Method for “real time” in-line quality audit of a digital ophthalmic lens manufacturing process

A non-transitory computer-readable medium storing a program including instructions that, when executed by a processor, causes the processor to perform the steps of a method for quality auditing of a digital manufacturing process of ophthalmic lenses, in real time, applicable to each ophthalmic lens produced, said lens having given lens and environment parameters, said method involving the following computer-implemented steps of:

• setting up a computable single lens global quality criterion (or LGQC), providing a quality quantification for every manufactured lens of a first selected representative learning set of measured lenses produced by the manufacturing process, so that the computation result is a computed LGQC able to match or to mimic, automatically and with high repeatability, a quality evaluation that would be obtained from an expert, wherein the single LGQC of a lens is a unique transformation between said lens and a unique value, based on a deviations or error map of an appropriate and sufficiently large selected number of measurement points spread on the whole lens surface, said deviations map being built as the difference at each point between the measured map of actual respective through or surface optical power and the corresponding theoretical or reference map of respective through or surface optical power;
• said measured map of actual respective through or surface optical power being corrected by a correction performed according to the knowledge of the expected systematic deviations due to one or more manufacturing process steps which are independently controlled and evaluated;
• learning a mathematical model and optimizing the same, based on a second selected representative learning set of measured lenses having given lens and environment parameters and produced by the manufacturing process as model input, the model output being an expected lens global quality criterion, or expected LGQC, which is defined as the LGQC of ophthalmic lenses produced using the manufacturing process in a given, generally stable and fixed manufacturing process state under control, and minimizing the difference between the model output for said expected LGQC and the computed LGQC based on said second selected learning set, said model learning and optimization including the identification, based on said second selected learning set, of relevant input lens and environment parameters playing a role on the lens quality during normal production, so that the LGQC model output will be based on a set of parameters gathered within these relevant input lens and environment parameters, said model being able to provide an expected LGQC for each lens further produced in normal production according to its lens and environment parameters, the latter being the input of the model;
• providing a first quality score, called process quality score for the digital lens manufacturing process, said process quality score being calculated from measurements on a single manufactured lens or on a limited number of manufactured lenses obtained by normal production, and being independent of lens and environment variants, said method further involving the following steps:
• evaluating the computed LGQC for every lens processed;
• selecting a number X of lenses, taking into account the time period allocated for at least one of the evaluation and statistical considerations regarding the signal-to-noise ratio (SNR) limit;
• quantifying the process quality score on the basis of an average of the normalization of the actual computed LGQC for any manufactured lens with its lens and environment parameters, said normalization being based on a comparison with the expected LGQC for said manufactured lens with its lens and environment parameters, and the expected lens global quality criterion for a given, generally stable and fixed manufacturing process state under control;
• using the process quality score as a quality feedback of the manufacturing process.