Research On Selective Assembly Method For Key Components Of Dynamical Servo Gyro Based On Drift Performance

The dynamical servo gyro is the core part of IR seeker,and its drift performance plays a decisive role in the tracking ability of IR seeker.Through selective assembly for gyro components,the drift performance can be improved.At present,the selective assembly for the key components of gyro is still mainly based on manual selection,but also subject to the drift performance,resulting in the low success rate of one-time selective assembly.In order to improve the assembly efficiency of dynamical servo gyro,this paper starts with the gyro drift performance,through the analysis of gyro assembly data,combined with assembly characteristics of the key components of gyro,establishes a selective assembly method based on residual networks with gated recurrent unit(RNGRU)for the key components of gyro.Aiming at the selective assembly between gyro rotor and drift-adjusting screw,a selective assembly method based on comprehensive grey relation order model and RNGRU regression model is proposed.In this method,the assembly parameters of gyro rotor are analyzed by using comprehensive grey relation order model,and the key assembly parameters that affect the selective assembly between gyro rotor and drift-adjusting screw are determined.Based on the key assembly parameters of gyro rotor,RNGRU regression prediction model is established to predict the mass of drift-adjusting screw and the drift performance of gyro rotor after selective assembly with drift-adjusting screw.The experimental results show that the proposed method can achieve selective assembly of gyro rotor components,and the RNGRU model has higher prediction accuracy in the experiment compared with traditional GRU model.Aiming at the selective assembly between gyro rotor and coil packages,a gyro assembly prediction method based on statistical classification of drift performance is proposed.In this method,drift performance data are classified by the combination of box-plot and range.Based on the results of statistical classification,the importance of gyro assembly parameters is analyzed by the method of Random Forest feature analysis.At the same time,RNGRU classification model is established to predict the drift performance of the assembly between gyro rotor and coil packages.Finally,the efficiency of the prediction method is verified by experiments.For gyro batch selective assembly,a selective assembly batch method based on elitist genetic algorithm(EGA)for the assembly between gyro rotor and coil packages is proposed.In this method,RNGRU classification model is used to establish the constraint conditions of the assembly between gyro rotor and coil packages.The batch selective assembly optimization model for gyro is built with the highest selective assembly success rate as the goal.EGA is used to solve the batch selective assembly optimization model.Experiments show that the proposed method can achieve gyro batch selective assembly and reduce the number of invalid assembly and surplus.Based on the selective assembly method studied in this paper,combined with the production requirements of enterprises,a gyro key component selective assembly prototype system is designed and developed to assist assembly personnel in the selective assembly for the gyro components.