IL272614B2 - Method for auditing in " real time" and in-line the quality of a digital ophthalmic lens manufacturing process - Google Patents

Description

METHOD FOR AUDITING IN "REAL TIME" AND IN-LINE THE QUALITY OF ADIGITAL OPHTHALMIC LENS MANUFACTURING PROCESS Field of the invention[0001] The present invention relates to the field of digital manufacturing processes ormethods used for manufacturing spectacle lens in the production laboratories of prescription (Rx) lens.[0002] The invention relates in particular to providing an automatic, reactive andaccurate method allowing to audit the manufacturing process in real time. Such an audit should be based upon the analysis of the results of a measured optical power mapping providing high resolution and upon accurate error map of said optical power on the surface of lenses. This audit applies to products originating from normal production (or daily production) of lenses to be delivered.[0003] The present invention also relates to a process that is applicable to powermeasurements of a single lens surface (reflection measurement, surface scan, etc.) assessing the design replication of this single machined surface of the lens, generally immobilized on a surfacing block, for auditing at least a limited part of the steps of the manufacturing process. [0004] Finally, the auditing method is applicable to any manufacturing process used formanufacturing spectacle lens.

Prior art and problem to be solved[0005] The ophthalmic industry has evolved from a situation where fixed and rigiddesigns were replicated in series (by molding, etc.) at the blank manufacturer’s, by a well- mastered mass-production process of which a few samples are checked at a few surface points to guarantee the replication of the design, up to the rise of digital surfacing which has completely changed the way of manufacturing lenses. Any individualized and customized design can now be directly surfaced (machined) on the lens (which is called "design replication") in the Rx lab.[0006] This evolution has a number of consequences:- the digital manufacturing process is delicate and requires continuous maintenance;- the expected quality of the final product is higher but much more difficult to guarantee given the greater difficulty to detect and quantify a great variety of defects appearing at any position on the surface, on individualized designs (all different); - the responsibility in the manufacturing of the progressive surface has been transferred from the blank manufacturer (mass production) to the Rx lab (individualized production).[0007] The digital manufacturing process involves a sequence of steps (multistepprocess) such as for instance, in the case of digital surfacing: blocking the blank, generating one (or two) lens surface(s), polishing the lens, engraving reference marks on the lens surface, etc. resulting in a great diversity of possible defects affecting the final quality of the lens. [0008] External parameters, called environment parameters, that are not connected tothe lens itself or to the manufacturing process, play a role on process quality (for instance lab temperature, etc.).[0009] Many lens parameters play a role on the quality of the product resulting from aprocess under control or not: material index (refraction), "design" parameters of the lens, and more generally geometrical and optical parameters, etc. In the production laboratories, owing to the huge diversity of products, numerous design parameters may play an important role on lens quality during normal production.[0010] Finally, some parameters related to the manufacturing process, that are notrelated to the lens itself, play a role on process quality (e.g. tool speed, etc.).[0011] Inspection machines are at the center of the lab quality management and arecrucial to maintain at a high level the quality for every product that is shipped.[0012] In many cases, inspection is performed by measuring the through power with afocometer on a limited number of points (one, two or three points for instance) defined by some standards, without assessing the design replication over the whole surface. The whole surface is only checked for strong defects by visual assessment of the lens quality using appropriate lighting for evidencing specific surface defects (for ex: projecting light through the lens and analyzing the image projected on a diffusing screen) without precise objective measurement. [0013] Among the available inspection solutions, inspection based on through powermapping is the most appropriate inspection solution for the RX lenses lab. Thus, inspection based on mapping using through power measurement has become crucial in order to guarantee the quality of the freeform manufacturing process.[0014] Often though, inspection machines based on power mapping are used toclassify the measured lenses in a "binary" way between "good" and "bad" products, but corrective actions are only taken when the reject rate reaches a given threshold, often a long time after the issue occurs in the production process, resulting in waste of time and money. [0015] In many cases, the mapping technology is used for quality audits of laboratoriesas well. Once in a while, a defined set of lenses is produced and the mapping results are analysed by experts. Although these audits provide useful and objective data on the production quality and already lead to better decisions, they are very often late.[0016] More frequent audits are based on the analysis of the evolution of one or moreparameter(s) measured on a given lens that is carefully selected and regularly produced by the lens manufacturing process in order to detect drifts of the latter.[0017] For example, document EP 2 214 868 A1 describes a process allowing tocontrol a process for manufacturing ophthalmic lens comprising the steps of:a) manufacturing a "master" lens according to a manufacturing process, using a specific manufacturing device,b) measuring at least one parameter of the master lens of step a), using at least a measuring device,c) recording the value of the parameter,d) regularly repeating steps a) to c) and checking the evolution of the parameter over time, wherein the evolution of at least one parameter of the manufacturing device used during the lens manufacturing process is controlled/checked over time and the evolution over time of at least one parameter of the master lens is related with the evolution over time of at least one parameter of the manufacturing device.[0018] This document associated with the principle of statistical process control (SPC)using graphical representations under the form of control charts with lower and upper limits for a measurable parameter, based on single measurements, allows for occasionally controlling the manufacturing process for a class of specific products and defects. This process is well known to the person skilled in the art. Accordingly, mass production process may be stopped over defined time intervals and used for manufacturing a master lens, i.e. a lens used as a learning sample, which design is well known and mastered, and includes a well-chosen measurement that is sensitive to the variations of the process parameter(s) to be studied. The aim thereof is to control the stability of the manufacturing process (parameter), from time to time (when the master lens is produced), by comparison with predetermined tolerances, and is not to assess the quality in real time by controlling each lens produced. There is no diversity in the master lens : all the lenses produced have exactly the same design. Only one, or a very limited number of measurements is carried out on each master lens, for example at a point corresponding to near and/or far vision.[0019] Within this frame, the Applicant is already providing on the market inspectiondevices using through power mapping with the resolution and accuracy needed for evidencing the lens defects based on an error map. The latter is calculated as the difference between the measured optical surface / through power map and the reference / theoretical surface / through power map expected from a perfect lens (i.e. with a perfect replication of the design).[0020] An automatic analysis of the error map making lens evaluation easier isprovided. It gives quantitative results under the form of a set of various criteria for the deviation from the theoretical design, calculated on a defined zone of the lens surface. Among the evaluated criteria, a single computable criterion of the global quality of each lens, named Global Mapping Criterion (in short "GMC™", Automation & Robotics, Verviers, Belgium), allows to take into account all types of defect and is valid for all variants of lens / process / environment parameters. This criterion of lens global quality combines a weighted quantification of any defect resulting from the manufacturing process and affecting the quality level of the lens design replication. It has been adjusted to match at best the field expert evaluation. This quantification is based on the error map values discretized into an appropriate number of measurement points spread on the whole surface of the lens. In some way, it provides automatically and with a high repeatability, an evaluation of the global quality of the design replication that is (very) similar to an expert’s evaluation.[0021] In summary, in prior art,- complex multistep process,- great diversity of lenses (see also [1]),- great diversity of defects,- high influence of many lens and environment parameters on quality, inducing varying quality of the lens for a process under control;- huge amount of data provided by the mapping inspection on the produced lenses, etc., make the automatic interpretation of power maps impossible to perform efficiently in terms of manufacturing quality.

