摘要
Performance of quality monitor models in spot welding determines the monitor precision directly, so it’s crucial to evaluate it. Previously, mean square error (MSE) is often used to evaluate performances of models, but it can only show the total errors of finite specimens of models, and cannot show whether the quality information inferred from models are accurate and reliable enough or not. For this reason, by means of measure error theory, a new way to evaluate the performances of models according to the error distributions is developed as follows: Only if correct and precise enough the error distribution of model is, the quality information inferred from model is accurate and reliable.
Performance of quality monitor models in spot welding determines the monitor precision directly, so it’s crucial to evaluate it. Previously, mean square error (MSE) is often used to evaluate performances of models, but it can only show the total errors of finite specimens of models, and cannot show whether the quality information inferred from models are accurate and reliable enough or not. For this reason, by means of measure error theory, a new way to evaluate the performances of models according to the error distributions is developed as follows: Only if correct and precise enough the error distribution of model is, the quality information inferred from model is accurate and reliable.
