中文摘要：

针对动态模糊神经网络（DFNN）在进行预测应用时容易陷入“局部极值”的缺陷，提出一种改进方案。综合人工免疫算法和遗传算法的优点，用免疫选择优化遗传算法的进化策略，提出一种新的免疫遗传算法。将免疫遗传算法对模糊神经网络的学习算法进行改进，增强其学习能力和算法的稳定性。结合复杂性强的短期电价预测问题，采集美国PJM电力市场的实际数据作为样本数据，对免疫遗传算法改进DFNN进行实证研究。研究结果表明：与DFNN的预测结果相比，改进后的方法在同样的运算条件下，预测精度提高4．5％，而运算时间仅增加6．4s，说明基于免疫遗传算法对DFNN模型的改进效果较好。

英文摘要：

An improvement for dynamic fuzzy neural network （DFNN） was presented to avoid its running into the local extreme. By integrating the excellence of the artificial immune algorithm and the genetic algorithm, an new artificial immune-genetic hybrid algorithm was proposed, and its evolution game of the genetic algorithm was optimized by the immune algorithm. The artificial immune-genetic hybrid algorithm was used to improve the learning tactic and the steadiness of the DFNN. Based on the actual electricity price data of the American PJM power market, the proposed algorithm and the DFNN were applied in the complex problem of short-term electricity price forecasting. The comparison results indicate that the precision of the proposed algorithm is enhanced by 4.5% and the time is only increased by 6.4 s, so the improved effect of the proposed method is obvious.