IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 58, NO. 7, JULY 2011 3077 Economic Analysis of Canary-Based Prognostics and Health Management Wenbin Wang and Michael Pecht, Fellow, IEEE Abstract—Prognostics and health management (PHM) is a dis- cipline of techniques utilizing in situ monitoring and analysis to assess and trend system degradation and to determine remaining useful life with a defined level of confidence for a specified coverage of anticipated events. One of the techniques for PHM is the use of a canary: an early-warning device derived from the use of a canary bird to detect the presence of poisonous gases in a mineshaft. This paper proposes a cost-based decision model to evaluate the economic benefit of the use of canary devices and the optimal replacement time of a canary-equipped system if the canary issues a warning. This paper also discusses the influence of scheduled preventive maintenance on the replacement policy of a canary-equipped system. The model is demonstrated with an example from the aviation industry. Index Terms—Canary, cost, economic, electronic systems, health management, prognostics, prognostics and health manage- ment (PHM), replacement. I. I NTRODUCTION P ROGNOSTICS and health management (PHM) is a method that permits the assessment of the reliability of a product (or system) under its actual application conditions [1]. The key benefits of applying PHM, as discussed in [1] and [2], include the following: advance warning of failures; minimizing unscheduled maintenance; extending maintenance cycles while maintaining system effectiveness through timely repair actions; reducing the life-cycle cost of equipment by decreasing inspec- tion costs, downtime, and inventory; improving qualification; and assisting in both the design and logistical support of fielded and future systems. There are many methods of implementing PHM, including physics-of-failure and data-driven methods [1]. Expendable devices, such as fuses, have been traditional protection meth- ods for structures and electrical-power systems. For example, circuit breakers are examples of elements used in electronic Manuscript received March 27, 2010; revised June 8, 2010; accepted August 18, 2010. Date of publication September 2, 2010; date of current version June 15, 2011. This work was supported in part by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (CityU8/CRF/09). W. Wang is with the Salford Business School, University of Salford, M5 4WT Salford, U.K. and also with the Prognostics and Health Manage- ment Centre, City University of Hong Kong, Kowloon, Hong Kong (e-mail: w.wang@salford.ac.uk). M. Pecht is with the Prognostics and Health Management Centre, City Uni- versity of Hong Kong, Kowloon, Hong Kong and also with the Center for Ad- vanced Life Cycle Engineering, Electronic Products and Systems, University of Maryland, College Park, MD 20742 USA (e-mail: pecht@calce.umd.edu). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIE.2010.2072897 products to sense excessive current drain and disconnect power. Fuses within circuits safeguard parts from voltage transients or excessive power dissipation and protect power supplies from shorted parts. Another example is a thermostat, which has been used to sense critical temperature-limiting conditions for the system concerned until the temperature returns to normal. In some systems, self-checking circuitry can also be incorporated to sense abnormal conditions and make adjustments to restore normal conditions or activate switching means to compensate for a malfunction [3]. Xiong et al. [4] developed prognostic warning systems for power-electronic modules of an electric vehicle. The word “canary” is derived from the use of canary birds in mines to warn of the presence of hazardous gases. Because canaries are more sensitive to hazardous gases than humans, the sickening of a canary is an indication to the miners to evacuate the shaft. Due to their embedded location, canaries experience substantially the same environmental and operational condi- tions as does the actual system. For example, the stresses that contribute to the degradation of a circuit may include voltage, current, temperature, humidity, vibration, and radiation plus many others, depending on the actual applications. Canaries are more than fuses in that the damage rate is expected to be the same for both the system and the canary circuits. A prognostic canary is thus designed to fail faster through increased stress on the canary structure by means of scaling, which may be achieved by controlling the increase of the stress (e.g., current density) inside the canaries. In canary- based prognostics, the output information from the canary is usually binary, i.e., failed or not failed, but physics-of-failure and data-driven algorithms can be embedded or used with the canary to assess system health and make prognostic decisions on a continuous basis. In this paper, we focus only on the binary type of canary output. Canary devices mounted on an actual system can be used to provide advance failure warning due to specific wear-out failure mechanisms [1]. For example, Mishra and Pecht [5] studied the applicability of semiconductor-level health monitors by using precalibrated cells (canaries) located on the same chip as the actual circuitry. Han et al. [6] proposed a concept of developing a “canary-containing” packet that can be attached externally to weapon casings to receive environmental loading identical to what the casings experienced. Wang and Calhoun [7] demonstrated a 90-nm 128-Kb test chip on which canary cells track changes in temperature and data-retention voltage to detect the minimal value of a supply voltage. Calhoun and Chandrakasan [8] proposed a closed-loop approach using canary flip-flops to enable power savings of over 40 times in a 0278-0046/$26.00 © 2010 IEEE