Aims of the Invention[0022] The present invention utilizes power error maps such as provided by aninspection process with through or surface respectively optical power mapping, providing information about defects related to most of the critical steps of lens manufacturing and as a result providing the advantage of assessing the quality of the final lens functionality (light refraction).[0023] One aim of the invention is to build an appropriate feedback on the processbased on the results of mapping inspection, thanks to an automatic use of all the information contained in the error map. id="p-24" id="p-24" id="p-24" id="p-24" id="p-24" id="p-24" id="p-24" id="p-24"

[0024] Another aim of the invention is to provide a reactive and accurate audit methodfor a digital lens manufacturing process. Such an audit should be based upon the analysis of the measurement results of an accurate, high-resolution optical power error mapping of the lens surface for the lenses from normal production.[0025] The invention also aims to provide an innovative quality audit of anymanufacturing unit used to carry out one of the lens manufacturing steps, based on high-end mapping inspection and on smart data analysis solutions, intended to enhance quality and yield and to lower costs.[0026] The process presented here is also applicable to surface power measurementsassessing the design replication of a single surface (reflection measurement, surface scan, etc.) of the lens (generally fixed on the surfacing block), for auditing a limited part of the steps, or each step, of the manufacturing process.

Main features of the invention[0027] The present invention is described in more details in the claims hereinafter. Itwill be noted that the term "method" rather designates the modelling and evaluation or auditing method of the invention while the term "process" rather designates the digital process or method for manufacturing ophthalmic lenses.[0028] According to one embodiment of the invention, a method for real-time modellingand quantitatively evaluating the global quality level of an ophthalmic lens is described, said lens having specific lens and environment parameters and being produced by a digital process or method for manufacturing lenses, said modelling and evaluation method being computer- implemented and comprising the following steps of :- based on a first representative learning set of lenses produced by the manufacturing process, setting up a unique computable criterion for the global manufacturing quality of each lens with a view to automatically and with high reproducibility reproduce the expert evaluation work, providing for each manufactured lens a global quality quantification evaluated from the deviation map between the theoretical/reference through or surface optical power map or an equivalent map, and the actual through or surface optical power map, measured and optionally corrected, or an equivalent map; said measured actual power map being optionally corrected based on the knowledge of the systematic deviations expected during one or more manufacturing process steps that are independently controlled and evaluated, the deviation map corresponding to the evaluation of said deviations at an appropriate and sufficiently large number of measurement points spread over the lens surface;- based on a second selected representative set of measured lenses produced by the manufacturing process, obtaining and optimizing by a learning method a mathematical model providing a transformation between the lens and environment parameters of each lens manufactured and the expected global quality criterion if this lens had been produced using the manufacturing process in a specific, generally stable, adjusted and fixed state, by minimizing the difference between the model output for said global quality criterion of the lens and the actual computed value of the global quality criterion of the lens, based on said second lens learning set, said learning method comprising the identification of relevant input lens and environment parameters playing a role on the quality of the lens produced during normal production, based on the second selected lens learning set, so that the model output of the unique measurable global quality of the lenses will be based on a set of relevant input lens and environment parameters.[0029] According to one embodiment of the invention, an audit method for the real­time, in-line quality of the "freeform" production line is provided, by means of a "manufacturing process quality score", built following the normalization of the actual global quality level calculated in relation to the expected value of this score for the lenses manufactured by this same process under control.[0030] According to one embodiment of the invention, a method for providing a"manufacturing unit quality score", for any unit of the digital manufacturing process to be evaluated, called "evaluated unit", is described, said unit quality score being calculated from values calculated on a limited number of manufactured lenses obtained by normal production, independently from variants of the lens and environment parameters, and independently from the control state of the other units involved in the manufacturing process.

