CAJ | 학술논문

为了提高S面控制器对环境的适应能力，提出一种基于预测模型的模糊参数自寻优方法。采用非线性自回归滑动平均模型对潜水器动力学特征进行描述，并使用Elman神经网络进行模型辨识，从而建立了系统的预测模型。对于在线辨识需求，从样本容量和模型结构两个方面对预测模型进行了改进，改善了预测模型在时变环境下的预测能力。最后将建立的预测模型应用到基于模糊规则的参数自寻优S面控制器中，并进行了仿真实验。实验结果表明：该参数自寻优方法在S面控制器参数调整中取得较好的效果，改进后的S面控制器具有较快的控制响应速度。
위료제고S면공제기대배경적괄응능력，제출일충기우예측모형적모호삼수자심우방법。채용비선성자회귀활동평균모형대잠수기동역학특정진행묘술，병사용Elman신경망락진행모형변식，종이건립료계통적예측모형。대우재선변식수구，종양본용량화모형결구량개방면대예측모형진행료개진，개선료예측모형재시변배경하적예측능력。최후장건립적예측모형응용도기우모호규칙적삼수자심우S면공제기중，병진행료방진실험。실험결과표명：해삼수자심우방법재S면공제기삼수조정중취득교호적효과，개진후적S면공제기구유교쾌적공제향응속도。
In order to improve the adaptability of the S surface controller , a fuzzy parameter self-optimized method based on the prediction model was proposed .Firstly, the nonlinear auto-regressive moving average ( NARMA) mod-el was adopted to describe the dynamic characteristics of submersibles , and then the prediction model was estab-lished by identifying the NARMA model using the Elman neural network .For the requirement of the on-line identifi-cation, two improvements were made , i.e.sample size and model structure , thus the Elman network could replace its weights based samples updated with the change of environment .Finally, the prediction model was applied to the fuzzy parameter selfo-ptimized S surface controller .The simulation experiment was carried out and the expected effect was obtained with the parameter adjustments to the S surface controller using the proposed method .The im-proved S surface controller achieved faster response speed .