FR2975688A1 - Cutting multiple pieces of glass in glass sheets, comprises reading information relating to defects in sheets, automatically providing optimum cutting plane for glass sheet, and cutting pieces of glass respecting optimum cutting plane - Google Patents

Abstract

The method comprises reading information relating to defects in glass sheets (8), automatically providing an optimum cutting plane for each of the glass sheet according to some of the information relating to faults, and cutting pieces of glass respecting the optimum cutting plane. The optimum cutting plane is provided according to acceptance criteria of defects that are different defects within predetermined areas (Z0, Z1, Z2) of the pieces to be cut with respect to another predetermined area of the pieces to be cut. The method comprises reading information relating to defects in glass sheets (8), automatically providing an optimum cutting plane for each of the glass sheet according to some of the information relating to faults, and cutting pieces of glass respecting the optimum cutting plane. The optimum cutting plane is provided according to acceptance criteria of defects that are different defects within predetermined areas (Z0, Z1, Z2) of the pieces to be cut with respect to another predetermined area of the pieces to be cut. One of the predetermined areas is included in the other of the predetermined areas of the same piece of glass to be cut. Three different acceptance criterias correspond to three respective zones of the same piece to be cut. The step of providing the optimum cutting plane comprises automatically generating the optimum cutting plane for each of the glass sheet according to some of the information relating to faults, where the automatic generation step is obtained by dynamic calculation. The dynamic calculation maximizes or minimizes an objective function of several variables that are subjected to constraints. The variables and/or constraints respectively include variables and/or constraints representing the acceptance criteria according to some of the information about defects. The variables include variables that represent a percentage of cutting one of the pieces in the set of glass sheets. The information on defects includes a position and/or size and/or type of defects. The method further comprises: analyzing the defects in the glass sheet; and storing information relating to defects detected in the glass sheet, where the storing step is performed by ink marking on the defects of the glass sheet or in an electronic memory. The reading step comprises reading an ink marked on the defects of the glass or reading an electronic memory containing the information.

Description

-1- METHOD FOR CUTTING ONE OR MORE WINDOWS

The present invention relates to the field of cutting glass pieces in large glass sheets. The glass is generally manufactured in the form of a continuous ribbon, for example a continuous ribbon of float glass or cast glass. This ribbon is then cut into sheets of glass called "motherglass" (literally "mother glass" in French even if this term is not used); Which sheets are for example "PLF" (Large Format Glass Trays), typically of dimensions 3.21m by about 6m or "DLF" of dimensions about 2.55m by 3.21m. A defect analysis step is performed prior to this cutting to check whether the glass ribbon corresponds to defect specifications. If there are out-of-specification defaults, the motherglass are cut out by excluding a certain length of the ribbon corresponding to the off-specification portion of the ribbon. Alternatively, the defects are for example marked with an ink so as to identify them later without a new analysis. After cutting, the motherglass can then be stacked in different stacks according to the defect specification classes. The motherglass can then undergo one or more transformation processes (for example depositing a layer, laminating, ...). After each transformation, the motherglass are for example analyzed for possible defaults and thus check whether the quality corresponds to a predetermined specification. Otherwise, the motherglass is rejected. US-A-2004/0134231 discloses a method of cutting glass substrates for LCD screens in motherglass. The motherglass are identified and information about the defects of each motherglass such as the position, size or type of defects are stored so as to optimize the cutting of LCD substrates of different sizes according to the fault information of each motherglass. . Various predetermined cutting planes are for example combined with different motherboards and with different acceptance criteria so as to maximize the number of LCD substrates that can be cut in a set of several motherglass.

An object of the invention is to provide a method for further reducing losses due to glass defects. According to one aspect of the invention, it is a method of cutting several pieces of glass in at least one glass sheet, comprising a step of reading information relating to defects of said at least one sheet of glass. glass, in which the method automatically provides an optimum cutting plane for each of said at least one glass sheet in accordance with at least some of the fault information, the optimum cutting plane being provided according to criteria of acceptance of defects, the criteria for accepting defects being different within a predetermined zone of one, several or each of the pieces to be cut with respect to another predetermined zone of the same piece to be cut, the method comprising a step of cutting the pieces of glass respecting the optimum cutting plane provided.

