Universal direct tuner for loop control in industry

This paper introduces a direct universal (automatic) tuner for basic loop control in industrial applications. The direct feature refers to the fact that a first-hand model, such as a step response first-order plus dead time approximation, is not required. Instead, a point in the frequency domain and the corresponding slope of the loop frequency response is identified by single test suitable for industrial applications. The proposed method has been shown to overcome pitfalls found in other (automatic) tuning methods and has been validated in a wide range of common and exotic processes in simulation and experimental conditions. The method is very robust to noise, an important feature for real life industrial applications. Comparison is performed with other well-known methods, such as approximate M-constrained integral gain optimization (AMIGO) and Skogestad internal model controller (SIMC), which are indirect methods, i.e., they are based on a first-hand approximation of step response data. The results indicate great similarity between the results, whereas the direct method has the advantage of skipping this intermediate step of identification. The control structure is the most commonly used in industry, i.e., proportional-integral-derivative (PID) type. As the derivative action is often not used in industry due to its difficult choice, in the proposed method, we use a direct relation between the integral and derivative gains. This enables the user to have in the tuning structure the advantages of the derivative action, therefore much improving the potential of good performance in real life control applications