HUMIDITY CONTROL OF A NEONATE INTENSIVE CARE UNIT USING CDM Felipe C. Freitas Francisco V. Andrade Bismark C. Torrico José C. T. Campos Universidade Federal do Ceará camaraf12@yahoo.com.br, vanierandrade@ifce.edu.br, bismark@dee.ufc.br, teles@dee.ufc.br Abstract. This article proposes a dead-time compensator using CDM (Coefficient Diagram Method) design method for controlling humidity in a neonate intensive care unit. It was used a Modified Smith Predictor (MSP) structure for dead- time compensation. Also was done a stability and robustness analysis according to the CDM Diagram. It was done simulations and experiments and is shown the results using the proposed controller. It was done still a simulation for showing results of modelling errors. Keywords: CDM, Dead-Time, Smith. 1. INTRODUCTION Dead-time or time delay is a common phenomenon found in control process and chemical industry. The dead-time occurs because of mass or energy transportation, to time lag of low order systems in series. Some systems has variable dead-time, as those used in distributed systems, whose information stream in the communication network is changing. When present in a feedback loop, the dead-time can be a serious obstacle for good process operation,because it reduces the gain and phase margins, that can drive the system to instability. When the ratio between dead-time and time constant is low, a classical controller as PID (Proportional - Integrative and Derivative) can be used, but when this ratio is higher than 0.5, is necessary using a dead-time compensator (Normey-Rico and Camacho, 2009). The first dead-time compensator was proposed by (Smith, 1957), but is still in use because it’s easy implementation. This compensator is effectively used in stable plants. It’s main advantage is that the dead-time is removed from the transfer function of control loop relating the output and set-point. However, this dead-time compensator has some disadvantages, because it has instability under model mismatching and variable delays, besides it has a poor disturbance rejection (Suksri and Tunyasrirut, 2009). For coping these problems, it has been proposed some modification to the original structure. In the last twenty-five years, dead-time has been received great attention by academic community(Normey-Rico and Camacho, 2009). Some authors have proposed different structures for dead-time compensation. In this article, was used a structure proposed by (Normey-Rico and Camacho, 2009), because it’s implementation facility. However, some controller tuning procedures are complex or have many parameters for using. In this article was used a CDM method for controller design. The experimental result was applied in a neonate intensive care unit. Recently, a controller GPC was applied to this kind of equipment (Freitas et al., 2014). This article is divided as following: Section 2 describes the Coefficient Diagram Method approaching some important aspects for this method, section 3 Proposed Controller, where is done the controller design using a dead-time compensator, the section 4 Simulation and results show simulation and experimental results obtained for controller application. The section 5 shows the Coefficient Diagram used for robustness analysis. 2. Coefficient Diagram Method The CDM is a controller design method using polynomial approach to present the system, this method was proposed by Shunji Manabe in 1992. The transfer function for numerator and denominator are considered separately each other. The CDM allocate the closed loop transfer function poles to get desired response, using as design parameters: the stability index (γ i ) and equivalent time constant (τ ). The CDM diagram is shown in Fig.1. For more details about the CDM method see (Manabe, 1998). The polynomial are defined as: D(s)= a n s n + a n-1 s n-1 + ... + a 1 s + a 0 N (s)= m m s m + b m-1 s m-1 + ... + b 1 s + b 0