Note that throughout the text, the term "automatic", an action performed by a machine running a recorded program. The term "dynamic generation" or "generation by dynamic calculation" means a construction of the cutting plane that is determined as the program is executed.

It should also be noted that by "cutting a sheet of glass" is meant cutting of a sheet of bare glass or on which a coating has been deposited. In addition, a glass sheet is not necessarily flat, although it is usually flat when cutting. This method has the advantage of making it possible to further optimize the process for cutting the pieces of glass in a large glass sheet or in a group of several glass sheets. In fact, taking into account information relating to the defect, in particular their position, their type and their size, makes it possible to discriminate the defects to be rejected or accepted according to the area of the piece to be cut in which the defect occurs. find. According to particular embodiments, the method has one or more of the following characteristics, taken in isolation or in any technically possible combination: one of said predetermined zones is included in the other of said predetermined zones of same piece of chopping glass; there are at least three different acceptance criteria corresponding to at least three respective zones of the same piece to be cut; Three of said predetermined areas are included one inside the other; the step of providing the optimum cutting plane comprises a step of automatically generating the optimum cutting plane for each of said at least one glass sheet according to at least some of the information relating to the defects, the automatic generation of the plane optimum cutting being obtained by a dynamic calculation; the calculation is iterative, the calculation being for example iterated from an initial cutting plane, for example predetermined; the dynamic calculation maximizes or minimizes an objective function of several variables, the variables being subject to constraints; The variables and / or the constraints respectively include variables and / or constraints representative of fault acceptance criteria as a function of at least some of the information relating to the defects, the criteria for accepting defects being, for example, besides different for different pieces of glass to be cut; The objective function provides a value representative of the number of pieces of glass to be cut including at least one non-acceptable defect and / or a sum of one or more dimensions of these pieces of glass and / or a sum of costs of rejecting these pieces of glass; the objective function is linear; The variables include variables representative of spatial coordinates of the pieces to be cut; some of the pieces to be cut have different dimensions, the variables including variables representative of one or more dimensions of at least some of the pieces to be cut, for example the width and / or the length in the case of a rectangle; the variables include variables representative of one or more angles of at least some of the pieces of glass to be cut with respect to one or more references; said at least one glass sheet comprises a plurality of glass sheets, the variables including for example at least one variable representative of a percentage of cutting of at least one of the pieces in the group of glass sheets; at least some of the variables can only take a finite number of values, for example all the variables; the constraints include at least one positioning constraint for the pieces of glass preventing the pieces of glass from overlapping with each other; the constraints include at least one positioning stress of the pieces of glass inside at least one of the glass sheet or sheets; the constraints are linear equations, representative of a convex polyhedron; the step of providing the optimum cutting plane comprises a simulation step for testing different predetermined cutting planes, for example on a group of several glass sheets and / or for example with different acceptance criteria for different pieces of glass. glass; the method comprises: - a step of analyzing defects in said at least one glass sheet; a step of memorizing information relating to the defects detected in said at least one glass sheet, the storage being for example carried out in particular by ink marking on the defects of said at least one sheet of glass or by storage in a electronic memory, the step of reading information including for example a step of reading a marked ink on glass defects or a reading step of an electronic memory containing said information. The storage step includes a step of storing said information in one or more electronic memories; - the information is accessible via the Internet or a local network; the storage step includes a step of marking said information on the corresponding glass sheet, for example by an ink marking the flaw or faults detected directly on the flaw or defects; the method comprises several fault analysis steps, the analysis steps being for example alternated with storage steps of the detected defects; the method comprises a step of identifying the at least one glass sheet; the identification step includes the inscription of an identification code on the corresponding glass sheet, for example of the barcode type, and / or the reading of this code; the information relating to the defects includes a position and / or a size and / or a type of the defects; the calculation is carried out by one or more electronic computers; - The glass sheet or sheets are cut in a continuous ribbon of glass; - The glass sheet or sheets are cut in a continuous ribbon of glass without rejecting a portion of the glass ribbon between two consecutive glass sheets cut in the ribbon; - The pieces of glass to be cut in the at least one glass sheet are able to form at least a portion of a glazing, including a building glazing, glazing for solar application, for example photovoltaic, glazing for OLED application , a mirror or automotive glazing; The invention will be better understood on reading the description which follows, given solely by way of example, and with reference to the appended drawings, in