Neural Network PID Research Articles

Self-Tuning Neural Network PID With Dynamic Response Control

PID controllers are widely used and adaptable to various types of systems. However, for the response to be adequate under different conditions, the PID gains must be adjusted. The tuning is made according to the difference between the reference value and the real value (error). This work presents a self-adjusting PID controller based on a backpropagation artificial neural network. The network calculates the appropriate gains according to the desired output, that is, the dynamic response desired which is composed of the transient part and the stationary part of the step response of a system. The contribution of the work is that in addition to using the error for network training, the maximum desired values of overshoots, settling times, and stationary errors were used as input data for the network. An offline training database was created using genetic algorithms to obtain the dynamic response data associated with PID gains. The genetic algorithm allows getting data in different operating ranges and allows using only stable gains combinations. The database was used for training. Subsequently, the neural network estimates an appropriate gain combination, adapting to the error and the desired response. The method performance is evaluated by controlling the speed of a direct current motor. The results indicate an average error of 4% for the database between the requested and system response. On the other hand, the gains estimated by the network in the test dataset (1544 combinations) did not cause instability and complying with the expected dynamic response in 86% of the dataset.

Disturbance learning controller design for unmanned surface vehicle using LSTM technique of recurrent neural network

In this study, to effectively control small unmanned surface vehicles (USVs) for marine research, characteristics of ocean current were learned using the long short-term memory (LSTM) model algorithm of a recurrent neural network (RNN), and ocean currents were predicted. Using the results, a study on the control of USVs was conducted. A control system model of a small USV equipped with two rear thrusters and a front thruster arranged horizontally was designed. The system was also designed to determine the output of the controller by predicting the speed of the following currents and utilizing this data as a system disturbance by learning data from ocean currents using the LSTM algorithm of a RNN. To measure ocean currents on the sea when a small USV moves, the speed and direction of the ship’s movement were measured using speed, azimuth, and location (latitude and longitude) data from GPS. In addition, the movement speed of the fluid with flow velocity is measured using the installed flow velocity measurement sensor. Additionally, a control system was designed to control the movement of the USV using an artificial neural network-PID (ANN-PID) controller [12]. The ANN-PID controller can manage disturbances by adjusting the control gain. Based on these studies, the control results were analyzed, and the control algorithm was verified through a simulation of the applied control system [8, 9].

Development of a Robotic Catheter Manipulation System Based on BP Neural Network PID Controller.

In the process of artificial interventional therapy, the operation of artificial catheter is not accurate, which will bring strong radiation damage to surgeons. The purpose of this study is to develop a catheter operating system of surgical robot to assist doctors in remote operation and avoid the influence of radiation. BP neural network plays an important role in the flexibility and rapidity of control. According to the actual output of the system, the control parameters of the controller are constantly adjusted to achieve better output effect. This paper introduces the practical application of BP neural network PID controller in the remote operation of the system and compares with the traditional PID controller. The results show that the new control algorithm is feasible and effective. The results show that the synchronization performance of BP neural network PID controller is better than that of traditional PID controller.

PIDNN control for Vernier-gimballing magnetically suspended flywheel under nonlinear change of stiffness and disturbance

Vernier-gimballing magnetically suspended flywheel is often used for attitude control and interference suppression of spacecrafts. Due to the special structure of the conical magnetic bearing, the radial component generated by the axial magnetic force and the change of the magnetic air gap will cause the nonlinearity of stiffness and disturbance. That will lead to not only poor stability of the suspension control system but also unsatisfactory tracking accuracy of the rotor position. To solve the nonlinear problem of the system, this article proposes a proportional–integral–derivative neural network control scheme. First, the rotor model considering the nonlinear variation of disturbance and stiffness parameters is established. Then, the weight of neural network is adjusted by the gradient descent method online to ensure the accurate output of magnetic force. Finally, the convergence analysis is carried out based on the Lyapunov stability theory. Compared with the general proportional–integral–derivative control and the radial basis function neural network control, the simulation results demonstrate that the proposed method has the highest tracking accuracy and excellent performance in improving stability. The experimental results prove the correctness of the theoretical analysis and the validity of the proposed method.

