which may cause impacts on the final performance of the assembly are analyzed. The necessity to data mining on the assembly data is proposed. The construction and basic process of assembly data mining is established.2. To solve the problem that the data collected from databases and documents of the manufacturing information systems for different departments in the same corporation or different corporations may be inconsistent, a global database can be established to raise the unified data model. And XML document is used to be the carrier of assembly data to realize integration technology.3. To work out the problem of data vacancy and mistakes, Python script language is adopted to be the algorithm tool and data preprocessing module is integrated to fill the imperfect data and noise data. The data mining algorithm warehouse based on Mlpy package and data mining module is established to achieve setting, analyzing and visualizing assembly data mining.4. Based on the structure of assembly data mining construction, the platform of assembly data mining for missile dynamic alignment is designed. The function for maintaining the basic data, inserting assembly data and data mining is demonstrated. Finally an actual example is put forward to realize regression analysis based on the platform.