Wireless ad hoc and sensor networks have been becoming increasingly essential in recent years because of their ability to monitor and manage real-time situational information for various novel services. Therefore, we propose an adaptive intelligent home energy management system (AiHEMS) using a wireless ad-hoc and sensor network, which consists of a monitoring sensor and a management gateway. The AiHEMS adopts an information convergence mechanism, so that when the system generates patterns it interconnects with the AiHEMS in other domains and then utilizes information from other domains for pattern generation. We also propose adaptive light-weight middleware architecture which is modified based on a system's role, a situation, and a policy, which leads to reduction in sensor's system resource. To verify the efficiency of our system, we built a test bed using two offices and conducted experiments based on light and air-conditioner control for saving power. The results show that the power saving using dynamic patterns without information convergence and with information convergence are approximately 18% and 26% respectively compared to fixed value based control.