Method for the generation of a sequence of layers for a glass surface

Abstract

Method (1) for the generation of a sequence of layers for a glass surface, said sequence of layers comprising, according to a layering position, layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields or a combination thereof each having its own thickness, comprises the steps of a) providing a machine learning generative artificial intelligence algorithm based on a conversational model, said artificial intelligence algorithm being trained to generate a sequence of layers for a glass surface necessary to make a glass surface by selecting one or more layers, a relative thickness for each selected layer and a layering position; b) providing by input on a user device (10) one or more initial statements relating to a glass surface design, said initial statements comprising performance parameters required for the glass surface selected from optical, energetic, static, acoustic, usage parameters or a combination thereof; c) processing the statements provided by means of said artificial intelligence algorithm so as to generate said sequence of layers, said sequence of layers being representative of the correspondence between the statements provided and the training of the artificial intelligence algorithm, said artificial intelligence algorithm being resident in a processing unit (20) in signal communication with the user device (10), d) generating by means of said artificial intelligence algorithm a final sequence of layers representative of the correspondence between the statements provided and the training of the artificial intelligence algorithm.

Description

[0001] Title: “Method for the generation of a sequence of layers for a glass surface”

[0002] DESCRIPTION

[0003] Technical Field

[0004] The present invention relates to a method and the relative system for the generation of a sequence of layers for a glass surface.

[0005] In particular, the present invention relates to the generation of a sequence of layers for a glass surface by means of artificial intelligence algorithms that take into account performance requirements and are capable of providing the sequence of layers, its relative position and the thickness of each layer as a function of the selected position.

[0006] State of the Art

[0007] It is known for the design of a glass surface to take into consideration various parameters such as optical, energetic, static and acoustic parameters.

[0008] The known methods for designing a glass surface take into consideration the required performance requirements on the basis of the aforesaid parameters and by trial and error a designer selects the layers to define the glass surface.

[0009] Specifically, the design of the glass configuration consists in deciding the sequence of layers as type and thicknesses that constitute it.The layers can be: glasses, coatings, plastic interlayers, gaps with air / argon, opaque elements and shields.

[0010] There are also various software programs that perform calculations according to regulations, but the designer must identify and try different configurations to find the right solution.

[0011] There are also databases with pre-calculated configurations that the designer can query using specific values or search ranges (e.g. light transmission between 40% and 60%, thermal transmittance lower than 2.0).

[0012] There are also software programs that use a neural network to qualitatively estimate the acoustic attenuation of a proposed sequence of layers as the acoustic attenuation cannot be calculated, but only measured in a laboratory.

[0013] Some software programs are also known that allow for providing qualitative advice on glass and curtain configurations.

[0014] Problems of the State of art

[0015] Disadvantageously, the sequence of layers is case sensitive.

[0016] In fact, the same layers in a different sequence generate a different result.Therefore, the designer, window / door frame maker or glazier, starts from the performance requirements and describes and calculates different sequences and by iteration finds a solution that meets the needs.

[0017] Each layer has thermal and spectral characteristics and the sum / interaction of these characteristics is carried out by means of a calculation according to regulations. Such operations are time-consuming and sometimes only approach the required performance requirements.

[0018] Disadvantageously, even the most technological solutions take into account only part of the required parameters and still require an iterative operation by the designer who, based on his own experience, selects the most suitable sequence of layers and proposes it within the software.

[0019] Disadvantageously, solutions that use artificial intelligence provide only qualitative and not quantitative evaluations, requiring the constant intervention of the designer for the correct evaluation of the sequence of layers and the subsequent calculation of the performance.

[0020] Object of the Invention

[0021] The object of the invention in question is to provide a method for the generation of a sequence of layers for a glass surface that is able to overcome the drawbacks of the known art cited above.In particular, it is an object of the present invention to provide a method for the generation of a sequence of layers for a glass surface that responds to the required performance requirements without subsequent calculation tests by the designer and which at the same time reduces the time necessary for the relative generation of the sequence.

[0022] The specified technical task and the specified objects are substantially achieved by a method for the generation of a sequence of layers for a glass surface, said sequence of layers comprising, according to a layering position, layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields or a combination thereof each having its own thickness, the method comprising the technical features set forth in one or more of the appended claims.

[0023] Advantages of the invention

[0024] According to the present invention, the claimed method allows to save time on the selection of the sequence of layers for the glass surface while also taking into account different glass configurations, glass models and brands.

[0025] According to the present invention, the claimed method allows to select the sequence of layers starting from the required performance requirements.According to the present invention, the claimed method allows to speed up the iterative process. In fact, artificial intelligence algorithms perform many iterations and calculations with lower precision but greater speed, so it is possible to evaluate many potentially interesting sequences of layers to derive the best one, and only then perform a precise calculation.

[0026] According to the present invention, the claimed method allows to optimize the power and calculation time by performing performance calculations only on the selected sequence.

