Laser cladding technology has become a central issue in industrial research and application recently. Due to the complexity of the processing and the difficulty of quantitative analysis, it is particularly important to effectively and reliably detect the forming quality of the cladding layer. Therefore, the laser cladding processing condition of metallic panels is monitored and identified based on acoustic emission technology in this article. Consequently, laser cladding acoustic emission signal of 316L stainless steel plate is first collected, and a multi-domain acoustic emission signal feature extraction method based on time–frequency domain and waveform parameters is constructed by integrating time–frequency, empirical, and inter-relationship diagraph analysis. Moreover, a feature optimization approach based on t-distributed stochastic neighbor embedding algorithm is proposed, which combines with correlation analysis to realize the de-redundancy and optimization of acoustic emission signal features. In addition, on the basis of optimizing the characteristics of acoustic emission signal, laser cladding acoustic emission signals under three technological parameters are collected. Second, a laser cladding condition identification method, which is a least square support vector machine algorithm based on niche particle swarm optimization, is proposed. The optimal parameter combination of niche particle swarm optimization algorithm is mainly selected to improve the accuracy of identification and classification. The results demonstrate the proposed t-distributed stochastic neighbor embedding feature optimization method can effectively extract the sensitive information related to the processing condition in the feature space, and the optimization of least square support vector machine parameters through niche particle swarm optimization can significantly improve the identification rate of classification. Specifically, the feasibility of the proposed t-distributed stochastic neighbor embedding–niche particle swarm optimization–least square support vector machine model to analyze laser cladding acoustic emission signal characteristics is verified, and the effectiveness of acoustic emission–based structural condition monitoring and identification of laser cladding plate-like structures is also demonstrated.
The Acoustic Emission Signal Recognition based on Wavelet Transform and RBF Neural Network