Brief description of the drawings[0031] FIG.1 schematically represents the overall scheme of the method according tothe present invention, defining the quality of a digital lens manufacturing process (quality of design replication).[0032] FIG.2 schematically represents the concept of computable global qualitycriterion for a lens, calculated from the quantification of the global error in the design replication. id="p-33" id="p-33" id="p-33" id="p-33" id="p-33" id="p-33" id="p-33" id="p-33"

[0033] FIG.3 represents the increase of the GMC parameter value as the importanceof the defect (here ring defect) increases.[0034] FIG.4 represents the case of a stable process under control producing similarlenses with different lens features (additions), the corresponding evolution over time of the GMC being an unstable signal, due to its sensitivity to the lens parameters (in this case the addition only).[0035] FIG.5A and FIG.5B schematically represent the definition of the quality of thelens manufacturing unit.[0036] FIG.6 shows the evaluation of the quality of a specific lens manufacturing unitfor a defined lens routing.[0037] FIG.7 shows all existing lens routings (linked units, either directly linked, notdirectly linked and not linked) to a specific production unit.[0038] FIG.8 shows the general configuration of the method of the invention where themanufacturing process model of the control lens is adjusted.

Detailed description of preferred embodiments[0039] For the sake of clarity, a number of definitions are given hereinafter, which arean integral part of the disclosure of the invention.

Definitions[0040] Lens (often named freeform lens): an optical lens, generally intended forophthalmic spectacles, often customized, with at least one digitally defined surface to be achieved using a digital manufacturing process.[0041] Quality of a lens (operational definition for easy understanding): ability of a setof intrinsic characteristics of a lens to satisfy its final function, namely light refraction. In practice, all considerations regarding "quality" may be applied to the corresponding error amplitude associated with the measured defects (quality of design replication).[0042] Digital lens manufacturing process: a process using one or more computer-driven means to produce smooth surface(s), digitally defined and possibly complex. Various successive means may be implemented including means for performing the step of material surfacing (on one side, two sides) from a blank or from an additive digital 3D manufacturing, etc. A selected part of the multistep lens manufacturing process is considered here, including one or more machines, tools or operations (see [1]). id="p-43" id="p-43" id="p-43" id="p-43" id="p-43" id="p-43" id="p-43" id="p-43"

[0043] Real time: the indicators are updated when the inspection results of a new lensmanufactured by the manufacturing process are available.[0044] Multistep manufacturing process: a sequence of steps in the manufacturingprocess such as, for digital lens surfacing: choosing and verifying the blank, blocking the blank, surfacing/machining the lens surface, polishing the lens surface, engraving reference marks on the lens surface, etc.[0045] Well-adjusted or under control manufacturing process: in SPC, anymanufacturing process that is stable, i.e. does not trigger the detection rules (such as Western Electric rules) of the control charts.[0046] Digital (lens) manufacturing process unit: the smallest manufacturing sub­process subject to selection in routing operations during lens manufacturing. In extreme cases, it may be either a tool used on a machine or a complete line of machines.[0047] Normal or routine production: as opposed to the production of specific lensesfor testing, a normal production designates the routine daily production of the prescription laboratory.[0048] Lens parameters: set of parameters specifying/defining a lens to be producedand playing a role on the quality of said lens when it is produced: material index (refraction), lens design parameters, and more generally geometrical and optical parameters, etc. In the production laboratories, due to huge diversity of products, numerous lens parameters play an important role on the lens quality during normal or routine production.[0049] Environment parameters: parameters that are external to the lens itself and tothe manufacturing process, but play a role on process quality (for ex. lab temperature, etc.). [0050] Relevant parameters: whether they be lens, environment or manufacturingprocess parameters, relevant parameters designate the parameters actually influencing the final quality of manufactured lenses in a given manufacturing process.[0051] Appropriate learning set or group: a large set of controlled lenses, i.e. a greatnumber of lenses ideally having characteristics or parameters as diversified as in normal production, needed to provide information (learning) on the effect of all the relevant variants from the space of the lens and environment parameters playing a role on the quality of the lens produced during normal production.[0052] Representative learning set or group: a large set of controlled lenses producedby the manufacturing process having all variants of lens defect.[0053] Learning method: whole set of learning methods, whether supervised or not(e.g. principal component analysis (PCA), linear or non-linear multivariate regression, etc.). allowing to perform the steps of selecting influence and modelling factors of the relationships between the influence parameters and the output to be modelled.[0054] Error map: difference between the measured optical through or surface powermap and the reference/theoretical (or target) optical through or surface power map expected from a perfect lens (i.e. with a perfect replication of the design). For the purpose of the present invention, the lens error map is obtained for an appropriate and sufficiently large number of points (e.g. 1,000 points) suitably spread over the whole surface of a lens (discretization of the lens surface).[0055] Error pattern: characteristic pattern qualifying the general contouring of an errormap, as a function of the spatial distribution of the error.[0056] Lens global quality criterion (see FIG.2): computable parameter providing asingle value that is representative of the quality of design replication of a manufactured lens, characteristic of the error pattern of the lens, taking into account all types of concerned defects and valid for all variants of the lens, environment and process parameters (see ref. [1]).[0057] In other words, the unique global quality criterion of each lens is a parameter,whose evaluation provides a single value resulting from the transformation of information from a given error map of a produced lens, said lens being discretized into an appropriate and sufficiently large number of points. This is a global, unique and absolute quantification criterion in the sense that it does not depend on the specific manufacturing tool or on other process and environment parameters. As mentioned above, this criterion is defined to match and to mimic, automatically and with high repeatability, the evaluation that would be obtained from an expert. In particular, the purpose of setting up a single quality criterion being to provide a tool for imitating the evaluation of the expert, this unique quality criterion should be defined on the basis of a set of lenses having sufficient size to scan all or most of the defects encountered during manufacturing. Mathematically speaking, according to one embodiment, it is built from local power errors measured and collected in the error map but it could also involve for example probabilistic mathematical measurements on the appropriate and sufficiently large number of points considered. An example of lens global quality criterion is GMCTM.[0058] The computer lens global quality criterion combines a weighted quantificationfor any type of defect due to the manufacturing process, and affecting the lens quality level using admitted rules or directly evaluated by the expert in the field. This quantification is based on the deviations, compared to the theoretical/reference values, of the actual measured optical power of an appropriate and sufficiently large number of measurement points spread over the surface of the lens. id="p-59" id="p-59" id="p-59" id="p-59" id="p-59" id="p-59" id="p-59" id="p-59"

