Modeling and Parameter Identification of a Cooling Fan for Online Monitoring

In many industrial fields, cooling fan systems have been widely implemented in electronic equipment such as network hosts, product line junction boxes, manufacturing facilities, computer numerical control (CNC) machine systems, cooling units of array servers, and many other power systems. These systems usually require high computation power and release high heat. Once the cooling fan systems malfunction or the cooling efficiency degrades, it could result in lower system performances or even cause serious damage to the core systems. Thus, there is a growing interest in monitoring and detecting the cooling fan operation status. As a result, the status of the cooling fan systems needs to be monitored in real-time. In this article, dynamics modeling, parameter identification, and online fan speed monitoring of a cooling fan system are presented. First, the nonlinear model of the cooling fan system is derived from blade aerodynamics with a driving motor. Next, a discrete model is applied based on the bilinear transformation of the description of the dynamic behavior and is further used for the least-square (LS) parameter estimation. To suppress the measurement noise, a regulation filter (RF) is further presented to improve the parameter identification precision. In addition, a Levenberg-Marquardt (LM) optimization is further applied for parameter refinement. Simulation comparison studies are considered to validate the proposed method. Moreover, many experiments are conducted to verify the feasibility and reliability for different types of fans. Unlike common conventional monitoring methods, the proposed framework does not apply constant threshold or need any training stages. The alarm threshold is adjusted automatically according to the current operation status. Finally, an embedded measurement and monitoring instrument is developed for demonstrating the effectiveness of the proposed method. Experiments firmly verify the novelty of the model-reference-based online cooling fan monitoring techniques.