Neural Tuner Research Articles

Rating the speed of the shearer’s electric motor drive load automatic control

Introduction. It is possible to improve productivity, effectiveness, and cost-efficiency of coal extraction due to the efficient use of physical resources, technical upgrade of mechanized longwall equipment, and introduction of advanced technologies and control methods. The existing method of shearer electric motor drive automation based on the automated load controller of Uran type has a significant drawback of low speed. In case the actuator (A) meets solid rock and the shearer’s (S) speed is not changed, it may result in heavy shock loads on A and its transmission, therefore, increased wear of the cutter or machine’s breakage, leading to production loss due to the reduced speed of travel along the face. The foregoing demands higher standards of the load controller’s speed, making the task of improving the control system’s development a relevant scientific task. Research aim is to synthesize the neural tuner for the coefficients of the proportional-integral controller (PI controller) in the control system of a shearer with increased speed as compared to the existing standard controllers. The research also aims to estimate its efficiency by the method of mathematical simulation. Methodology. Mathematical model has been developed which has made it possible to compare the performance of standard controllers with an adaptive PI controller. The structure and parameters of the neural network underlying the controller have been substantiated. The proposed controller was compared to the standard PI controller and to the MPC controller (microprocessor-based speed controller) by the method of simulation experiment. Research results. The adaptive PI controller has been synthesized based on the neural network which allows changing the coefficients of the PI controller as soon as coal strength changes. Summary. The simulation experiment has shown that the PI controller with the neural network tuner for its coefficients in the control system will make it possible to increase the load controller’s speed by 1.5 to 3 times on average as compared to the classical controller. Therefore, it is going to be possible to avoid critical overload and breakage of mechanical parts in the shearer’s transmission in case of the sudden contact of its actuator with solid inclusion.

On Development of Module for Neural Tuner to Adjust D-part of PID-controller Online

Nowadays nonlinear industrial plants are controlled by linear P, PI and PID-controllers with constant parameters. This results in both transients quality and energy efficiency decrease. An adaptive neural tuner to adjust P and PI controllers parameters has been proposed to solve these problems and tested both for heating furnaces and DC drives. But it is not able to adjust a D-part of the controller, which is widely used for the heating furnaces. So the aim of this research is to extend the functionality of the tuner with a new software module to adjust KD parameter online. In order to achieve this, a new structure of the neural network for the tuner is chosen, its rule base is enhanced with the developed rules for KD, and the sustainability criterion for the system with the tuner and the PID-controller is developed. The new version of the tuner is applied to control a heating furnace to keep the required transients quality under the condition of the furnace parameters non-stationarity. Comparing to the PID-controller, it allows to reduce transients time averagely by 11.6% and the furnace energy consumption by 8.4%.

Applying a neural tuner of the PI­controller parameters to control gas heating furnaces

In order to optimize the quality of transition processes on a heating object of control, it is proposed to apply a neural-network tuner, which changes parameters of the PI-controller in real time. The aim of present study is to determine effectiveness of application of the tuner using a model of the heating furnace, containing a gas supply control circuit and a controlling element of this circuit. Simulation was performed on the model of a gas furnace obtained through recalculation of thermal power from the model of an electric furnace. The study confirms the capability of the proposed adaptive system to effectively execute adaptation of parameters of the controller in the presence of a controlling mechanism whose dynamics may negatively affect the quality of control.The result of applying the tuner is a decrease in the time of transition process by 25.8 % and a reduction in the total controlling influence by 22.85 %. The presence of the controlling element in this case had no significant effect on the work of a neural-network tuner. The result of research makes it possible to extend the class of objects for which a neural-network tuner can be applied. Previously, the tuner demonstrated its effectiveness only for electric furnaces where influence of the controlling element is minimal. Result of the present study makes it possible to scale up the solution for gas thermal furnaces despite a markedly greater influence of the controlling element

Concerning the Use of a Neural Tuner for Adjustment of the Speed P-controller of a Rolling Mill's Main Drive

