Experimental and numerical insights on the CFRP propeller: comparative interlaminar mechanical simulation, and strain-time prediction with LSTM neural network

The interlaminar mechanical characteristics and strain prediction method of carbon fibre-reinforced polymer propellers are key issues in practical engineering. Carbon Fiber Reinforced Polymer (CFRP) is a high-performance composite material composed of carbon fibers and a polymer matrix, abbreviated as CFRP in the article. In this study, the custom-designed CFRP propeller is taken as the research object. First, the detailed summary of CFRP propeller numerical modeling techniques with two lamination structures are provided, including cladding blade and symmetrical blade. Then, the mechanical tests and fluid-structure interaction numerical simulation are conducted to comparatively analyze and screening the mechanical characteristics of CFRP propeller blades with two lamination structural structures. The results reveal that the cladding blade exhibits superior stiffness and interlaminar stress distribution compared to the symmetrical blade, indicating better mechanical properties. Last, the cladding CFRP propeller is taken as the investigation subject for its better mechanical characteristic, and “an underwater test is conducted to explore the strain prediction method. Long-Short Term Memory is a special type of recurrent neural network, abbreviated as LSTM in the article. Based on the test results, LSTM neural network is employed to evaluate the strain of the cladding CFRP propeller during rotation, the coefficient of determination and error are all in acceptable range which proves that LSTM neural network is capable to fit the strain-time series data. The research on exploring interlaminar mechanical characteristic, developing the strain-time prediction with LSTM are of great significance to the application of CFRP propeller in the ocean engineering.