基于BP神经网络的海洋压力模拟设备PID控制

PID Control of Marine Pressure Simulator Based on BP Neural Network

摘要: 深海压力的不断变化会给深海勘测装置带来巨大的不确定因素，因此海洋压力模拟对海洋资源开发有着重要的意义。文章介绍了比单电机压力模拟有着更好的精度一种新型的双电机活塞式海洋压力模拟设备，在说明结构和工作原理后，建立了进给系统的数学模型。压力模拟设备具有非线性、强时变性、强干扰性和强耦合性的特点。由于传统PID无法满足控制要求，因此，文中提出采用基于误差反向传播(error Back Propagation)神经网络的PID控制算法用于控制该海洋压力模拟设备。在Matlab中进行仿真实验验证可行性后，在搭建的实验平台上进行实验验证。实验结果表明BP神经网络PID控制能在一定程度上克服传统PID控制的缺陷。在1 MPa阶跃压力实验中，减少了1.5 s响应时间、8.7%相对超调量，增强了抗干扰性。有效提高压力模拟设备的性能。

Abstract: Marine pressure simulation is of great significance to Marine resource development. Ocean pres-sure simulators help in the development of deep water. This paper introduces a new type of Marine pressure simulator with double motors and pistons. After introducing the structure and working principle, the mathematical model of the system is established. Pressure simulator has the charac-teristics of nonlinearity, strong time-variability, strong interference and strong coupling. Tradi-tional PID cannot meet the control requirements. Therefore, BP neural network PID control pres-sure simulation equipment is proposed. After the feasibility is verified by simulation experiment in matlab, the experimental platform is built for experimental verification. The experimental results show that BP neural network PID control can overcome the defect of traditional PID control to some extent and improve the performance of pressure simulation equipment. In the step pressure experiment of 1 MPa, the response time of 1.5 s and the relative overshoot of 8.7% were reduced and the anti-interference property was enhanced.