Applying back propagation neural networks to remote monitoring of ocean-going ships

In order to satisfy the need for remote monitoring of ocean-going ships by more effective methods,back propagation(BP) neural networks were applied to improve existing monitoring systems by anticipating failure and so outputting warning signals instead of failure signals.During this process,the alarm apparatus on board could be actuated while relevant failure identification codes are sent to company management ashore.Subsequently,the best method to solve the problem could be worked out between the ship's crew and engineers ashore.To analyze the application of BP neural networks to remote surveillance,changing trends in the exhaust temperature of a six cylinder main diesel engine were simulated.Making use of the learning ability of BP neural networks,a warning model was proposed for when continually rising exhaust temperature potentially leads to engine failure.Five more non-warning models were established for other conditions.The errors between the samples and the results simulated by BP neural networks were smaller than 5%,consequently BP neural networks can judge the trend of failure and improve monitoring systems by implementing failure prediction functions.