论文部分内容阅读
针对600MW超临界机组主汽温对象由于其具有迟延大、时变的特点而引发的调节特性差、经常出现过热器汽温超调等问题,提出1种基于实数编码小世界优化算法优化RBF神经网络的PID控制策略。该新型小世界优化算法通过应用n维球面构造小世界网络邻域和非邻域节点集,避免了现有算法由于编码、解码而造成的算法繁琐和运行时间长等问题,函数测试表明其具有较强的快速性及较高的寻优精度,非常适合于控制器的实时优化。经与传统PID和基于混沌算法优化RBF神经网络的PID控制策略的控制效果进行仿真比较,表明新方法在不同负荷下的主汽温控制均获得很好的调节品质,具有较强的稳定性、鲁棒性和自适应性。
Aiming at the problems of large steam temperature variation and large time-varying characteristics of 600 MW supercritical units, the control characteristics of the main steam temperature caused by superheater steam temperature overshoot are often poor. A real-coded small-world optimization algorithm is proposed to optimize RBF nerves Network PID control strategy. The novel small-world optimization algorithm uses n-dimensional spheres to construct small-world network neighborhoods and non-neighborhood node sets, avoids the cumbersome algorithms and long running time of existing algorithms due to encoding and decoding, and the function tests show that they have Strong fastness and high precision, ideal for real-time optimization of the controller. The simulation results of PID control strategy based on traditional PID and RBF neural network based on chaotic algorithm show that the new method achieves good regulation quality and good stability under different loads, Robustness and adaptability.