CAJ | 학술논문

针对海洋环境影响下武器装备作战效能的评估问题，建立了基于BP神经网络的组合预测评估模型。利用神经网络，将多个方案的预测结果作为网络输入，其真实值作为输出，以组合预测模型均方误差最小为目标，属于一种非线性组合预测模型。同时针对BP网络收敛速度缓慢，容易形成局部极小解的缺点，提出了AGABP(自适应遗传反向传播)算法对其进行改进。最后针对某作战平台作战效能的评估实例，在多种单项预测模型的基础上利用AGABP算法进行组合预测，得到了较好的实验结果。
침대해양배경영향하무기장비작전효능적평고문제，건립료기우BP신경망락적조합예측평고모형。이용신경망락，장다개방안적예측결과작위망락수입，기진실치작위수출，이조합예측모형균방오차최소위목표，속우일충비선성조합예측모형。동시침대BP망락수렴속도완만，용역형성국부겁소해적결점，제출료AGABP(자괄응유전반향전파)산법대기진행개진。최후침대모작전평태작전효능적평고실례，재다충단항예측모형적기출상이용AGABP산법진행조합예측，득도료교호적실험결과。
The combinational forecasting model based on BP neural network is established to evaluate the operational effectiveness of weapon equipment under the influence of marine environment. The model takes advantages of neural network,and takes the prediction re-sults of multiple solutions as the inputs and the true values as output of neural network aiming at getting the smallest mean squared error. It is a kind of the nonlinear combination forecasting model. To solve the problem of BP network slowly converging and easily falling into the local minimum points,an algorithm called AGABP ( Adaptive Genetic Algorithm based Back Propagation) is proposed to improve it. Finally,aiming at the operational effectiveness evaluation instance of an operation platform and based on several single-forecasting-mod-els,the AGABP model is used for combinational forecasting and gets good experimental results.