Design and application of supervisory control based on neural network PID controllers for pressurizer system

The pressurizer system (PRZ) plays a vital role in the operation of pressurized water reactors (PWRs). The PRZ is inherently a non-linear, time-variant system, and hence the PRZ modeling with transfer function or linear time-invariant state-space forms is not accurate enough. In this paper, the thermal-hydraulic behavior of the PRZ is modeled by RELAP5 thermal-hydraulic code that is one of the best modeling codes offered for the PRZ system. Although the RELAP5 best-estimate code is appropriate for PRZ modeling, the RELAP5 is not capable to implement the advanced controllers. As first-time, to eliminate this issue, the RELAP5 code and MATLAB software are coupled to use the capabilities of MATLAB (the ability to implement advanced controllers) and RELAP5 (the best model for the PRZ system) simultaneously. Accordingly, this coupling provides a new platform for designing and implementing various intelligent and advanced controllers for PRZ pressure and level in RELAP5 code. Likewise, after providing the platform for implementing intelligent controllers, a novel supervisory control based on neural network PID (NN-PID) controllers is designed for the PRZ pressure and level control. Also, the PRZ system is modeled as a MIMO system with consideration of the PRZ pressure and level interactions. Furthermore, as a case study, the advanced designed control system is applied to adjust the pressure and level of PRZ in a VVER-1000 type nuclear power plant. The study results show that this newly designed control system is able to control the PRZ pressure and level efficiently and effectively in several conditions. Therefore, the proposed methodology as a novel design of PRZ control can be applied and developed in new generations of nuclear power plants and pressurized test loops.

Data-Driven Tracking Control Based on LM and PID Neural Network With Relay Feedback for Discrete Nonlinear Systems

In this article, a hybrid algorithm of Levenberg-Marquardt (LM) and proportional-integral-derivative neural network (PIDNN) with the relay feedback (LM-PIDNN-RF) is proposed to solve control problems for unknown discrete nonlinear systems. First, the PIDNN is initialized by the relay feedback to solve the problem of assigning initial weights; meanwhile, the LM neural network is regarded as an identifier to fit system input/output quickly and accurately. Second, the partial derivative of the system output to system input is transferred to the PIDNN, which ensures that the weights of the PIDNN can be updated correctly in time. The hybrid algorithm can update the weights of the neural network controller correctly against the errors caused by system instantaneous disturbance, and the controller has only one parameter to be tuned manually. Moreover, the stability of the closed-loop system is proven by using the Lyapunov stability theory. The proposed hybrid algorithm can significantly improve tracking performance in comparison with PIDNN and RF-PID. The results of three simulation examples and a physical experiment are presented to show superior tracking performance of the proposed algorithm.

An Improved Method for PID Control of Time-Delay Systems

The industrial systems has became more complex, because of the rapid development of industrial production. Large time-delay, time variation, and highly nonlinear have placed higher requirements on industrial control systems. Smith predicting controller is a controller specifically used for time-delay systems. It estimates the dynamic characteristics of the system’s basic disturbances in advance, and designs an estimator for compensation control, which can effectively improve the system response. But Smith predicting controller depends too much on the mathematical model of the controlled object. This paper proposed a control system based on Smith-PIDNN, which combines Smith controller with neural network PID (PIDNN), uses the neural network’s ability to simulate non-linear systems and self-learning ability, and adjusts control parameters online. And it was simulated by MATLAB software, the effectiveness of this combination method was proved.

Control of Hydraulic Power System by Mixed Neural Network PID in Unmanned Walking Platform

To speedily regulate and precisely control a hydraulic power system in a unmanned walking platform (UWP), based on the brief analysis of digital PID and its shortcomings, dual control parameters in a hydraulic power system are given for the precision requirement, and a control strategy for dual relative control parameters in the dual loop PID is put forward, a load and throttle rotation-speed response model for variable pump and gasoline engine is provided according to a physical process, a simplified neural network structure PID is introduced, and formed mixed neural network PID(MNN PID)to control rotation speed of engine and pressure of variable pump, calculation using the back propagation(BP) algorithm and a self-adapted learning step is made, including a mathematic principle and a calculation flow scheme, the BP algorithm of neural network PID is trained and the control effect of system is simulated in Matlab environment, real control effects of engine rotation speed and variable pump pressure are verified in the experimental bench. Results show that algorithm effect of MNN PID is stable and MNN PID can meet the adjusting requirement of control parameters.

