Study On Time-domain Evolution Characteristics And Controllable Mechanism Of Atmospheric Argon Plasma Streamer

Atmospheric low temperature plasma has been widely applied in materials processing,biomedical and environmental engineering applications because of their properties.As an important process of atmospheric low temperature plasma,the process of the morphological evolution of the streamer is important for the application of plasma.During the process of evolution,there are many external factors influencing the morphology of plasma streamer,and the correlation between theory and experimental phenomena is complicated,whether the morphology of streamer at atmospheric pressure is controllable or not is still controversial in theory,and it is difficult to carry out research on the specific methods of control,therefore,the study on the controllability of the morphology of plasma streamer and the method of control for it is a very important study.Based on these problems,we use a high-speed camera to shoot the streamer and build a data set.A deep learning technology is applied to study the correlation between the discharge parameters and the morphology and explore the main influencing factors and controllability of the streamer.(1)In order to study the morphological characterization of streamer in the evolution process,we increased the dimension of parameters by changing the voltage,frequency,gas flow,electrode structure and electrode materials to,and used high-speed cameras to photograph the processes.The experimental results show that under these different conditions,the local characteristics of the morphology evolve randomly,but the overall morphology evolves regularly with the change of the discharge conditions.(2)In order to understand the distribution of the morphological feature point and the problems of stochastic of streamer,the SIFT algorithm was used to extract and match features of streamer in this paper,and the characteristic distribution of its morphology was analyzed.The final results show that there are four different evolution states of streamer under the same discharge condition,and it changes periodically,moreover,the feature matching degree between image data of different evolution states is low,and the feature matching degree is high between image data of the same evolution states,and the characteristic point matching threshold of the streamer varies between 20 and 148.In addition,the feature matching degree of image data with different discharge conditions and the same evolution state is still high,but lower than that of image data with the same discharge conditions and evolution state.Therefore,it can be proved that the morphology of plasma streamer does not belong to random variation during the evolution process and it changes periodically,and the image data can be distinguished from the feature point distribution.(3)Aiming at the problem of control the morphology of the streamer,the deep learning network(Res Net)was used to analyze the sorted image data set.The discharge condition of the image data is taken as the label of the data.By using a network model to train and analyze the image data,we can learn the complex relationship between the form of streamer and the discharge conditions,and finally recognize the corresponding discharge conditions of the image data,the final experimental results show that deep learning network can be used to identify the plasma streamer image data within a certain range of changes in discharge conditions.Therefore,the experiment proves that deep learning technology can solve the problem of the controllability of streamer.The paper concludes that the morphologic evolution of streamer at atmospheric pressure under different discharge conditions is complicated,and the local characteristics of the streamer show some random evolution characteristics,but the overall morphology of the streamer shows a evolution law with the change of the discharge conditions,and the SIFT found the change of the overall morphology of streamer under the same discharge condition is not a random evolution process,but a periodic change,and a deep learning technology can be used to prove that the morphology of plasma streamer is controllable.