which: FIG. 1 is a diagram illustrating schematically an example of a method for manufacturing building glazing, glazing for solar applications, for example photovoltaic, OLED glazing, mirrors or automotive glazing, illustrating the main steps and an example of a supply chain; Figure 2 schematically shows an example of motherglass for which different defects have been listed; FIG. 3 illustrates a possible implementation of the positioning of a piece to be cut (called "primitive") with a view to an optimization calculation; FIG. 3bis illustrates other possible forms of pieces to be cut; FIG. 4 diagrammatically shows an example of a cutting plane in the motherboard of FIG. 2, the cutting plane being generated by a computer according to the information relating to the defects and as a function of the criteria for acceptance of the defects. Figure 1 is a non-limiting example of a manufacturing process to which the various aspects of the invention can be applied. In this example, the upper part of the diagram concerns the steps of manufacturing a motherglass at a glass manufacturer, and the second part the steps of manufacturing an application glass such as a glass for automotive glazing, glazing for solar application, for example photovoltaic, glazing for OLED application, mirror or building glazing at a second manufacturer, client of the first. All of the steps may alternatively be performed by the same manufacturer or the division of labor be of any suitable type. In this particular example, the first manufacturer produces in a plant 2 called "float glass", a continuous ribbon 4 of float glass on a bath of tin. Defects of the ribbon 4 are analyzed by a detection device 6 (of any suitable type), then the ribbon 2 cut into motherglass 8. Note that the detection device is for example a device called "scanner" in industry and for analyzing the glass for defects. Conventionally, any areas of the ribbon with defects considered unacceptable are for example excluded when cutting motherglass. We will nevertheless see below that rejecting ribbon areas between motherglass is not necessary with the invention. Information concerning the defects relating to each motherglass is stored in a database 10. For this purpose, the defects are marked with an identifier 12, for example a barcode, an RFID chip or an other means. of any suitable type. The marking of the identifier is for example made in ink or laser. The stored fault information includes, for example, the position, the size and the type of the faults detected by the detection device 6. In a variant, the faults are not memorized in this way, that is to say by inscription in an electronic memory. They are for example marked by an ink on the defects of the glass, which ink will for example be read by a camera. The term "memorize" must thus be understood in a broad sense, throughout the text, the marking of defects by an ink being considered as a storage of information relating to defects, which information is written on the glass. The motherglass are, for example, then stacked 14 and stored 16 before being conveyed for a transformation process 18, for example for the deposition of a coating by a "coater", typically at least one conductive or dielectric coating, transparent or reflective, and having thicknesses of a few tens or hundreds of nanometers thick, or for example for a lamination process or a mirror formation. After treatment, the motherglass are for example analyzed by a second detection device 20, in particular for the purpose of detecting defects in the treatment or treatments carried out. The detection device 20 is for example a "scanner" as mentioned above. The detection device 20 is able to identify the motherglass 8, for example by means of a barcode reader. It is furthermore for example connected to the database 10 so as to be able to use the stored fault information for each motherglass, for example for a more detailed inspection of the areas with known defects, and so that the new ones can be memorized. defect information generated by the detection device 20 for each motherglass 8 analyzed. In the case where ink tasks have previously been written on the defects, the detection device 20 comprises for example in addition to or in replacement of the "scanner", a camera detecting the position of the tasks at the entrance of the line of transformation. Database 10 is optional. It may be an alternative of a removable memory medium and read by the detection device 20 or a tool connected to the detection device 20. The motherglass 8 are again stacked 22 and stored 24, for example on the basis of stored fault information before being routed to a customer. The customer is the one who will cut the motherglass into pieces of glass, typically several sheets of glass with identical dimensions. Note that alternatively, it is not a customer but the first manufacturer itself, for example an internal transformer. The customer has a calculator tool 28 in which stored programs are able to provide an optimum cutting plane for example on groups of several motherglass or on one, in order to minimize the amount of glass to be rejected. For this purpose, the customer has for example an identifier reader to identify each motherglass 8 and for example has access to the database 10, which is for example connected to the calculator tool 28 by an information system such as the Internet. The information is for example filtered by a filter 30, so that only the information useful to the customer is accessible or so that this information is accessible in a compatible format. As a variant, especially in the case where ink stains have previously been written on the defects of the glass, the customer is, for example, equipped with a camera detecting the position and / or the color and / or the size of the tasks at the same time. input of the cutting line and transmitting information to the computer 28. The programs of the computer will be described in more detail below, with regard to essential aspects of the invention. Once, the generation of an optimum cutting plane achieved, the motherglass are cut 32 according to the cutting planes that the computer 28 has selected for each motherglass 8.