[0059] Normalization of a value based on a reference: arithmetic quantification of thedistance between the "value" and the "reference" in a given space. Here, a normalization is applied to a measured error (using the measured error as the "value") to get a comparison with the expected error (the expected error becomes the "reference"). So, a unique mathematical transformation of the measure value is used to make it comparable with the reference value (linear normalization, ... ). Another formulation for "normalization" could be the comparison results, the difference or the ratio, etc.[0060] Expected systematic deviations: the deviations between theoretical lens andproduced lens caused at a particular and controlled step of the manufacturing process, which can be independently evaluated. For example, a uniform power shift of a blank can be measured before machining. This measurement will allow to correct the final evaluation of the machining process quality (i.e. not taking into account the blank error) from the lens error measurement (which takes into account the blank error).[0061] Quality of a digital lens manufacturing process: a quality score of the digital lensmanufacturing process, independent of the variants for lens and environment parameters, obtained by the normalization of the computed lens global quality criterion (ex. GMC) on any lens produced, related to the expected specific lens global quality criterion calculated for the same lens in a situation of well adjusted (under control) process using a model taking into account the relevant parameters from the parameter space (lens and environment parameters). In practice, a mathematical expectation value (average, weighted or not) on a limited set of lenses can be calculated in order to limit the statistical fluctuations. To be rigorous, this definition applies either to a homogeneous set of manufacturing units (same manufacturing units for each manufacturing step), or to a given routing (defined as a sequence of manufacturing steps), or to a homogeneous set of routings, or, in case of a non- homogeneous set of manufacturing units, it may apply to a fixed representative set of routings. [0062] Average: a mathematical expectation (average, weighted or not)[0063] Quality of a digital lens manufacturing process unit: a quality score of the digitallens manufacturing process unit, independent of the variants for lens and environment parameters, obtained by the normalization of the computed lens global quality criterion (ex. GMC) on any lens produced, when the manufacturing process involving the unit to be evaluated is under control (except for the unit to be evaluated), related to the expected specific lens global quality criterion calculated for the same lens in the case of the well-adjusted (under control) process by a model taking into account the relevant parameters from the parameter space (lens and environment parameters). In practice, a mathematical expectation value (average, weighted or not) on a limited set of lenses can be calculated in order to limit statistical fluctuations.[0064] Appropriate number of measurement points of the lens (on lens surface):number of points selected to have an optimal measurement resolution with a spatial distribution of sampling on the lens surface, taking into account the Shannon theorem applied to the spatial frequency of the power deviations on the surface linked to the defects to be detected. The number of points is for example comprised between 100 and 100,000 for the defects generated by the digital freeform lens manufacturing process.[0065] Manufacturing process or manufacturing process unit under control:manufacturing process or manufacturing process unit well adjusted, in the general meaning of statistical process control.[0066] Control state (of a manufacturing process or of a manufacturing process unit) :indicator that tells if the process is well adjusted or not, in the general meaning of statistical process control.[0067] FIG.1 shows the overall scheme of the method according to the presentinvention for assessing the quality of a digital lens manufacturing process :- definition of lens and environment parameters among all possible variants of lens and environment variables, characterizing a reference design;- elaboration of a lens manufacturing process model "under control" leading to an expected error;- definition of a multistep manufacturing process (for example : only a single routing is shown on FIG.1) leading to the assessment of the global quality of the manufacturing process from the normalization of the actual measured manufacturing error relative to the expected manufacturing error.[0068] This global evaluation of the process quality brings a feedback to the labmanagement and in case of tool failure (drift, dropout, etc.), suitable actions can be undertaken.[0069] The detailed steps of the auditing method, as well as numberous applicationsthereof, are disclosed hereinafter.

