Method for adjusting parameters of a coating process to manufacture a coated transparent substrate

A method for adjusting at least two parameters of a coating process to manufacture a coated transparent substrate comprising a multi-layered coating according to a targeted value for at least one quality function for said coated transparent substrate,

• wherein said method takes, as input, a dataset of values for at least one quality function and a dataset of values for at least two coating process parameters;
• wherein said datasets come from a manufacturing history dataset of coated transparent substrates in one or more coating systems;
• wherein said method provides, as output, optimized values for the at least two coating process parameters to manufacture said coating transparent substrate, and
• wherein said method comprises the following steps:
• (a) computing, on a computer, a dataset of at least two different mathematical prediction models between differences of the values of the at least one quality functions and differences of the values of the at least two coating process parameters,
• wherein said at least two different mathematical prediction models are based on different machine learning methods, and wherein said different machine learning methods are trained on differences of the values of the dataset for the at least one quality function as features data and on differences of the values of the dataset for the at least two coating process parameters as target data;
• (b) manufacturing a coated transparent substrate with standard values for the at least two process parameters;
• (c) measuring or predicting a value of the at least one quality function of said manufactured coated transparent substrate;
• (d) computing, on a computer, the difference between the measured or predicted value of the said at least quality function and a targeted value for said at least one quality function;
• (e) computing, on a computer, differences for the at least two coating process parameters from said difference of measured or predicted value for said at least one quality function with each mathematical prediction models of the dataset;
• (f) manufacturing a coated transparent substrate with adjusted values for the at least two coating process parameters which are adjusted according to differences computed at step (e);
• (g) reiterating steps (c) to (f) until the targeted value for the quality function is reached according to a given error interval or until the variation of said at least two coating process parameter is below a given threshold upon subsequent iteration, the values obtained for the at least two coating process parameters to reach said targeted value for the quality function are the optimized values of the said at least two coating process parameters.