Note that in the alternative the cutting plane is not generated by the computer but simulated from predetermined cutting planes. In general, the calculator provides an optimum cutting plane.

As illustrated, the cut pieces are for example washed 34 before being optionally analyzed by a third detection device 36 and then for example assembled into a multiple building glazing or automotive glazing. In the case for example of a windshield of a motor vehicle, two pieces of glass will be curved and laminated together via a thermoplastic interlayer for example PVB type. The various aspects of the invention concerning the provision of an optimum cutting plane, in particular by automatic and dynamic generation or by simulation, will be described in more detail below. The term "dynamic generation" or "generation by dynamic calculation" is understood to mean a construction of the cutting plane, which is determined as and when the program is executed. In a second part, we will mention possible generalizations of the example of Figure 1 to other manufacturing processes. As explained above, the invention more particularly relates to a division of the pieces to be cut into different areas corresponding to different criteria for accepting defects. According to a first aspect of the invention, it is in fact to automatically provide an optimum cutting plane for each of the glass sheets according to the information relating to the defects, the optimum cutting plane being provided as a function of acceptance criteria for different defects for different areas of the pieces to be cut. According to another aspect of the invention, it is a question of automatically and dynamically generating an optimum cutting plane for each of the glass sheets according to the information relating to the defects, the optimum being obtained by a dynamic calculation, for example by a linear optimization. An example of a dynamic generation method will be described below. 2975688 - 10 - We will explain later the possibility of providing an optimum cutting plan by simulation, but this is less advantageous. FIG. 2 illustrates an example of motherglass for which various defects have been listed, namely, a pinhole type defect (literally "pinhole" in French) on the coating, a bubble type defect 38, a scratch-type defect 40 on the glass, and a defect 42 of the surface defect type. To explain the generation of an optimum cutting plane, let us begin with the simplest example, namely the dynamic generation of an optimum cutting plane in a single sheet of glass with pieces of cut glass of identical size, defects of a single type and size that are not accepted in chopped (or "primitive") pieces of glass. In this example, different fault acceptance areas in the glass pieces are ignored. We will see later how to take it into account. This example is explained with reference to FIG. 3. The dynamic generation is, in this example, carried out by a linear optimization, that is to say by the iterative resolution of a problem of optimization of a linear function. on a convex polyhedron representing constraints on the variables, the constraints being linear equations. In a variant, it is an optimization program of any suitable type. The advantage of linear programming is its speed of calculation. The objective chosen is to minimize a function representative of the number of primitives including at least one defect. We will see below how the value of this function can be calculated. Alternatively, the function provides a value representative of the number of pieces of glass cut in the generated cutting plane and / or a sum of one or more dimensions of the cut pieces of glass such as the total area of the pieces of glass. cut and / or a sum of the cost of cut pieces of glass. In general, it is a performance indicator of the cutting plane of any suitable type. In this example, the pieces to be cut, also called "primitives" in industry, are rectangles (see Figure 3). In general, it is a polygon or even more generally a closed figure (i.e. the edges are not necessarily rectilinear, see Figure 3bis). The various aspects of the invention can of course be applied to the cutting of glass pieces forming automotive glazings, which typically have non-rectilinear contours. Note that the image is for example pixelated and a polygon, whether it is a rectangle, a parallelogram or other, is then a combination of pixels. For each primitive, here being rectangles, two variables and two parameters are used here to define its position relative to the motherglass. Indeed, the rectangles have in this example always the same orientation, that is to say, an orientation with the length parallel to the length of the motherglass. As illustrated in FIG. 3, the abscissa coordinates x;,; n; and of ordinates y;, n; for example from the lower left corner of each primitive i are for example chosen as variables to represent the position of each rectangle. Alternatively, it is another point of the primitive or other types of coordinates. In another variant, the variables indicate an angle of the primitive with respect to a reference, so as to be able to move the primitive in rotation during the optimization.

