Many previous studies have concerned with the effects of land use/land cover changes and landscape patterns on the urban heat island. However, accurate identification of key patches and nodes with high contribution to urban heat island need to be urgently studied. Using Yunlong Demonstration Zone of Zhuzhou, a new-type city, as the case study area based on morphological spatial pattern analysis (MSPA) and landscape connectivity model, this study constructed an urban heat island network to identify the spatio-temporal dynamic characteristics of core thermal landscape patches and nodes in process of urban development. The results showed that:(1) from 2007 to 2018, heat island patches tended to be concentrated in Yunlong Demonstration Zone, and the area of heat island increased significantly compared with the initial stage of the study. From the spatial distribution of land surface temperature, the contrast between urban area and surrounding countryside was more distinct. (2) During the study period, the area of heat island core class was the largest, and the area proportion showed an obviously rising trend. While the area proportion of edge class, branch class, islet class, bridge class, loop class and perforation class showed a decreasing trend to varying degrees, which was largely related to the increase of aggregation degree between heat island patches. (3) The top 20 thermal landscape patches with higher importance value of landscape connectivity were identified, and the importance value and area of the top 3 patches were relatively large, which should be deserved more attention and optimization. (4) Patch-based cooling measures adopted to avoid the connection between node could effectively alleviate the heat island effects in the whole region. The results of this study have important practical significance for the planning and sustainable development of small and medium-sized cities in China.