Research on anomaly detection algorithm for spacecraft payload action execution status

Addressing issues such as high-dimensional satellite telemetry data and data missing, the need for mass telemetry data for long-term training of payload action status monitoring models, and the limited scalability of existing monitoring method, this paper proposes a telemetry parameter model-oriented payload action patterns and a telemetry parameter comparison algorithm based on payload control plans. The proposed model utilizes principal component analysis (PCA) to identify key telemetry parameters specific to particular payload action patterns, significantly reducing data dimensionality and computational complexity. Additionally, the model generates simulated telemetry sequence to serve as a reference for evaluating actual telemetry parameter sequence. Building on these key telemetry parameters and the simulated sequence, the proposed algorithm processes the telemetry data by filling in missing values, removing null telemetry values, and standardizing the data. The algorithm then employs an improved entropy weight method to calculate the weights of telemetry parameters for payload action determination and compress the telemetry data, thereby reducing false alarm rate due to data supplement and lowering the computational load, to achieve the judgment of the execution of the satellite payload action. Validation using real telemetry data from a specific satellite model demonstrated a high accuracy of 87.5% in determining the execution status of satellite payload actions, surpassing the performance of direct sequence comparison post-preprocessing. This method provides ground-based operation control personnel with a reliable basis for assessing the operational status of satellite payloads.