In general, they are variables giving an indication of positioning of the primitive compared to the motherglass to be cut. The two parameters (constant by definition) chosen here are the length and the width of the rectangle, which make it possible to calculate, from the coordinates of the lower left corner of the piece to be cut, the ordinate y, the end of the upper left corner and the x abscissa, fn of the lower right corner. As a variant, these are parameters of any type adapted to indicate the dimensions or the orientation of the primitive. An intersection constraint of two primitives is introduced. In this example, the constraint "Intersection (i, j)" of two primitives is equal to 1 if two primitives overlap each other and equal 0 otherwise. This constraint must of course be equal to 0. These values are for example stored in a matrix n X n, where n is an integer corresponding to the number of primitives which one wishes ideally to be able to cut in the sheet. Intersection (i, i) is of course not considered. In this example, Intersection () actually contains 4 constraints, namely xi, ini xj, end Yi, ini Yj, end xi, ini xi, end Yj, ini Yi, end At least one of these four constraints must be checked so that the constraint Intersection (i, j) is equal to 0. Finally, the value of the function is calculated by creating a matrix of n rows and m columns, m being an integer corresponding to the number of defects.

Each defect is defined by a rectangle whose positioning is defined for example in the same way as the primitives, namely with xi, ini, yi, ini, xj, end and Yi, end. In the same way as for the primitives, is more generally a closed figure of any suitable type, for example a polygon.

A function Fault (i, j) = 1 in case of intersection of the primitive rectangle i with the default rectangle j and equal to 0 otherwise by checking at least one of the four inequalities mentioned above for the constraint Intersection(). Unlike Intersection (i, j), Default (i, j) is not a constraint but a value used to calculate the objective function to maximize. The calculator then calculates 1 Default (i, j) for each primitive i. An array of size n is created with IsGood (i) values. IsGood (i) = 0 if 1 Default (i, j)> _ 1, and J IsGood (i) = 1 if EDefault (i, j) = O.

The objective function = EIsGood (i), which must be maximized.

To implement this program, a linear solver using a simplex algorithm is for example used. Initially an initial cutting plane was saved in memory. The iterations are performed from this initial cutting plane, for which the function to be optimized is calculated during a first initialization step. Several generalizations of this program will be explained below. The first generalization concerns the application of this program to the first aspect of the invention. As illustrated in FIG. 4, the various areas of acceptance of defects are, for example, rectangles included in each other inside the piece to be cut. The positioning of each zone inside the primitive (z1 and z2 in FIG. 4) is for example defined by four parameters, namely for example the relative coordinates of its lower left corner relative to the lower left corner of the primitive, its length and width. These four parameters make it possible to calculate, from the coordinates of the lower left corner of the primitive, the coordinates of abscissa x;, zi, n; (for the zone z1) and ordinates the ordinate yi, zl, end of the upper left corner and the abscissa x;, zi, fn of the lower right corner. It is the same for the zone z2 inside zone z1. In addition, the number of zones in a primitive is for example an additional parameter of the primitive.

