变异蚁群神经网络及其对DTC转速的辨识

将变异机制引入基本蚁群算法中，然后利用这种变异蚁群算法去优化神经网络的权值，有效地解决了神经网络容易陷入极小点的缺点，同时又远比只采用单一的基本蚁群算法提高了收敛速度，从而得到一种时间效率和求解效率都比较好的启发式方法，即变异蚁群神经网络．通过对直接转矩控制中电机转速进行辨识的仿真实验，结果表明：这种变异蚁群神经网络兼具了神经网络和蚁群算法两方面的优点，不仅具有广泛的映射能力，还明显提高了运算效率，用变异蚁群神经网络构造的转速辨识器能够准确地跟踪电机转速的变化，使系统具有良好的动态性能．

Mutated ant colony algorithm-based neural network and speed identification in DTC

Mutated evolutionary algorithm was added into ant colony optimization to form a new algorithm,mutated ant colony optimization.The weights of neural networks were optimized by using this new algorithm.Formed is a mutated ant colony neural network,namely a new type of heuristic algorithm,characterized by high speeds of the convergence and solving solutions.The speed identification of motors in direct toque control(DTC) was simulated.The results show that the mutated ant colony neural network,having the advantages of both neural network and ant colony optimization,not only is able to map widespread,but also to enhance the operation effectiveness.The speed identification sensor consisting of mutated ant colony-based neural network can trace the variation of motor speed accurately,thus the motor system showing its favorable dynamic properties.