Model for the lens global quality criterion[0070] According to particular embodiments of the invention, a method for providing amodel allowing to assess the expected global quality level of an ophthalmic lens is described , said lens having lens and environment parameters, and being produced by a digital lensmanufacturing process, said method involving the following computer-implemented steps of :- based on a first selected representative learning set of lenses produced by the manufacturing process, setting up, with a view to reproduce the evaluation work of an expert, a single computable criterion for the global manufacturing quality of each lens providing for every manufactured lens a global quality quantification evaluated from the deviation map, between a theoretical or reference respectively through or surface optical power map and the actual, corrected, measured through or surface respectively optical power map, said actual measured power map being corrected based on the knowledge of the expected systematic deviations during one or more manufacturing process steps which are independently controlled and evaluated, the deviation map corresponding to the evaluation of said deviations in an appropriate number of measurement points spread over the lens surface. In concrete terms, the global quality quantification is advantageously built on the basis of a weighted combination of statistical values (e.g. mean, standard deviation) and/or global decomposition coefficients (e.g. coefficients of an error map model under the form of Zernike polynomials, Fourier decomposition) evaluated on the deviations map between the reference/theoretical optical through or surface power map and the actual optical through or surface power map, an/or on any mathematical transform of this map (e.g. derivatives of any order). The global mapping criterion (or GMC, see above) developed by the Applicant as a qualitative/ quantitative error criterion is considered as a lens global quality criterion. GMC is a single number representing the global quality of the design replication based on the error map. It works on any error pattern, independently of the type and position of the defect. As illustrated in FIG.3 for a ring pattern error, the GMC value increases when the importance of the defect increases. GMC, which is a reactive and significant value, based on relevant and significant mapping inspection results, is chosen to create a feedback signal on the process.- based on a second selected and representative learning set of lenses, obtaining by a learning method, a mathematical model and optimizing the latter, said mathematical model providing a transformation between each manufactured lens defined by its lens and environment parameters, and the unique computable criterion for the global quality of each lens expected for ophthalmic lenses produced considering the manufacturing process for a given, generally stable and fixed, manufacturing process state, and minimizing the (one) difference between the output of the expected model for said lens global quality criterion and the computed value of the lens global quality criterion based on said second set of lenses, said learning method including the identification, based on said second learning set of measured lenses, with relevant input lens and environment parameters playing a role on the lens quality during normal production, so that the output of the unique global computable quality criterion of each lens is based on a set of parameters gathered among these relevant input lens and environment parameters.[0071] The first intention of the invention is to set up a model based on the lens globalquality criterion (such as the GMC of the Applicant).[0072] The model reproduces or mimics the behavior of the manufacturing tool. Themodel has to be adjusted by minimizing the deviations to correspond to the operation of this tool.[0073] When the tool/process is under control, it should provide an actual errorevaluated by the lens global quality criterion close to the expected error given by the model.

Process quality score[0074] According to particular embodiments, the method is further providing a qualityscore, called "process quality score" for a digital lens manufacturing process, said quality score being calculated from evaluations on a single manufactured lens or on a limited number of manufactured lenses obtained by normal production, and being independent of variants of the lens and environment parameters, said method further involving the following steps:- digitally evaluating the unique global quality criterion for each lens processed;- selecting a number X of lenses needed for evaluating the quality level of a digital lens manufacturing process independently of the variants of lens and environment parameters, taking into account the time period allocated for the evaluation and/or statistical considerations regarding the influence of the signal-to-noise ratio (SNR) limit on the "process quality score";- quantifying the "process quality score" based on the average of a normalization of the actual measured value of the unique computable global quality criterion of any lens produced, related to the expected value of the global quality criterion of the specific lens by said mathematical model providing the transformation between said lens and its environment parameters for a given state, generally stable and under control, of the manufacturing process.[0075] The model transforms the data corresponding to a specific lens in a specificenvironment into an "expected" GMC. id="p-76" id="p-76" id="p-76" id="p-76" id="p-76" id="p-76" id="p-76" id="p-76"

[0076] The process quality score is quantified by averaging the difference (or ratio)between the actual measured GMC for a manufactured lens and the "expected" GMC.[0077] Analysing actual production data shows that the quality of design replication andthe expected GMC depend on numerous lens parameters because a lens that is difficult to manufacture, is more likely to have a high replication error and thus a high GMC.[0078] Among those numerous lens parameters are: addition, decentration, materialand curvature. For example, the higher the addition, the higher the difficulty to achieve the lens by usual machining and the higher the expected GMC values. Thus, an accurate and reactive feedback providing a quality quantification of the manufacturing process has to be insensitive to lens parameters. Hence, instead of directly using the GMC values, the normalized value of the computed GMC based on the expected GMC should be used if the process is stable and under control. This expected value is provided by a process model under control, calculated on real measurement data from the lab.[0079] Therefore, for a process that is stable and under control, the production of abatch of the same lens leads to the same GMC output, it is a case of mass-production. However, freeform lens production is individualized industry rather than batch production. Therefore, even if the process is stable and under control, at a specific quality level, this process will produce lenses with different GMC, depending on the difficulty to produce the specific lens.[0080] FIG.4 presents a case of stable and under control process producing similarlenses whose only different parameter is addition. The corresponding time evolution of the GMC provides an unstable signal, due to its high sensitivity to the lens parameters, characterizing the intrinsic difficulty to produce each lens. Hence, the GMC signal value is not stable for a process under control at a given quality level, the GMC value corresponds more to the quality of the product than to the quality of the process.[0081] The process quality score that is looked for has to be stable for a process thatis stable and under control, even if the studied lens is more or less difficult to be manufactured. [0082] Therefore, in order to build an appropriate feedback signal on the state of theprocess based on the GMC, the latter should be made insensitive to the lens parameters. Hence, according to the present invention, instead of using the GMC signal itself, the inventors used either the difference or ratio between the calculated GMC and the one expected for the process supposedly under control. In this way, if the GMC is equal to the one expected in the control, the process is still considered as being under control. However, if the calculated GMC is higher than the expected GMC under control, it means that the process has drifted and is no longer under control.[0083] To calculate the expected GMC, which is the expected value of the GMC for theprocess under control, a model of the process under control has to be built. This model should include the relationship between the lens parameters and the corresponding expected GMC, for example, in the case illustrated in FIG.4, the relationship between addition and GMC. [0084] A model calculated using a non-linear multivariate regression leads to goodresults. The normalization of the process quality score as defined above, based on the GMC value, which is a product quality score, has been evidenced for a particular generator (not shown). As expected, as these values are independent of the lens parameters, the signal quantifying the manufacturing process quality is less noisy and more stable than the signal directly built on the GMC measurement of the produced lenses.[0085] The process quality score is therefore a significant and reactive signal to beused to obtain a feedback of the process.[0086] The following three feedback signals were compared: score based on rejectrate, score based on power deviation at the two ISO inspection points and process quality score based on GMC (not shown). In the example, the studied period of time ended with a maintenance step performed in the lab due to the increased number of rejected lenses. The signals based on the reject rate and on the power error analysis at one reference point show an increase at the end of the time period, which corresponds to an increase in the number of rejected lenses. However, the third signal shows a constant growth during the time period. This increase is the sign of a drift of the generator. Consequently, if the lab had monitored this last value, they could have performed preventive action on the generator, avoiding an increase in the number of rejected lenses.[0087] The process quality score is given for each lens, and provides results that aresimilar to the quality feedback signal of a mass-production process, although each lens is different.

