论文部分内容阅读
化工过程中的温度对象往往具有惯性大、滞后时间长的特征,严重影响常规PID控制算法的性能。神经网络算法能够拟合操作人员的操作过程,往往能够取得常规控制难以达到的效果。本文针对反应过程,根据现场操作经验离线训练1个3层神经网络模型,在中控ECS-700集散控制系统(DCS)的VF软件平台上实现,并在维生素生产的一个反应中投运,结果验证了神经网络算法在温度控制上的实际有效性。
The temperature objects in the chemical process often have the characteristics of large inertia and lag time, which seriously affect the performance of the conventional PID control algorithm. Neural network algorithm can fit the operation of the operator, often can achieve the effect of conventional control is difficult to achieve. In this paper, a three-layer neural network model is trained offline according to the field operation experience and implemented on the VF software platform of DCS-controlled ECS-700, and put into operation in a reaction of vitamin production. As a result, Verify the practical effectiveness of neural network algorithm in temperature control.