Rotation Speed Control of the Rotary Valve in MWD Tools Based on Speed Feedforward Compensation

Abstract

The rotary valve speed control, seriously affected by the nonlinear characteristic of rotary valve load torque, affects the generation of drilling fluid pressure phase shift keying (PSK) signal and its quality. The calculation model feedforward of the load torque acts on the speed control system and enables the motor voltage to change according to the law of calculation model, and the linearization correction of the speed system is performed. Additionally, the flow measurement is introduced into the calculation model of the load torque to track the load torque change with the flow, suppressing the influence of large changes in flow rate on open-loop control of rotary valve speed. The closed-loop proportional-integral-derivative (PID) control is formed by negative feedback of speed, and the PID parameter value rule is established based on the rapid decay control of transient component in the rotary valve speed step response, which speeds up the tracking of the rotary valve speed following the control voltage pulse and reduces the distortion of the drilling fluid pressure PSK signal, increasing the frequency of the drilling fluid pressure carrier and improving the transmission rate of downhole information. Simulink simulation indicates that the closed-loop PID control of rotary valve speed can track the change of control voltage pulse quickly and strongly suppress the interference influences from flow measurement error and load torque calculation model deviation.

Neural Network-Based Adaptive PID Controller Design for Over-Frequency Control in Microgrid Using Honey Badger Algorithm

“…The schematic diagram of the proposed MLP neural network architecture is shown in Figure 11. Finally, Equation (21) shows the output of the neural network in terms of the weights of the output layer and the hidden layer. The input to the Artificial Neural Network (ANN) consists of the interrupted power P 12 , which represents the loss of load.…”

“…w j,l ϕ l +w j,0 +W i,0 (21) where W i,j are the weights of the output layer, w j,l is the weights of the input layer, f j is the activation function of neuron j in the hidden layer, nh is the number of neurons in the hidden layer, and nϕ is the number of neural network inputs.…”

“…The load frequency control system consists of three parts regulating the speed of the steam turbine and the inertia of the system, and to increase the efficiency of the thermal cycle and the useful life of the turbine blades, multi-story turbines are used. In this control system, by opening or closing the steam valve, the steam flow on the turbine is regulated and finally controls the output active power of the generator [19], [20], [21], [22], [23].…”