Crack Monitoring And Diagnosis In Laser Cladding Of Nickel Based Alloy

In the metal laser cladding process,many process parameters have complex effects on the properties and quality of cladding coating,and typical defects such as cracks and pores are easy to appear in the cladding process.These defects will greatly affect the physical and mechanical properties of the cladding coating,so real-time monitoring of each process parameter and the quality of the cladding layer in the cladding process is a relatively efficient and novel solution.The work done in this paper is the monitoring and diagnosis link in the above-mentioned monitoring and feedback adjustment process.The main research contents of this paper are as follows:(1)The current research background and significance of real-time monitoring and feedback adjustment in laser cladding process are systematically introduced.Monitoring scheme and research results of laser cladding process based on different online monitoring technologies.With the development and popularization of intelligent diagnosis technology,the application of machine learning,deep learning and other cutting-edge theories in defect diagnosis of laser cladding is introduced as well.(2)A method to distinguish the crack impact signal from the base molten pool activity signal is proposed.Firstly,a large number of monitoring signals with and without cracks in the cladding process were obtained through experiments.Then the energy level of the monitoring signals without cracks is counted,and the impact signals with energy higher than this level in real time are inferred to be generated by crack activity.This method can give full play to the potential of acoustic emission detection technology in laser cladding process monitoring.(3)A method of cladding state recognition and crack diagnosis based on one-dimensional convolutional neural network is proposed.A laser cladding state recognition and crack diagnosis model(SRCD)was constructed based on one-dimensional convolutional neural network.The current cladding state(normal state or abnormal process parameter)can be identified from the acoustic emission detection data,and the process causes for cracks(if there is crack activity)can be diagnosed.This method proves the application potential of acoustic emission detection technology in the cladding process,and more fully excavates the characteristic information about molten pool activity and crack activity contained in AE monitoring signals.(4)A crack prediction method based on temperature field monitoring in cladding area and LSTM is proposed.AE signals from the substrate and temperature field distribution in the cladding area were monitored by using acoustic emission detection technology and infrared thermal imager.The neural network model based on CNN and LSTM was built to judge whether the current cladding process would cause cracks due to the high(too low)of one of the above process parameters.This method takes into account the process of the temperature field in the cladding area changing with time,and combines the acoustic emission detection results in parallel to "predict" the crack activity,which provides a new idea for the laser cladding process monitoring and online feedback control method.In the work of this paper,some innovative experiments have been done for the laser cladding process monitoring,and some meaningful attempts have been made for the monitoring data processing and information mining.The method presented in this paper can accurately diagnose the cause of the process parameter setting of the equipment end in the process of metal laser cladding.It provides an idea for real-time monitoring and feedback control of laser cladding process and reducing crack defect of cladding layer.