The emerging internet of things (IoT) is all about connecting machines and systems together through sensors and actuators. The meaningful information collected from the sensors can be processed timely and possible actions can be taken either by the actuator or human to enhance productivity, efficiency, and reliability. We will see IoT devices and embedded systems become more and more prevalent in our daily lives. One such area is the power generation industry. Not surprisingly, the power generation industry has been one of the early adopters of IoT. The expensive electricity generator is embedded with massive internet-connected sensors for its operation condition monitor to prevent potential catastrophic failure. This dissertation presents the development of a low-cost, easy-to-deploy, self-powered wireless sensor nodes and integrates them with existing solutions into wireless sensor networks to monitor temperature and vibration information of the instruments in a power plant (e.g., motor, pump, etc.). Each sensor node includes: (a) energy harvesting components (which convert the vibrational motion of the instruments into electricity) made by a piezoelectric material; (b) a power management application-specific integrated circuit (ASIC) for AC to DC, DC to DC and voltage regulation; (c) the multi-physical sensor components, including temperature and vibration sensors; (d) an ultralow-power wireless communication unit using ZigBee protocol. We design a customer-specific printed circuit board (PCB) to integrate the sensor node components and a robust package to protect the sensor node for field deployment and harsh environment applications. The sensors communicate with a powered gateway unit intermittently while constantly harvesting and storing energy, maintaining always-on operation without external energy input. Attribute reporting allows the system to transmit data in a more intelligent and energy-efficient way. These sensor nodes can be installed by the end-users (requiring no professional training) in a non-intrusive way by retrofitting or mounting onto an equipment's surface using strong magnetic tape. The proposed wireless sensors can be implemented in any industrial power plant and require minimal maintenance while providing continuous, long-term monitoring of equipment with the real-time readout of temperature and vibration, and possibly other parameters of interest.