Production unit (or machine) quality score[0088] According to particular embodiments of the invention, the method is furtherproviding a second quality score, called "production unit quality score", for any unit of the digital lens manufacturing process to be evaluated, called "evaluated unit", calculated from the evaluation of the process quality performed on a limited number of manufactured products obtained by normal production, and independently of the variants of lens and environment parameters, said method further involving the following steps:- for any digital lens manufacturing unit, recording the "process quality score" for each lens processed;- selecting a number Y of lenses needed for evaluating the manufacturing unit to be evaluated, or "evaluated unit", taking into account the time period allocated for the evaluation and/or statistical considerations regarding the influence of signal-to-noise ratio (SNR) limit on the "unit quality score";- calculating the "unit quality score " of the "evaluated unit", as the average "process quality score" relative to all the Y lenses processed on the "evaluated unit".[0089] The definition of the quality of the lens manufacturing process unit is given inFIG.5A and FIG.5B. The unit quality score is quantified based on a specific number of measured lenses having been processed on the evaluated unit, by determining an average process quality score only taking into account the lenses selectively processed on a particular unit. These data can be used to provide the lab manager with real time machine quality indicators in case of mixed-flow production (lab not organized in lines).[0090] These process quality results may be linked with the production routing data toprovide a real-time feedback on production. Production routing data gather at least the list of machines used to produce a given lens and the entry and exit times of the various corresponding manufacturing units (FIG.6).

Improved unit (or machine) quality score (all existing lens routings with this unit)[0091] In a multistep process, the inspection data of all the lenses that were run throughthe same polisher, for example, will be influenced by the performance of the next and previous machines in the production flows, such as the generator(s) supplying the polisher.[0092] According to particular embodiments of the invention, the method is furtherproviding a third quality score, called "improved quality score of the production unit", for any unit of the digital lens manufacturing process to be evaluated, called "evaluated unit", calculated from the evaluation of a limited number of manufactured products obtained by normal production, independently of variants of lens and environment parameters, and independently of the control state of the other units involved in the manufacturing process, further involving the following steps:- for each digital lens manufacturing unit, recording the "process quality score" for each lens processed; - determining all the manufacturing units of ophthalmic lenses linked by a routing, called "units linked to the evaluated unit" or "linked units";- selecting a set of lenses Y’ processed on the "evaluated unit" and/or on the "linked units", said set being needed for evaluating the evaluated unit, taking into account the time period allocated for the evaluation and/or statistical considerations regarding the influence of signal-to-noise ratio (SNR) limit on the "improved quality score of the unit";- while taking into account the set of "process quality scores" for the set of selected lenses and all the corresponding routings, by mathematically determining the most probable "improved quality score of the production unit" of each unit from the evaluated and linked units.[0093] According to these embodiments, the other units on which the lens has beenprocessed are taken into account. The aim is to eliminate the effect of possible defective units by means of statistical analysis.[0094] The units linked to the evaluated unit can be the units linked directly or not tothe steps before (resp. after) the evaluated unit or of the manufacturing step concerned by the evaluated unit, but not linked directly thereto (see FIG.7).

Application - Adjusting the manufacturing process parameters[0095] According to particular embodiments of the invention, the base method can beused for adjusting the process parameters, by means of the monitoring of the quantification of the expected global quality level of a lens, with any lens and environment parameter, to be produced by the digital lens manufacturing process with said adjustment of the process parameters.[0096] According to these embodiments, the manufacturing process or environmentparameters are modified. The model is adjusted to take into account the lens, process and environment parameters that are modified (see FIG.8).[0097] According to particular embodiments, the manufacturing process parametersare adjusted, leading to a quantification of the quality score of a digital ophthalmic lens manufacturing process or "process quality score", with said adjustment of the manufacturing process parameters.[0098] According to particular embodiments, the process parameters are adjusted,leading to a quantification of the quality score of any digital ophthalmic lens manufacturing process unit, or "unit quality score", with said adjustments of the manufacturing process parameters. id="p-99" id="p-99" id="p-99" id="p-99" id="p-99" id="p-99" id="p-99" id="p-99"

[0099] According to particular embodiments, the evaluated production unit parametersare adjusted, leading to a quantification of the quality score of the digital ophthalmic lens manufacturing process unit, or "production unit quality score", with said adjustment of the manufacturing unit parameters.