To determine if a fault is in at least one of the zones, the "Fault" function described above can be adapted as follows. Criteria for accepting defects for the different zones are for example defined as additional parameters of each zone. In addition, the defects are for example assigned parameters such as their size or type (bubble, scratch ...) to accept them differently in each zone. This is however necessary in the simplest case where each zone accepts either all the defects taken into account, or none. For example, there will be for example a PositionFunction function with, for example for zone z1: 2975688 -14- PositionFault (i, z1, j) = 1 in case of intersection of the area rectangle z1 with the default rectangle j and equal to 0 in the opposite case by checking at least one of four inequalities similar to those mentioned above for the intersection of the primitives. This function checks for the presence of the fault in the

5 zone.

If PositionFault (i, z1, j) = 1, it is checked whether the acceptance criteria of the zone z1 are compatible with this defect, we then have for example ZeroFault (i, zl, j) = 0 in case of compatibility , and DefaultZone (i, zl, j) = 1 otherwise.

This is done for each zone z1, z2, ... inside the primitive and for the rectangle of the primitive, which corresponds to the zone "zO".

Then we will have Default (i, j) = 1 if EDefautZone (i, z, j)> 1 z

(ie ZoneFault (i, zO, j) + ZoneFault (i, zl, j) + ZoneFault (i, z2, j) + ...? 1), 15 and Fault (i, j) = 0 if 1 ZoneFault ( i, z, j) = 0. The program then proceeds in the same manner as described above for calculating the objective function.

To discriminate on the size or the type of defect, one will proceed for example to the computation, in the case where DefaultPosition (i, zl, j) = 1, of DefaultType () and DefaultSize () with for example:

DefaultType (i, zl, j) = 1 if the type is not accepted for zone z1 and 0 otherwise, and

FaultSize (i, zl, j) = 1 if the type is not accepted for zone z1 and 0

25 otherwise. More specifically, it is also possible to

check the size only for the part of the fault inside zone z1. Then, if DefaultType (i, zl, j) = 1 or DefaultSize (i, zl, j) = 1 then

DefaultZone (i, zl, j) = 1, and

DefaultZone (i, zl, j) = 0 otherwise.

The program then proceeds in the same manner as described above for calculating the objective function. Another possibility to extend the above example is to consider primitives of different sizes and / or with different orientations. One way of doing, for primitives rectangles, is to consider as variables, in addition to the coordinates (x;, n ;, y;, n;) of the lower left corner, the length 5 and the width to determine the size , and an orientation angle of the rectangle to determine the orientation. It is also possible to generate an optimum cutting plane by considering a possible positioning of different primitives on different sheets of glass. The glass sheets are then for example considered contiguous and forming a single sheet of glass. The overlap of the primitives with the junctions between sheets are for example prohibited considering the intersection with these junctions as a prohibited constraint. This is for example of interest in the case of primitives of different sizes, so as to meet an ideal specification of distribution of these 15 different types of primitives. Compliance with the specifications is, for example, integrated into the objective function or considered as a constraint. It should be further noted that linear programming is only one possibility among others of generating an optimum cutting plane by a dynamic calculation, as well as how to pose the problem to be solved and to solve it. In general, it is an automatic optimization method using a dynamic calculation. This is for example a dynamic calculation that maximizes or minimizes a function of several variables, the variables being subject to constraints. The function may not be linear, as well as the equations induced by the constraints. In addition, it is not necessarily an optimization, but for example a simulation step for testing different predetermined cutting planes, for example on a group of several glass sheets and / or for example with criteria for cutting. different acceptance for different pieces of glass. A simulation method of the type described in US-A-2004/0134231 will then be used for example but it is a variant of a suitable type of process. In addition, as explained above, the various aspects of the invention can be applied to many glassmaking processes. The example of FIG. 1 can be generalized to manufacturing processes of any suitable type.

