The Internet of Things (IOT) management platform is used to manage and transmit data from a variety of terminal devices in the power system. In terms of detecting abnormal data, the existing IOT management platform has a low data processing efficiency and a high error rate. In addition, the optimal selection and determination of the structural parameters of a convolutional neural network (CNN) have a substantial effect on its prediction performance. On this basis, the paper proposes a decision algorithm for locating anomalous data in an IOT integrated management platform using a CNN and a global optimization decision of key structural parameters of a CNN using an improved particle swarm optimization (APSO) algorithm. Initially, an index model is developed to identify whether the data obtained from the IOT management platform is abnormal. Second, the structure of the CNN‐based anomaly detection approach is investigated. Next, an improved particle swarm optimization approach is designed to optimize the structural parameters of the CNN, and an APSO‐CNN with higher performance for anomalous data localization is constructed. Using the Adam optimizer, the accuracy, feasibility, and efficiency of the established method were assessed. The results demonstrate that the developed APSO‐CNN‐based decision method for anomaly data localization offers significant advantages in terms of precision and execution speed, with potentially intriguing application potential.