Research On Quality Decision And Optimization Of Array Antenna Assembl

Intelligent manufacturing is an important direction of electronic product manufacturing,and it is of great significance to promote the development of the electronic information industry.At present,array antenna has become a key component in the fields of space exploration,navigation,and national defense,and has an important position at the national strategic level.As a complex electro-mechanical product,array antenna involves numerous parts and assembly procedures,with complex manufacturing processes,making it difficult to make scientific assembly quality decisions and to implement refined management on assembly activities.On the hand,due to its requirement of the high precision and strong electro-mechanical coupling,the sequence of assembly is complicated,changeable,and difficult to control.As a result,the assembly performance often fails to meet the design requirements,requiring continuous rework and reassembly,causing delays in assembly delivery.In this situation,it is urgent to introduce a comprehensive,accurate and efficient assembly quality decision-making method and an effective assembly process control to achieve the improvement of the assembly quality of the array antenna.Regarding the issue above,this paper proposes a digital assembly quality decision model for the assembly activities of the array antenna.Based on the principle of the decision model,a feed-forward optimization strategy is proposed for the assembly process of the key components.The main contents in this paper are as follows:The selection and quantification of assembly process feature index is the key to make digital decision.In this paper,combining the feature of the assembly process,in view of human factors engineering,equipment status,material quality,assembly process,measurement and environment in assembly workshop,a multi-feature index evaluation system for assembly quality is established.After summarizing methods of relevant literature,the corresponding scoring standards for invisible feature is proposed.Due to the variable and redundant feature of assembly quality samples,random forest algorithm is used to screen out the feature that have a greater impact on assembly quality,and assign them corresponding index importance score.Then,the gaussian mixture model is used to cluster the historical data in assembly activities,and the assembly quality index corresponding to each cluster is calculated according to the importance scores,thus enabling the classification of assembly quality grades.The antenna sub-array unit is one of the core components of the array antenna,and its assembly precision greatly affects the overall working performance of the array antenna.In order to reduce the assembly error of the sub-array unit,shorten the assembly cycle,and improve the assembly quality of the array antenna,this paper establishes an online prediction model of assembly error based on Auto-Encoder and Boosting-OSKELM(online sequential extreme learning machine with Boosting).With the goal of minimizing the assembly error,the NSGA-Ⅱ algorithm is applied to adjust the assembly craft to achieve the feed-forward optimization on assembly process.This paper analyzes and studies the assembly quality decision and optimization problems of the array antenna,providing ideas for quantification,optimization and intelligence of the assembly process for complex electro-mechanical products.