[0027] BRIEF DESCRIPTION OF THE FIGURES

[0028] Further characteristics and advantages of the present invention will become more apparent from the indicative and therefore non-limiting description of a preferred but not exclusive embodiment of a method for the generation of a sequence of layers for a glass surface as illustrated in the accompanying drawings:

[0029] - figure 1 shows a block diagram of a method for the generation of a sequence of layers for a glass surface in accordance with an embodiment of the present invention;

[0030] - figure 2 schematically shows a system for implementing the method for the generation of a sequence of layers for a glass surface in accordance with an embodiment of the present invention.

[0031] DETAILED DESCRIPTIONEven if not explicitly highlighted, the individual features described with reference to the specific embodiments should be understood as accessory and / or interchangeable with other features, described with reference to other examples of embodiment.

[0032] The present invention relates to a method for the generation of a sequence of layers for a glass surface.

[0033] In particular, the sequence of layers generated by the method comprises, according to a layering position, layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields or a combination thereof each having its own thickness. It should be noted that the shields can be for example curtains with their relative functionalities and components. According to one aspect, opaque elements and shields are for example useful for the blind panels of the glazings that cover the floor slab, or also partially opaque.

[0034] Said sequence of generated layers has the purpose of responding to required performance parameters selected from optical, energetic, static, acoustic, usage parameters or a combination thereof.

[0035] Specifically, the performance parameters have the following characteristics:

[0036] - Optical: Luminous Transmittance, Luminous Reflectance, color, transmission and reflection spectrum from 300nm to 2500nm, etc.

[0037] - Energetic: Solar Factor, Thermal Transmittance, Solar Transmission, Temperatures, etc.- Static (deflections and stress of the glass): glass tensions, allowable tensions, deflections,

[0038] - Acoustic: Decibels of acoustic attenuation and its corrective factors based on the type of noise;

[0039] - Usage: how the glass surface is used.

[0040] It should be noted that the generated sequence of layers comprises:

[0041] i) a layering position for each layer, i.e. the position of the layer between a surface A and a surface B;

[0042] ii) a layer selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields;

[0043] iii) a thickness for each layer as a function of the selected layer of the relative layering position.

[0044] Specifically, the method of the present invention provides for the use of a machine learning generative artificial intelligence algorithm based on a conversational model trained by means of at least the product technical sheets and the entire company knowhow for the generation of the aforesaid sequence of layers as a function of the performance parameters.

[0045] In particular, the technical sheets and the know-how are formalized in HTML-type web pages, text and image files and design and exchange files such as for example XML or JSON.Said technical sheets and the know-how therefore represent the training sources of the artificial intelligence algorithm and said sources can be public and / or non-public (i.e. private, that is with conditional access by means of, for example, a password or similar tools).

[0046] The artificial intelligence algorithm, as mentioned, can be continuously trained through the experience that the company continues to develop and acquire in the realization, assembly, experimentation and calculation of sequences of layers.

[0047] Specifically, the artificial intelligence algorithm is intended to be trained with sequences of layers associated with relative performance parameters of the glass surface obtained with such a sequence.

[0048] It should be noted that the method of the present invention, through the use of the artificial intelligence algorithm, allows to automate the generation of the sequence of layers as it is trained to extrapolate what is necessary from the information provided on the glass surface project thanks to the relative technical training received and continuously updated.

[0049] Said use of the trained artificial intelligence algorithm reduces the time necessary to obtain the sequence of layers, automates the relative generation of the sequence of layers by reducing the number of necessary personnel, exploits the entire company knowledge in continuous updating and reduces or eliminates errors on the sequence of layers.It should be noted that the method of the present invention allows the interested user to interact with only the user device, described below, to obtain what is requested, improving the relative comfort and reducing response times.

[0050] The method of the present invention, also with reference to figure 2, is implemented by means of a computer system 100.

[0051] Said computer system 100 comprises a user device 10 with access to the web, i.e. to the Internet.

[0052] Said user device 10 can be a mobile device (tablet, smart phone or similar), a personal computer or a terminal. Said user device 10 is known to the person skilled in the art and not further described.

[0053] The computer system 100 comprises a processing unit 20 configured to process the received data which is in signal communication with the user device 10.

[0054] Preferably, the artificial intelligence algorithm is resident in the data processing unit 20.

[0055] To this end, the processing unit 20 may comprise one or more processors 21 and relative servers 22 suitable for executing the artificial intelligence algorithm andconfigured to store the information relating to the training of the algorithm as well as the algorithm itself.

[0056] It is noted that the processing unit 20 can be connected to the Internet and therefore to the relative public web pages.

[0057] The computer system 100 can also comprise a training device / program 30 in signal communication with the data processing unit 20 and configured to train the algorithm by loading HTML pages, text and image files, for example, photographs etc. and design and exchange files such as for example XML or JSON.

[0058] Said training device 30 is a device for inputting information and instructions to train the artificial intelligence algorithm.

[0059] Preferably, the computer system 100 comprises a database 40 in signal communication with the processing unit 20 through a link suitable to support an Internet communication protocol.

