Submarine cable insulation aging condition monitoring and dynamic threshold alarm algorithm based on Bayesian optimization

The submarine cable insulation aging alarm threshold is difficult to adapt to complex state changes. Directly collected cable status data is used, resulting in poor dynamic adjustment stability. Therefore, a submarine cable insulation aging status monitoring and dynamic threshold alarm algorithm based on Bayesian optimization is proposed. Using the Bayesian optimization algorithm, a Gaussian process regression model is constructed. Through sampling, evaluation, and updating, the minimum insulation impedance modulus is calculated and the predicted value is updated through comprehensive evaluation. The upper and lower dynamic thresholds are determined using Bayesian optimization, and the over-limit ratio is calculated. Experimental results show that the average insulation resistance of a well-insulated submarine cable is approximately 1500 MΩ, which drops to approximately 800 MΩ for moderately aged cables and below 300 MΩ for severely aged cables. Furthermore, in a healthy state, the electric field strength distribution is relatively uniform, while in advanced aging, the distribution becomes disordered and concentrated areas increase. The designed algorithm threshold adjustment range within 24 hours is relatively small, ranging from 0.08 to 0.13.