چكيده به لاتين
Control charts are one of the most powerful tools for detection of reasonable deviations in a process. As a control charts alarms for an out of control occurrence, any delay will result in quality costs, such as the cost of modifying the process. While detection of real-time changes leads to a significant saving in time or cost, Change point detection, expedites diagnosis of the origin of the deviation and decreases decisions dependence on the previous knowledge. In most applicative programs, the performance of a process or quality of a product is defined by a regression relation, named profile in statistical process control literature, between a response variable and one or more explanatory variables.
In manufacturing mechanical products and items, the shape of a product is considered as a qualitative characteristic. Many points on the surface are measured with great precision to detect existing defects. Such observations, describing geometric characteristics and have a spatial correlation, are named profile. Roundness geometric profile is a special type of the profiles that each of the data is dependent on preceding and succeeding data.
In this research, a method for detecting step changes point and gradual changes in the roundness geometric profile, using maximum likelihood function in the situation that change occurs in large-scale coefficients and small-scale coefficients, will be provided.
