Online Prediction Product Control Based on Information-Aware Weight and Error Compensation

This paper applies the information-aware weight method to the field of engineering where the validity of the data characteristics of the prediction object fluctuates or changes rapidly within a limited range. It clarifies the relationship between point similarity and regional similarity, so that the online prediction algorithm has a more acute data characteristic perception ability in the process of responding to data characteristic changes, and can control the life and performance of the product in real time. The error compensation algorithm is based on the relevant theory of the prediction ability solidification after the prediction model training and the use of prior error to approximate the posterior error in the prediction process of the online extreme learning machine derivative algorithm. By establishing a mapping relationship between the input data and the prediction error of the IPW-OSELM algorithm, an error compensation algorithm is formed. Starting from the cost function theory, the IPW-OSELM algorithm effectively integrates multiple parameters such as forgetting factor, regularization parameter and information-aware weight, which improves the overall performance of the WE-OSELM algorithm online prediction and can effectively manage large systems.