Other applications[0100] According to particular embodiments of the invention, the method comprises anadditional step of providing a display, for selected lenses and specific environment parameters, possibly reduced to a 1D, 2D, etc. parameter space, of the error between the actual global quality criterion calculated on the lenses from the learning set and the expected global quality criterion of said lenses as provided by the model, providing a means for detecting possible dropout of the manufacturing process under control, for specific conditions regarding particular values of lens or environment parameters of the parameter space.[0101] The display can be provided under the form of a dashboard representing forexample a production machine quality score at a given time. Each machine of the lab can be represented for example with an error bar. The importance of the machine quality score can be represented with a color (for ex. green: ok; yellow, red : not ok), while the length of the bar can represent the amount of lenses that was recently produced by the machine.[0102] It is also possible to get a continuous audit of the production quality to controlhow the machines perform during a longer period of time. The time evolution of the machine quality score can be monitored for several generators during a certain period. In the example situation discussed, a generator shows a higher error during the whole period of time studied. The visual analysis of the error maps confirms a systematic error. The studied lab also confirmed that this generator is the one that produces the highest amount of rejects (not shown).[0103] Machine quality scores can also be used to detect a drift and to know whenmaintenance is required. A supervision software could detect this drift and display a warning message to alert the lab manager and suggest action.[0104] Flow management is often used to manage and optimize the production flow interms of quantity. Combining this approach with the information provided by the machine quality score allows to manage the production flows with an integrated notion of quality optimization.[0105] For example, if the dashboard shows that a first engraver performs well but isless used while a second engraver is not used and a third engraver, which is mostly used, produces more errors, a smart supervision software could decide to distribute the lenses to the two first engravers, at least until the problems are fixed on the third (not shown).[0106] According to particular embodiments of the invention, the method comprises anadditional step of providing an automatic detection of the drifts of the manufacturing process under control, for specific conditions regarding particular values of lens or environment parameters, for example a specific area of the parameter space.[0107] According to particular embodiments of the invention, the method comprises anadditional step allowing to automatically detect the drifts of the manufacturing process possibly out of control, for specific conditions regarding particular values of lens or environment parameters, for example a specific area of the parameter space.

Bibliography[1] FOGLIATTO, Flavio S., DA SILVEIRA, Giovani J.C., and BORENSTEIN, Denis. The mass customization decade: An updated review of the literature. Int. J. Production Economics, Elsevier, 2012, vol. 138, no 1, p. 14-25.

Claims (15)

1./ CLAIMS 1. A method for modelling and quantitatively evaluating the global quality level expected of an ophthalmic lens in real time, said lens having particular lens and environment parameters, and being produced by a digital lens manufacturing process, said modelling and evaluation method being computer-implemented and comprising the following steps: - based on a first selected, representative learning set of measured lenses produced by the manufacturing process, setting up, with a view to automatically and with high repeatability reproduce an expert's evaluation work, a single computable criterion of the global manufacturing quality of each lens, providing for every manufactured lens a global quality quantification that is evaluated from the deviation map between the reference or theoretical respectively through or surface optical power map and the actual measured and corrected through or surface respectively optical power map, said correction of the actual measured power map being performed based on the knowledge of the systematic deviations expected during one or more of the manufacturing process steps, independently controlled and evaluated, the deviation map corresponding to the evaluation of said deviations in an appropriate and sufficiently large number of measurement points spread over the lens surface; - based on a second selected representative learning set of measured lenses produced by the manufacturing process, obtaining and optimizing by a learning method a mathematical model providing a transformation between the lens and environment parameters of each lens manufactured and the expected global quality criterion if this lens had been produced using the manufacturing process in a given, generally stable, adjusted and fixed state, by minimizing the difference between the model output for said global quality criterion of the lens and the actual computed value of the global quality criterion of the lens, based on said second selected learning set of measured lens, said learning method comprising, based on said second selected learning set of measured lens, the identification of relevant input lens and environment parameters 272614/ playing a role on the quality of the lens produced during normal production, so that the model output of the unique measurable global quality criterion of each lens will be based on a set of parameters selected from among the relevant input lens and environment parameters.

2. The real-time modelling and evaluation method according to Claim 1, characterized in that the appropriate and sufficiently large number of measurement points spread over the lens surface is determined on the basis of the Shannon theorem, taking into account a maximal spatial frequency of the power deviations and is comprised between 100 and 100,000, preferably between 1,000 and 100,000.

3. The real-time modelling and evaluation method according to Claim 1, characterized in that the appropriate and sufficiently large number of measurement points spread over the lens surface is 1,000.

4. The real-time modelling and evaluation method according to Claim 1, characterized in that the learning method of the mathematical model is an automatic learning method, supervised or not, as a linear or non-linear multivariate regression or a principal component analysis, PCA.

