Sensor Reliability Evaluation Index System for Main Equipment of Power Distribution IoT

Background: Power distribution Internet of Things is a smart service system in which traditional distribution networks comprehensively apply modern computing technologies, such as cloud computing, mobile Internet of Things, and big data to realize the interconnection of all things in the power system and human-computer interaction. In the power distribution Internet of Things, there are many types of main power distribution equipment, and each type of power distribution equipment is connected with multiple sensor devices. These sensor devices play an important role in the smooth operation of the power distribution Internet of Things, so it is necessary to build a reliability evaluation system for these sensor devices. Objectives: The objective was to evaluate the reliability of the sensor device connected to the main equipment in the power distribution Internet of Things. Methods: Firstly, the G1-CV method (entropy weight method-coefficient of variation method) is used to construct the initial evaluation index system, and the comprehensive weight of each index is determined and ranked. Then the improved TriMap algorithm (ITriMap) is used to reduce the high-dimensional complex index system. Results: Simulation verification results show that the algorithm has a smaller reconstruction error than traditional dimensionality reduction algorithms such as t-SNE, Largr-Vias, UMAP, etc. The reduced index can scientifically and objectively measure the main equipment sensors of the power distribution network. Conclusion: An index reduction method based on the ITriMap algorithm is proposed to evaluate the reliability of sensors. Experimental results show that this algorithm greatly reduces the reconstruction error compared to other traditional methods. Discussion: In the future, on the basis of this article, we intend to construct a sensor reliability evaluation model for the main equipment of the power distribution Internet of Things using the 5- dimensional features after dimensionality reduction as input.