Considerations on LAMDA Controller With BB–BC Tuning Toward Hardware Implementation

In real-world industrial processes, achieving self-tuning of control unit parameters using an intelligent optimization algorithm offers significant advantages in adapting to the diverse characteristics of dynamic systems. Conventional controllers often fall short in real-time application scenarios and reliability requirements. In response, adaptable learning algorithms have gained acceptance across various industrial control applications. Handling complex computational tasks, extensive training data, and multivariable demands requires a robust platform to achieve optimal results. In this article, we present a multi-input, multioutput (MIMO) control system that utilizes the Big Bang–Big Crunch optimization algorithm alongside the learning algorithm for multivariate data analysis (LAMDA). Our modifications to the LAMDA controller involve using the ordered weighted average instead of
T
-norm or
S
-norm as aggregate functions. Additionally, we adopted field-programmable gate array (FPGA) technology to improve execution speed and energy efficiency compared to conventional platforms (e.g., microcontrollers). We modeled the MIMO system for heating, ventilation, and air conditioning system using state space equations. The hardware system was designed with (very high-speed integrated circuit) hardware description language (VHDL). To demonstrate our approach, we implemented the entire system on an AMD/Xilinx Virtex-7 FPGA, and experimental results confirmed the feasibility of our proposed optimization strategy.