Since Internet of Things (IoT) gadgets are restricted in energy resources, energy consumption is a serious problem. In IoT networks, conventional routing protocols cannot take energy consumption into account, which results in early network downtime and a shorter lifespan of the network. Because of this, the article suggests a dynamic energy-efficient routing protocol that can react to the varying energy levels of network nodes by using machine learning (ML) to forecast the amount of energy consumed and modify routing patterns to reduce energy consumption. Using cluster- based technologies to shorten the distance between sensor nodes is one common way to lower the energy consumption of wireless sensor networks (WSNs). The Variance Ratio (VR) technique is used initially to determine the ideal number of clusters. To accomplish the major objective of this work, this improvement will occur through two stages that pass using better ways to enhance the chain-greedy hierarchical routing protocol (CGHRP). The fuzzy c-means (FCM) technique was then used to increase the network lifetime. As a consequence, data is carried via several shorter parallel lines instead of a single lengthy route. The Python tool is used for simulating the protocol, and the findings, especially regarding energy savings, are clear and useful. The findings from the simulation demonstrate that in terms of network lifespan and energy consumption, the suggested protocol (CGHRP-FCM) performs better than the state-of-the-art energy-efficient routing protocols. The protocol is suited for IoT networks wherein nodes are portable and have restricted resources for energy because of its dynamic nature, which enables it to adapt to modifications in the network architecture and energy levels of nodes.