神经网络预测及电力主变室多参数优化控制

Neural Network Prediction and Multi-parameter Optimization Control of Power Transformer Room

针对变压器室通风散热这类多变量、非线性和时变的复杂控制系统,采用神经网络作为优化反馈控制器求解优化反馈解;利用预测控制的滚动优化具有克服室外温度干扰和不确定性影响的优势,通过滚动优化算法训练神经网络模型,同时对控制系统中负荷电流变化也采用神经网络进行预测,以实现被控对象的实时预测;利用该方法对变压器室通风散热系统进行理论分析和仿真,仿真结果表明系统具有较强的鲁棒性;最后应用于变压器室智能通风散热系统实际工程中,获得较好降温的效果。

Transformer room ventilation is a multivariable, nonlinear and time--varying control system, a neural network served as the optimal feedback controller, which was trained with optimization algorithm based on the method of the rolling optimization of predictive con- trol to compensate for disturbances and uncertain plant nonlinearities. The controller can approximate the optimal feedback solution for non- linear--time--varying systems without the complexities of computation. Additional neural networks were used to predict load current param- eters to realize the real--time predication of the dynamic behavior. An optimal control system was designed to control ventilation and heat dissipation system of the transformer chamber, which aimed at implement the theoretical analysis and simulation, Simulation results show that the system has strong robustness. Finally, applied to the ventilation and heat dissipation intelligent system of the main transformer chamber in practical engineering, which has got better effect.