Rolling production is one of the most energy-consuming sectors of the metallurgical industry. The most powerful plants are the rolling mills producing rough rolling of the steel casts. They are based on the technology of the reverse rolling. It requires variation of the roll mill parameters. Such a variation may also be due to replacement of the worn out rolls. So, the controllers with constant parameters usage results in the deterioration of the transient quality for the rolling mill main drive. An adaptive control system can be developed to solve the problem. A brief analysis of the linear controller tuners is presented in this research and a neural tuner is proposed to solve the problem. It combines an artificial neural network and a rule base. It does not require identification of the plant model or use of the test signals. In this paper the authors consider a control system for the main DC electric drive of a two-roll reversing rolling mill. A structure for the neural network is selected and the rule base is described. It defines the moments, when a controller should be adjusted (training of the neural network) as well as an appropriate learning rate. Experiments for application of the neural tuner were conducted using the rolling mill DC electric drive model, as well as a physical model of this drive. An analysis of the experimental results shows, that the neural tuner adjusts the speed of the controller effectively. The purpose of the further research is to apply the neural tuner to compensate for the disturbances acting on the rolling mill and caused by the steel cast engagement.

On PI-controller Parameters Adjustment for Rolling Mill Drive Current Loop Using Neural Tuner

We consider a problem of an on-line tuning of linear controller parameters used for a rolling mill drive control. A neural tuner of PI-controller for current control loop is developed. It allows to adjust KP and KI according to armature winding parameters values change. Moreover, if initial values of controller parameters do not provide required transients quality, proposed system will tune them to solve this problem. A functional block diagram, a neural network structure, and a rule base are shown. Two sets of modelling experiments are conducted with a roll mill model. Obtained results show transients quality improvement in speed and current control loops. In addition to this, energy consumption of the control system with the tuner is lower by 1-2% comparing to conventional PI-controller basic system.

A New Robust Online PID with NN Gain Tuner for Pressure Process through PLC

The process variation in the pressure process tank is so rapid that it makes the PID more susceptible to the variations in the pressure and consequently presents a sluggish response. Even the neural network tuned PID controller suffers to overcome this problem. Therefore this work contributes a new robust online PID controller with back propagation (BPN) gain tuner for process control applications like pressure process. This proposed controller incorporates the versatility of the PID controller and the robustness of the neural network gain tuner. The control program is developed using the ladder logic functions of the PLC and interfaced with the pressure process tank. This online tuned PID with BPN gain tuner is validated by comparing the steady state responses of conventional PID, modified PID and neural network tuned PID controllers for the pressure process tank through PLC. The robustness of the proposed controller is validated for servo regulatory and disturbance rejection capabilities.

On PID-Controller Parameters Neural Tuner Usage for Nonlinear Plants Control

The problem of a control system efficiency increasing for nonlinear plant control is considered. It is shown that proposed neural tuner based on neural networks is an efficient approach to solve the problem of PID-controller parameters online tuning. The backpropagation method is used to train the network. This method is modified by adding a rule base containing conditions on choosing learning rate for neurons. Experiments are made for plant models with various parameters values (with a time constant about 103 seconds and 101 seconds). The required quality for each transient is determined. The system with the neural tuner allows to achieve 10% decreasing of the amount of time and 9% of energy-savings in comparison with the conventional PID-controller.

About Heating Plants Control System Developing on Basis of Neural Network Usage for PID-Regulator Parameters Optimization

An adaptive control system implementation is described. Such system is based on a neural network, used for online PID-regulator coefficients tuning. The backpropagation online training method is used. It is modified by adding a rule base. It contains conditions on choosing neural network learning rate. PID-regulator with neural tuner and conventional PID-regulator were used as regulators during the process of heating furnace control modeling. Such experiments were made for different loading furnace modes and setpoint schedules. The 11% economy of time on setpoint schedule realizing was achieved with the help of proposed neural tuner..