论文部分内容阅读
文章针对神经网络存在局部最优、收敛速度慢以及大样本等缺点,将改进的粒子群算法、灰色模型和神经网络模型有机结合,构建了改进粒子群优化灰色神经网络预测模型(IPSO-GMNN)。并与其他预测模型进行比较,实证结果表明:IPSO-GMNN预测模型能够克服神经网络预测模型的不足,更好地识别时间序列的非线性和突变性特征。在对我国专利授权数量的预测应用中,新模型对非线性时间数据预测表现出更好的预测精度和稳定性。
In order to overcome the shortcomings of local optimization, slow convergence speed and large sample size, this paper proposes an improved Particle Swarm Optimization (PSO) neural network prediction model (IPSO-GMNN) by combining the improved particle swarm optimization algorithm, gray model and neural network model. . Compared with other forecasting models, the empirical results show that the IPSO-GMNN forecasting model can overcome the deficiencies of the neural network forecasting model and identify the nonlinear and catastrophic features of the time series better. In predicting the number of patents granted in our country, the new model shows better prediction accuracy and stability for nonlinear time prediction.