Secondary and Tertiary Voltage Control of a Multi-Region Power System

This paper presents techniques for the application of tertiary and secondary voltage control through the use of intelligent proportional integral derivative (PID) controllers and the wide area measurement system (WAMS) in the IEEE 39 bus system (New England system). The paper includes power system partitioning, pilot bus selection, phasor measurement unit (PMU) placement, and optimal secondary voltage control parameter calculations to enable the application of the proposed voltage control. The power system simulation and analyses were performed using the DIgSILENT and MATLAB software applications. The optimal PMU placement was performed in order to apply secondary voltage control. The tertiary voltage control was performed through an optimal power flow optimization process in order to minimize the active power losses. Two different methods were used to design the PID secondary voltage control, namely, genetic algorithm (GA) and neural network based on genetic algorithm (NNGA). A comparison of system performances using these two methods under different operating conditions is presented. The results show that NNGA secondary PID controllers are more robust than GA ones. The paper also presents a comparison between system performance with and without secondary voltage control, in terms of voltage deviation index and total active power losses. The graph theory is used in system partitioning, and sensitivity analysis is used in pilot bus selection, the results of which proved their effectiveness.

Research on Semiactive Control of Civil Engineering Structure Based on Neural Network

In order to improve the strength of civil engineering structure, a semiactive control model of civil engineering structure based on neural network is proposed, and the control constraint parameter model of semiactive regulation of civil engineering structure is constructed. Combined with the controlled object model, the semiactive control model of civil engineering structure is designed, the mechanical analysis model of civil engineering structure is established, and the semiactive regulation of civil engineering structure is carried out by the small disturbance suppression method. The semiactive adjustment of civil engineering structure is carried out by using the structural strength fusion tracking method. Taking the internal strength and shock yield response of civil engineering structure as constraint parameters, the semiactive control of civil engineering structure is carried out and PID neural network is used to optimize the control system. The simulation results show that the semiactive control of civil engineering structure with this method has good stability, and the strength and yield response strength of civil engineering structure are improved, and it has good control efficiency.

A Back Propagation Neural Network with Double Learning Rate for PID Controller in Phase-Shifted Full-Bridge Soft-switching Power Supply

This paper mainly focuses on the control strategy for phase-shifting full-bridge soft switching electrolytic silver power supply based on Zero Voltage Switching (ZVS) soft switching technology. Taking into consideration the low performance of traditional PID control for phase-shifting full-bridge soft-switching, this paper introduce a PID improved by Back Propagation (BP) neural network with one single learning rate which is used to calculate weights from the input layer to the hidden layer and weights from the hidden layer to the output layer. After testing, it is found that setting independent learning rate for calculation of weights from the input layer to the hidden layer and weights from the hidden layer to the output layer which will not have an adverse effect on the design of the controller. Instead, the learning rate can be set according to the respective characteristics of the weights between the two layers, which is called double learning rate BP neural network PID. The simulation results indicate that compared with the single learning rate BP neural network PID control, the double learning rate BP neural network control has higher response speed, less over-shoot, short time to enter the steady state and strong immunity.

Multi-packet transmission aero-engine DCS neural network sliding mode control based on multi-kernel LS-SVM packet dropout online compensation.

In view of the strong nonlinear characteristics of the multi-packet transmission Aero-engine DCS with induced delay and random packet dropout, a neural network PID approach law sliding-mode controller using sliding window strategy and multi-kernel LS-SVM packet dropout online compensation is proposed. Firstly, the time-delay term in the system model is transformed equivalently, to establish the discrete system model of multi-packet transmission without time-delay; furthermore, the construction of multi-kernel function is transformed into kernel function coefficient optimization, and the optimization problem can be solved by the chaos adaptive artificial fish swarm algorithm, then the online predictive compensation will be made for data packet dropout of multi-packet transmission through the sliding window multi-kernel LS-SVM. After that, a sliding-mode controller design method of proportional integral differential approach law based on neural network is proposed. And online adjustment of PID approach law parameters can be achieved by nonlinear mapping of neural network. Finally, Truetime is used to simulate the method. The results shows that when the packet dropout rate is 30% and 60%, the average error of packet dropout prediction of multi-kernel LS-SVM reduces 29.21% and 44.66% compared with that of combined kernel LS-SVM, and the chattering amplitude of the proposed neural network PID approach law sliding-mode controller is decreased compared with other five approach law methods respectively. This controller can ensure a fast response speed, which shows that this method can achieve a better tracking control of the aeroengine network control system.