First, the number of fault analysis steps is of any suitable type. An advantage of the identification of the motherglass 8 or the ink marking of the defects is to enable these different analyzes to be carried out independently, each detection device then being for example provided with one or more readers for identifying the motherglass and connected to the database 10. With regard to the identifier, especially in the case of a marking of the identifier, it will advantageously be a marking on the edge of the motherglass, so as to easily read these last once stacked. Rather than identifying each motherglass and having a database of fault information storage, it is possible, alternatively, as mentioned above, to mark the defects with an ink of such or such color and / or size. on the defect itself. The client is then able to identify the different types of faults, their size and their position and can, for example by automatic readers, for example cameras, generate itself defect information useful to the optimization program. cutting plans. It should also be noted that the relative steps concerning the transformation of the glass sheets are optional since some sheets are not treated before cutting.

Another aspect of the method of Figure 1 relates to the location and timing of the optimization. In the method of Figure 1, the optimization of the cut is performed at the customer, that is to say in the one who cuts. Nevertheless, this optimization can of course be carried out at the motherglass manufacturer, insofar as the information concerning the pieces of glass to be cut and the criteria for accepting defects are known to him. This optimization at the manufacturer of motherglass will be all the more advantageous that it will allow him to perform cutting optimizations on more large motherglass for example by grouping the motherglass for different customers. In this way, instead of sending motherglass judged to comply with specifications to a particular customer, without taking into account an optimization of the cutting at the customers, the distribution of the motherglass to the different customers can be divided into the results of the optimization, avoiding in this way to send to a customer a motherglass that will not be optimal while this motherglass would have been more optimal to cut at another customer. The manufacturer of motherglass can of course also make a first cut of a motherglass for example to send a part to a first client and the other party to a second client, customers performing a second cut in these pieces.

Claims (10)

REVENDICATIONS1. A method of cutting a plurality of pieces of glass in at least one glass sheet (8), comprising a step of reading information relating to defects of said at least one glass sheet (8), wherein the method provides such automatically an optimum cutting plane for each of said at least one glass sheet (8) according to at least some of the fault information, the optimum cutting plane being provided according to fault acceptance criteria, the acceptance criteria for defects being different within a predetermined zone (zO, z1, z2) of one, several or each of the pieces to be cut with respect to another predetermined zone (zO, z1, z2) of the same piece to be cut, the method comprising a step of cutting the pieces of glass respecting the optimum cutting plane provided. The method of claim 1, wherein one of said predetermined areas (z0, z1, z2) is included in the other of said predetermined areas (z0, z1, z2) of the same piece of glass to be cut. 3. Method according to claim 1 or 2, wherein there are at least three different acceptance criteria corresponding to at least three respective zones (zO, z1, z2) of the same piece to be cut. The method of claim 3, wherein three of said predetermined areas (z0, z1, z2) are included in each other. 5. Method according to any one of the preceding claims, wherein the step of providing the optimum cutting plane comprises a step of automatically generating the optimum cutting plane for each of said at least one glass sheet (8) according to at least some of the information relating to the defects, the automatic generation of the optimum cutting plane being obtained by a dynamic calculation. 6. The method of claim 5, wherein the dynamic calculation maximizes or minimizes an objective function of several variables, the variables being subject to constraints. The method of claim 5 or 6, wherein the variables and / or constraints respectively include variables and / or constraints representative of fault acceptance criteria as a function of at least some of the related information. to defects. 8. A method according to any one of claims 5 to 7, wherein said at least one glass sheet comprises a plurality of glass sheets, the variables including for example at least one variable representative of a cutting percentage of at least one of the pieces in the group of glass sheets. The method of any of the preceding claims, wherein the fault information includes a position and / or a size and / or type of the defects. The method of any of the preceding claims, comprising: - a step of analyzing defects in said at least one glass sheet; A step of storing information relating to the defects detected in said at least one glass sheet, the storing being for example carried out in particular by ink marking on the defects of said at least one sheet of glass or by storage in an electronic memory, the step of reading the information including for example a step of reading an ink marked on the defects of the glass or a reading step of an electronic memory containing said information.