Improved genetic algorithm for direct current motor high speed controller implemented on field programmable gate array

Direct Current (DC) motors are widely used in robotic systems in case of their simplicity, more accessible to be linear control and good speed controllability. However, these systems still poor performance if not comprise with a controller. Proportional Integral (PI) controller one of the most significant controllers that use to improve the speed performance of DC motors. There are many researches have been done to optimize PI controller based evolutionary algorithm, such as Genetic Algorithm (GA). However, it has several drawbacks come from randomly searching constraints that cause lousy optimization. There are very little studies to analyse the influence of modifying initialization constraints on GA based PI controller problems objectively. On the one side, the estimation Transfer Function (TF) of these motors is considered a significant problem in most previous studies which causes bad controller design, if the low estimation accuracy. Additionally, based on multi experiments that applied to tune PI controller, it is not necessarily all simulation results based on tuning gains such as negative values, are applicable in hardware design. On the other side, there is little pay attention to improve the speed of the DC motor controller to be measured in the microsecond unit. All of these problems have been considered in this research to be fixing by proposing multi new methodologies. The main objective is to improve the speed performance of DC motor based PI controller in terms of dead (td), rise time(tr), settling times(ts) by estimating precise TF, improving GA performances, and enhancing architecture design to be integrated on Field Programmable Gate Array(FPGA). It is chosen three different direct current motors, and there are three methodologies proposed. Firstly, to propose an accurate TF for the tested DC motors by designing High Speed Motor Data Acquisition System (HSMDAQS) to collect data in data to be imported into System Identification (Sys Ident). The obtained results show that the TF achieved an accurate estimation by increasing the best fit to 95 %. Secondly, is to improve the GA performance based PI controller, by Modified Initialization Fitness Function (MIFF) to overcome the downsides of random searching. Afterward, it is suggested a new procedure to Optimize GA Parameters and Operators (OGA_P0). Simulation results show that the proposed PI controller based Improved GA (IGA) for motors 1,2,3 produces a better improvement for Reduction Step Response Ratio (RSRR) compared with classical GA by 8,9,35 times and over Particle Swarm Optimization (PSO) by 3,3,10 times. The third methodology is to integrate the proposed controller on FPGA, using a new method to run the design based simulink model. Experimentally, it is observed that the Steady State Time (SST) to achieve maximum speed for motors1,2,3 minimized by 10.68%, 8.67%,3.91% respectively, where the significant reduction is achieved in motor 2 to capture 4000 (Revolution Per Minute) RPM at 12.4μs. Finally, the PI controller based IGA providing better speed performance to all experimental motors in terms of response time characteristics to be measured experimentally in the microsecond unit.

Text FK 2020 10 ir.pdf Download (1MB)

Actions (login required)

View Item