Protection measurement circuits are the cornerstone of power system relay protection, and their error assessment is crucial for the stable and secure operation of the power grid. Aiming at the risk that the static hidden errors in protection measurement circuits may lead to protection relay maloperation or failure and are difficult to monitor online, this paper proposes a data-driven error assessment method based on recursive principal component analysis and extreme learning machine optimized by grey wolf optimization (RPCA-GELM). First, using historical and real-time data of the power system under normal operation, RPCA is applied to update the principal component feature model online, reducing the assessment time. Then, four classical statistical algorithms are introduced to generate four types of error monitoring feature quantities, and a comprehensive error evaluation method is constructed to optimize feature selection to improve the accuracy of error assessment. Next, considering that the model assessment accuracy depends on the key parameters C and , an infinite folding chaotic mapping strategy is introduced to optimize the gray wolf algorithm, improving parameter optimization accuracy and convergence speed. On this basis, combined with the ELM algorithm, an error assessment method for protection measurement circuits is proposed using the GELM algorithm. Finally, multiple sets of comparative experiments vilify that the proposed method can optimize the model performance and effectively improve the accuracy and precision of error assessment in protection measurement circuits compared with other methods.