Dynamic solid state lighting

Abstract

Energy conservation concerns will mandate near-future environments to regulate themselves to accommodate occupants' objectives and best tend to their comfort while minimizing energy consumption. Accordingly, smart energy management will be a needed and motivating application area of evolving Cyber-Physical Systems, as user state, behavior and context are measured, inferred, and leveraged across a variety of domains, environments, sensors, and actuators to dynamically mitigate energy usage while attaining implicit and explicit user goals. In this work, the focus in on the efficient control of a LED-based lighting network. This thesis presents a first-of-its-kind pentachromatic LED-based lighting network that is capable of adjusting its spectral output in response to ambient conditions and the user's preferences. The control of the intensity is formulated as a nonlinear optimization problem and the mathematics governing sensed illuminance, color, and corresponding control (feedback and adjustment) are formally defined. The prototype adjustable light source is capable of maintaining an average color rendering index greater than 92 (nearly the quality of daylight) across a broad adjustable range (2800 K - 10,000 K) and offers two modes of control, one of which is an energy efficient mode that reduces the total power consumption by 20%. The lighting network is capable of measuring the illuminance and color temperature at a surface and adjusting its output with an overall update rate of 11 Hz (limited by the MATLAB kernel). The sensor node features an optical suite of sensors with a dynamic range of 10000 : 1 lx (rms error: 2 lx). The sensor node measures the color temperature of daylight within ±500 K (kelvin). Device testing and validation were performed in a series of experiments in which the radiant power was collected using a radiometrically calibrated spectrometer with an expanded uncertainty (k = 2) of 14% and validated against a model derived by measuring the individual spectra of the system using custom MATLAB tools. A digital multimeter measured the current in the experiments. The work concludes by estimating the energy savings based on the measured optical and electrical data. In environments with moderate ambient lighting, the networked control reduces power consumption by 44% with an additional 5-10% possible with spectral optimization.