Global Optimization Method for Planetary Hybrid Power System with Constrained Terminal State

Abstract

The invention provides a terminal state restricted planet type hybrid power system global optimization method, and belongs to the technical field of new energy vehicles. The method comprises the stepsthat before DP backward motion optimization, system boundary calculation is carried out, an dboundary restriction of a state variable of every moment is obtained, so that the boundary restriction isconsidered in the backward iterative optimization process, and electric quantity balance of the system is achieved. By means of a solution of the boundary restriction, no penalty function is needed, and a large amount of debugging work conducted in order to achieve the electric quantity balance is avoided; meanwhile, robustness of an algorithm is no longer affected by model parameters and operation conditions, the amount of calculation and the time cost are lowered, and the efficiency of optimizing the algorithm is significantly improved.

Description

technical field [0001] The invention provides a global optimization method for a planetary hybrid power system with constrained end states, belonging to the technical field of new energy vehicles. Background technique [0002] Hybrid power has the requirement of power balance, and the current energy management strategy optimization based on DP global optimization usually uses a penalty function to meet the system's terminal state constraints. However, the penalty function in most studies requires researchers to conduct multiple debugging based on experience, which will bring a larger amount of calculation and increase the calculation time by several times, which will not be conducive to the automatic implementation of the global optimization algorithm. In addition, since the model parameters used for optimization will change with the passage of time or the vehicle state, and the target operating conditions of the global optimization will also change with the historical opera...

AI Technical Summary

Problems solved by technology

However, the penalty function in most studies requires researchers to conduct multiple debugging based on experience, which will bring a larger amount of calculation and increase the calculation time by several times, which will not be conducive to the automatic implementation of the global optimization algorithm

In addition, since the model parameters used for optimization will change with the passage of time or the vehicle state, and the target operating conditions of the global optimization will also change with the historical operating data, these factors will lead researchers to produce The calibrated penalty function is not robust, further reducing the application value of the penalty function method

For example, the patent for invention published on June 8, 2016: application publication number: CN 105644548 A, energy control method and device for hybrid electric vehicles, the method is based on stochastic model predictive control and neuron dynamic programming algorithm to realize the energy control of hybrid electric vehicles Energy management control, neuron dynamic programming algorithm for global optimization solution, by setting the reward and punishment function to keep the system power balance, need to constantly debug the factors of the reward and punishment function, there is a huge debugging workload, long calculation time, and low efficiency of global optimization solution And other issues

Images