Research On Multi-stage Data Fusion And Algorithm Based On Gas Gun Experiment

The gas gun plays an important role in high-load experiments.With the continuous improvement of experimental indicators,higher requirements are put forward for the gas gun experimental platform.Due to the wide application of gas gun experiments,different experimental purposes do not deal with the same data.Systematic and complete experimental data processing methods are needed.Aiming at the air cannon experimental data,this topic uses multi-stage data fusion theory to make the study and designs a data fusion software system suitable for the air cannon experimental platform.According to the experimental principle of air cannon and the characteristics of experimental data,this paper studies in the following aspects:1?Pre-processing on velocity data: In the gas gun experiment,when using the coil target to measure the muzzle velocity,the wavelet theory is used to denoise the zero-crossing signal,after considering the interference of noise on the zero-crossing signal.The simulation results show that the method has good denoising effect for white noise and colored noise.When the muzzle velocity is measured by high-speed,it is caused by a large number of tiny particles at the muzzle.The frame difference method is difficult to extract,so the Laplacian pyramid algorithm is used to fuse the video image to extract the target speed.The feasibility and accuracy of the method are verified by the application of image fusion examples.2?Data level fusion of target muzzle velocity: this paper measures the velocity value of the target through the muzzle with a variety of speed measurement methods.In order to reduce the influence of inaccurate measurement data on the final fusion value,select the appropriate trust function to calculate trust.The degree matrix uses a trust matrix to assign weights to multiple measurement data for fusion.The experimental simulation shows that the method can still guarantee the fusion precision and has strong anti-interference ability in the case of single measurement data failure.3?Decision-level fusion of experimental data: Firstly,the influence of various parameters in the gas gun experiment on the target muzzle velocity is analyzed.The BP neural network model is used to fuse various parameters of the experiment,by adjusting the number of neurons,changing the learning rate and using regularity.The training method was used to optimize the BP model,and the specific parameters of the model were finally determined.The prediction of the muzzle velocity under a certain experimental condition is realized,and the magnitude of the pressure parameter required to accelerate the target to a certain fixed value is also predicted.4?Design data fusion application software: the software achieves the following functions.1)Data level fusion calculation of the target muzzle velocity;2)predicting the speed of the target at the muzzle under the conditions determined by the experimental parameters;3)in the case where the target weight,the high pressure chamber volume,and the length of the launch tube are determined,Predicts the amount of pressure required to accelerate a target to a certain speed value.