This study focuses on an asset inventory system for Internet of Things (IoT) technologies, proposing a performance optimization strategy based on edge computing. We have established a novel system model that leverages heterogeneous computing resources to achieve reduced data processing and transmission delays. During the research, we introduced a task scheduling strategy optimized by a deep learning algorithm, which can intelligently analyze the characteristics of asset data and automatically adjust the task distribution between the cloud and edge nodes, thereby effectively alleviating the computational burden on the central cloud server. Additionally, we incorporated a fast caching structure based on capacity constraints to enhance the access performance of asset information. Experimental results confirm that our optimization strategy significantly improves system performance, particularly in high-concurrency scenarios, where the average response time is shortened by 30% and resource utilization is increased by 40%. Our research holds significant theoretical and practical implications for the application of edge computing in industrial IoT and smart city domains.