Research on remote intelligent control technology of throttling and back pressure in managed pressure drilling

In order to reduce the non production time of drilling, improve the efficiency and safety of drilling, improve the economic effect of managed pressure drilling (MPD), and realize the intelligent control construction of digital oilfield. Based on the pressure control in MPD, this paper analyzes the pressure control drilling system, takes the wellhead back pressure as the controlled parameter, calculates the mathematical model of the throttle valve according to the characteristics of the throttle valve, the basic parameters and boundary conditions of pressure control drilling, and puts forward an improved particle swarm Optimization PID neural network (IPSO-PIDNN) model. By means of remote communication, VR technology can realize remote control of field control equipment. The real-time control results of IPSO-PIDNN are compared with those of traditional PID neural network (PIDNN) and traditional Particle Swarm Optimization PID neural network (PSO-PIDNN). The results show that IPSO-PIDNN model has good self-learning characteristics, high optimization quality, high control accuracy, no overshoot, fast response and short regulation time. Thus, the advanced automatic control of bottom hole pressure in the process of MPD is realized, which provides technical guarantee for the well control safety of MPD.

Research on speed regulation system of water turbine generator unit based on PID neural network

For the water turbine generator unit speed control system control error is large, long adjustment time, low control accuracy, poor stability and other problems, the PID neural network speed control strategy is proposed. First of all, the PID neural network (PIDNN) is used to design the speed control system of the turbine generator unit, and a complete model of the turbine generator unit speed control system is established. Secondly, using MATLAB for simulation, the step response of the water turbine generator unit speed control system is obtained. Finally, compared with the traditional PID controller, the PID neural network controller has obvious advantages in controlling error, adjusting time, over-tuning inhibition, etc., and has good stability.

Design of Estimation Algorithm of Island Intelligent Tourist Volume Based on Data Mining

Zhang, J. and Wang, B., 2020. Design of estimation algorithm of island intelligent tourist volume based on data mining. In: Malvarez, G. and Navas, F. (eds.), Global Coastal Issues of 2020. Journal of Coastal Research, Special Issue No. 95, pp. 985-990. Coconut Creek (Florida), ISSN 0749-0208.The traditional intelligent neural network PID control method is not conducive to the estimation of the quantity control of the island intelligent tourism. In this paper, a particle swarm optimization (PSO) algorithm is proposed to estimate the tourist volume of island intelligent tourism. The fuzzy PID control method is used to establish the intelligent output scheduling control model. In this paper, the adaptive scheduling weighting coefficient of island tourism is introduced, and the radial optimization of iterative steps of particle swarm optimization is used to estimate the island intelligence tourism. On this basis, a mathematical model of intelligent tourism evaluation based on data mining theory is established. The simulation results show that the algorithm has high precision and good convergence, and improves the robustness of island tourism scheduling control.

Intelligent Thermal Control Algorithm Based on Deep Deterministic Policy Gradient for Spacecraft

In practical applications, it is difficult for the control parameters of the proportional integral derivative (PID) thermal controller to be self-tuning online. As the control object or environment changes, the control parameters are required to change accordingly. An intelligent thermal controller based on the deep deterministic policy gradient, called DRLTC, is proposed. Two types of reinforcement learning agents were designed in DRLTC, which can automatically adjust the control parameters of the thermal controllers and self-optimize online after training. Both theoretical and experimental results revealed that, when the control object was the main mirror support, the DRLTC achieved a control precision of 0.01°C. Additionally, the steady-state error was reduced by 40.2, 62.5, and 33.3% in the simulation and by 5.6, 80.6, and 85.7% in the experiment, compared with the reinforcement learning PID, neural network PID, and adaptive PID control based on fuzzy control, respectively. When the control object was changed to the main mirror installation, the DRLTC achieved a control precision of 0.02°C, and the steady-state error was reduced by 87.5, 91.7, and 90.9% in the simulation and by 80.2, 90.6, and 85.7% in the experiment, compared with the above-mentioned thermal control strategies, respectively. Therefore, the DRLTC has better universality, has stronger robustness, and saves more energy.

