Wind Speed Prediction Based On ELM Algorithm And Its Application In The Station-keeping Control For A Stratosphere Airship

The wind speed prediction in horizontal direction of the stratosphere is studied,and then the predicted wind speed is applied to the station-keeping control for a strato-sphere airship.For a stratosphere airship with station-keeping as the control target,its basic flight mode is to fly against the wind.So the wind speed prediction can effectively reduce the adverse effect of the random wind field in the stratosphere on the station-keeping control,and this means that it can improve the precision of station-keeping control of the airship,and reduce the amount of energy of the propulsion control sys-tem,and protect the motor system at the same time.Finally,the research and design of this paper mainly include the following aspects:(1)For the off-line wind speed prediction,the combination forecasting model of MS-EMD algorithm and ELM algorithm is proposed.Firstly,the basic principle of the masking signal-based empirical mode decomposition(MS-EMD)algorithm is an-alyzed,and then the MS-EMD algorithm is used to reduce the instability of the wind speed sequence.Secondly,the prediction process of the extreme learning machine(ELM)algorithm is described,and the BP and Elman neural network algorithm are al-so introduced in detail.Finally,the above three algorithms are used to forecast the wind speed respectively,and it can be concluded that the ELM algorithm has great advantage in wind speed prediction through the prediction results of them.(2)For the on-line wind speed prediction,a wind speed forecasting model on account of I-ELM algorithm is proposed.First of all,the on-line wind speed forecasting model is introduced.Secondly,the incremental extreme learning machine(I-ELM)algorithm is combined with the rolling prediction model to predict the measured wind speed,among them,the I-ELM algorithm is an improved form of the ELM algorithm.Finally,the effectiveness of the on-line rolling prediction model is illustrated by the forecasting results.(3)For the high-precision station-keeping control of the stratosphere airship,a one-dirmensional particle model of stratosphere airship in horizontal random wind field is constructed.When the airship flies back to the target origin from far away by the PID controller and reaches a steady state at the origin,the airship has a small shock amplitude in the horizontal direction.However,in order to improve the precision of station-keeping control of the airship,and reduce the energy consumption of the air-ship propulsion system at the same time,the wind speed prediction compensation is added to the PID controller after the airship reaching the steady state,in other words,a feedforward-feedback PID controller is used to control the airship.Finally,in or-der to show the effectiveness of the proposed control method,the control effect of the feedforward-feedback PID controller and the PID controller on station-keeping con-trol for the stratosphere airship are compared with each other.From the simulation results,the feedforward-feedback PID controller not only can reduce the amplitude of stratosphere airship in the horizontal direction and reduce the energy consumption of airship propulsion control system,but also can play a role in protecting the motor of stratosphere airship.The simulations also show that the wind speed prediction is quite necessary to the station-keeping control for a stratosphere airship.This paper puts forward the following innovations based on previous research.Firstly,the MS-EMD algorithm and ELM algorithm are combined and applied to the off-line prediction of wind speed series.Secondly,this paper creatively establishes an online wind speed forecasting model based on the I-ELM algorithm.Lastly,the on-line wind speed prediction is applied to the high-precision station-keeping control of stratospheric airship,and this means that the feedforward-feedback PID controller is used to control the airship.