5. The real-time modelling and evaluation method according to Claim 1, characterized in that it further provides a first quality score, called "process quality score", for the digital lens manufacturing process, said process quality score being calculated from digital evaluations on a single lens manufactured or on a limited number of lenses manufactured by normal production, and independently of the variants of the lens and environment parameters, said modelling and evaluation method further involving the following steps: - digitally evaluating the measurable global quality criterion for each lens processed; 272614/ - selecting a number X of lenses required for evaluating the quality score of the digital lens manufacturing process, independently of the variants of the lens and environment parameters, taking into account the period of time allocated for the evaluation and/or statistical considerations related to the influence of the signal-to-noise ratio limit on the "process quality score"; - quantifying the "process quality score" from the normalization of the actual calculated value of the global quality criterion for each lens manufactured relative to the expected value of the global quality criterion provided by said mathematical model, providing the transformation between each manufactured lens and its lens and environment parameters and the expected value of the global quality criterion of the lenses for a manufacturing process in a given state, generally stable, adjusted or under control and fixed.

6. The real-time modelling and evaluation method according to Claim 5, characterized in that it further provides a second quality score, called "quality score of the manufacturing unit", for any digital lens manufacturing process unit to be evaluated, called "evaluated unit", said unit quality score being calculated from evaluations based on a limited number of lenses manufactured by normal production, and independently of the variants of the lens and environment parameters, said modelling and evaluation method further involving the following steps: - for any digital lens manufacturing unit, recording the "process quality score" for each lens processed; - selecting a number Y of lenses required for evaluating the manufacturing unit to be evaluated, or "evaluated unit", taking into account the period of time allocated for the evaluation and/or statistical considerations related to the influence of the signal-to- noise ratio limit on the "unit quality score"; - calculating the "unit quality score" of the "evaluated unit", as an average "process quality score" relative to all the Y lenses processed on the "evaluated unit". 272614/

7. The real-time modelling and evaluation method according to Claim 5, characterized in that it provides a third quality score, called "improved quality score of the manufacturing unit", for any digital lens manufacturing process unit to be evaluated, called "evaluated unit", said improved unit quality score being calculated from evaluations based on a limited number of lenses manufactured by normal production, independently of the variants for lens and environment parameters, and independently of the control state of the other units involved in the manufacturing process, said method further involving the following steps: - for each digital lens manufacturing unit, recording the "process quality score" for each lens processed; - determining all the ophthalmic lens manufacturing units linked by a lens production routing/flow, called "units linked to the evaluated unit" or "linked units"; - selecting a set of lenses Y processed on the evaluated unit and/or on the "linked units", said set being required for evaluating the evaluated unit, taking into account the period of time allocated for the evaluation and/or statistical considerations related to the influence of the signal-to-noise ratio limit on the "improved quality score of the unit"; - while taking into account the set of "process quality scores" for the selected set of lenses Y and all the corresponding routings, mathematically determining the most probable "improved unit quality score of the manufacturing unit" for each unit from the evaluated and linked units.

8. The real-time modelling and evaluation method according to Claim 1, characterized in that the manufacturing process parameters are adjusted, resulting in a quantification of the expected global quality level given by the mathematical model for a lens, having lens and environment parameters, produced by a digital lens manufacturing process with said adjustment of the manufacturing process parameters. 272614/

9. The real-time modelling and evaluation method according to Claim 1, characterized in that it comprises an additional step of providing a display, for selected lenses and for environment parameters, possibly reduced to a 1D, 2D, etc. parameter space, of the error between the actual global quality criterion of the measured lenses of the first learning set and the expected global quality criterion of the lenses as provided by the model, this display thus providing a means for detecting dropout of the manufacturing process under control, for specific conditions regarding specific values of lens and/or environment parameters, for example part of the parameter space.

10. The modelling and evaluation method according to Claim 1, characterized in that it comprises an additional step of providing a display, for selected lenses and environment parameters, possibly reduced to a 1D, 2D, etc. parameter space, of the error between the actual global quality criterion of measured lenses and the expected global quality criterion of the lenses as provided by the model, this display providing a means for detecting drift of the manufacturing process possibly out of control, for specific conditions regarding particular values of lens and/or environment parameters, for example part of the parameter space.

11. The real-time modelling and evaluation method according to Claim 9, characterized in that it comprises an additional step of providing an automatic detection of the drifts of the manufacturing process under control, for specific conditions regarding particular values of lens and/or environment parameters, for example part of the parameter space.

12. The real-time modelling and evaluation method according to Claim 10, characterized in that it comprises an additional step providing an automatic detection of the drifts of the manufacturing process possibly out of control, for specific 272614/ conditions regarding particular values of lens and/or environment parameters, for example part of the parameter space.

13. The real-time modelling and evaluation method according to Claim 5, characterized in that the manufacturing process parameters are adjusted, resulting in the quantification of the quality score of the digital lens manufacturing process or "process quality score", with said adjustment of the manufacturing process parameters.

14. The real-time modelling and evaluation method according to Claim 6 or 7, characterized in that the manufacturing process parameters are adjusted, resulting in the quantification of the quality score of the digital ophthalmic lens manufacturing process unit, or "unit quality score", with said adjustments of the manufacturing process parameters.

15. The real-time modelling and evaluation method according to Claim or 7, characterized in that the evaluated unit parameters are adjusted, resulting in the quantification of the quality score of any digital ophthalmic lens manufacturing process unit, or "unit quality score", with said adjustment of the unit parameters.