Design of Constant-Speed Control Method for Water Medium Hydraulic Retarders Based on Neural Network PID

The water medium hydraulic retarder is the latest type of auxiliary braking device and has the characteristics of high power density, large braking torque, and compact structure. During traveling, this device can convert the kinetic energy of a vehicle to the heat energy of the cooling liquid and replace the service brake under non-emergency braking conditions. With regard to the constant-speed function of the water medium hydraulic retarder, this study designs a controller based on the neural network proportional–integral–derivative (PID) algorithm to achieve the steady traveling of the vehicle at constant velocity during a downhill course by controlling the filling ratio of the water medium hydraulic retarder. To validate the algorithm’s effectiveness, the dynamic model of the heavy-duty vehicle in the downhill process and the physical model of the water medium hydraulic retarder are developed. Three operating conditions, including a fixed slope, step-changing slope, and continuous changing slope, are set, and a simulation test is carried out in the MATLAB/Simulink environment. The neural network PID algorithm has better adaptability in controlling than the traditional PID algorithm. Thus, it controls the water medium hydraulic retarder such that the braking requirements of heavy-duty vehicles under a changing slope working condition are satisfied, and it performs constant-speed control when the vehicle travels downhill. Therefore, the proposed control method can significantly improve the safety of road traffic.

EKF-Neural Network Observer Based Type-2 Fuzzy Control of Autonomous Vehicles

This paper proposes a novel robust path-following strategy for autonomous road vehicles based on type-2 fuzzy PID neural network (PIDT2FNN) method coupled to an Extended Kalman Filter-based Fuzzy Neural Network (EKFNN) observer. Uncertain Gaussian membership functions (MFs) are employed to self-adjust the universe of discourse for MFs using the adaptation mechanism derived from Lyapunov stability theory and Barbalat's lemma. External disturbances are significant in autonomous vehicles by changing the driving condition. Furthermore, parametric uncertainties related to the physical limits of tires and the change of the vehicle mass may significantly affect the desired performance of autonomous vehicles. The robustness of the proposed controller against the parametric uncertainties and external disturbances is compared with one active disturbance rejection control (ADRC) algorithm, and a linear-quadratic tracking (LQT) method. The obtained results in terms of the maximum error and root mean square error (RMSE), demonstrate the effectiveness of the proposed control algorithm to reach the minimized path-tracking error.

Research and Practice of Neural Network PID Process Control Algorithm Based on Offshore Oil and Gas Platform

Since the exploitation of offshore oil and gas, after decades of development, it has gradually played an important role in world economic development. The central control system of offshore platforms is the core equipment of offshore oil and gas fields. Offshore platform industrial process control requires PID control most of the time. However, in practical applications, the tuning of PID parameters has not been well resolved. This paper combines neural network and PID control to build a neural network PID controller. At the same time, using OPC communication technology, the intelligent control PID algorithm is perfectly combined with the central control system, and the optimized model and parameters are applied to the actual control object of the oil level control of the offshore oil production platform, and the control effect is obtained satisfactorily.

Fault Diagnosis of HVAC System Considering LMBP Neural Network Method

Given the complexity of the mathematical model for the HVAC system and the constant changes of multiple external disturbances and working conditions, the application of intelligent control in the HVAC control system to improve the control performance of the HVAC control system has become a research hotspot. The Levenberg-Marquart algorithm (LM algorithm) is studied to address the defects of slow convergence speed and prone to falling into local minimum in the learning process of the BP neural network. To solve the two issues (the choice of the learning rate, and the solution of the inverse matrix) that significantly affect the training time and convergence accuracy, we improved and optimized the LM algorithm using the LU decomposition method and implemented the algorithm through the MATLAB programming language. The LMBP neural network method obtained is applied to the control loop of the HVAC chilled water cycle. The simulation comparison study is performed on its control effect and that of the PID control algorithm and BP neural network PID control algorithm. The research results show that the LMBP neural network method has significantly improved the performance in aspects such as reducing overshoot, accelerating convergence speed and reducing steady-state error, etc.