Research On Energy Management Strategy Of Hybrid Ship Based On SVM And MPC

The marine hybrid power system can effectively reduce fuel consumption and pollutant emission by combining different types of power sources,take advantage of complementary advantages.However,because of its more complex power structure composition and operation mode,the energy management of hybrid ships is also more complicated.At present,the energy management strategies for parallel hybrid ships are mainly divided into rule-based control strategy and optimization-based control strategy.The rule-based control logic is relatively simple and less calculation time,but the control effect is limited.Global optimization control strategies,need to grasp all working condition information in advance for off-line optimization,which can not be applied to practical engineering applications.The real-time optimization strategy,has limited optimization effect and poor robustness.Energy management strategy needs further research and development.In view of the above problems,in order to reduce the fuel consumption of power system,this paper combines on-line condition recognition with real-time optimization strategy,and proposes an energy management strategy of hybrid ship based on support vector machine and model predictive control.The research work of this paper is as follows:(1)The model of ship’s parallel hybrid power system is established.Taking a hybrid recreational fishing boat as the research object,the mathematical models of important components of power system such as diesel engine,motor,battery and super capacitor are established by combining experimental data and theoretical formula,and the simulation model of ship’s parallel hybrid power system is established.It is used to carry out simulation analysis and verify the performance of the proposed energy management strategy.(2)The on-line recognition method of ship working conditions is studied.On the basis of analyzing the data of typical historical working conditions of ships,an on-line recognition method of ship working conditions based on support vector machine is studied and proposed.Taking the actual operation data of the "Meiwei Kaiyue" cruise ship as the sample data and integrating the actual operation status of the ship,an online working condition recognition model based on support vector machine is constructed and trained.Through the comparison between the actual operation condition of the ship and the recognition results of the on-line working condition recognition model,the accuracy of the recognition results of the on-line working condition recognition model is verified.(3)A hybrid ship energy management strategy based on condition recognition and prediction is constructed.In order to improve the control performance of real-time energy management strategy,a method combining on-line recognition of working conditions and predictive control is proposed,and an energy management strategy of hybrid electric ship based on support vector machine and model predictive control is constructed.Through the simulation analysis of different strategies,the effectiveness and fuel economy of the proposed energy management strategy are verified.Compared with the MPC-based control strategy,the proposed energy management strategy can save 4.55% of fuel consumption under the premise of meeting the demand power,which is very close to the fuel consumption of DP-based strategy.