[0060] The computer system 100 further comprises a machine learning generative artificial intelligence algorithm based on a conversational model 1 resident in said processing unit 20.

[0061] It should be noted that said artificial intelligence algorithm is interrogable through the user device 10.The system 100 can also comprise a terminal 50 in signal communication with the database 40 for accessing theinformation stored therein such as the final sequences.

[0062] In accordance with a preferred embodiment, the computer system 100 further comprises a calculation algorithm for a performance estimate, preferably said algorithm allows to perform calculations according to regulations. The calculation algorithm is resident in the processing unit 20 and executed by means of one or more processors 21 and relative servers 22. It should be noted that the calculation algorithm allows in turn to train the artificial intelligence algorithm as a function of the comparison between the required performance requirements and the calculated estimate.

[0063] The method object of the present invention comprises the steps reported below, executed according to a preferred embodiment and shown in figure 1.

[0064] The method comprises a preliminary step of accessing a web page by the user through the user device 10 in order to input statements relating to a glass surface project.

[0065] According to one aspect, said project, as clarified below, provides for providing information relating to the performance parameters selected from optical, energetic, static, acoustic, usage parameters or a combination thereof.Preferably, the project can also specify the type of environment in which the glass surface is to be installed in order to identify the mode of use (for example children's room, bedroom, living room, offices etc.) and the dimensions of the environment of interest.

[0066] It should be noted that it is precisely from the performance parameters provided by the user to the artificial intelligence algorithm 1 that said artificial intelligence algorithm is able to understand the type of project to be realized and, on the basis of the training performed (by meansof the aforesaid public and / or non-public sources), it is always said algorithm that is able to generate the sequence of layers necessary for the realization of the project itself.

[0067] The method comprises a step a) of providing a machine learning generative artificial intelligence algorithm based on a conversational model 1. For example, the machine learning generative artificial intelligence algorithm based on a conversational model 1 is based on known AIs such as ChatGPT®, Gemini® or Copilot® or LLAMA® or similar artificial intelligence programs.

[0068] Said artificial intelligence algorithm 1 is trained by means of public and / or non-public sources which can be of the written and / or photographic type found in at least HTML-type pages, and / or text and image files, design and exchange files such as for example XML or JSON.Preferably, said public and / or non-public sources are provided to the processing unit 20 by means of the training device 30.

[0069] According to one aspect, the artificial intelligence algorithm is trained to generate a sequence of layers for a glass surface necessary to make a glass surface by selecting one or more layers, a relative thickness for each selected layer and a layering position.

[0070] In accordance with a preferred embodiment, step a) of the method comprises a step al) of preliminarily training the artificial intelligence algorithm by providing nonpublic sources (or information) such as HTML-type pages, text and image files and design and exchange files such as, for example, XML or JSON having as their object technical information relating to the design of a glass surface.

[0071] Specifically, said technical information provided comprises technical sheets, calculations, sequences of layers for glass surfaces associated with relative performance parameters realized and / or calculated, notions relating to technical assistance for the design and implementation of a glass surface.

[0072] Said non-public sources are therefore information that is part of the company knowhow and that are provided to train the algorithm by the company itself through the training device / software 30.

[0073] Alternatively or in combination with the preceding embodiment, step al) provides for training the artificial intelligence algorithm 1 by providing public sources (orinformation) such as for example written and photographic sources found in at least HTML-type pages, text and image files and design and exchange files such as, for example, XML or JSON having as their object technical information relating to the design of a glass surface.

[0074] For example, the public HTML-type pages comprise all the information that is present in the so-called FAQs (i.e. Frequently Asked Questions) publicly available.

[0075] Said public sources are therefore freely obtainable information that is part of the common knowledge and that is provided to train the artificial intelligence algorithm 1 by the company itself through the training device / software 30.

[0076] In other words, while the non-public sources represent the company know-how for the implementation of possible glass surface projects, the public sources represent the freely accessible documentation having as its object the instructions, suggestions, known experiences for the implementation of possible glass surface projects.

[0077] In both cases of public / non-public sources, it is preferably the company that instructs the artificial intelligence algorithm by means of the training device / software 30.

[0078] It should be noted that the step of inputting the information for training through the training device 30 occurs in a direct manner, i.e. by means of the manual loading by an operator of said information and / or in an indirect manner in which said device 30is programmed to perform access to the web and / or to the information from company databases to retrieve updated information on the subject.

[0079] In accordance with an embodiment combinable with the preceding one, step a) comprises a step a2) of training the artificial intelligence algorithm by providing additional implementation statements once the glass surface has been installed and technical support has been requested. According to one aspect, in this case experimental tests may also have been performed to test the performance parameters.

[0080] Specifically, step a2) provides for updating the training of the artificial intelligence algorithm as a function of new know-how acquired during the implementation of the project and / or other projects for the implementation of other glass surfaces.

[0081] Said additional statements may further comprise text and image files and / or similar supports and / or design and exchange files such as, for example, XML or JSON.

[0082] The method comprises step b) of providing by input on the user device 10 one or more initial statements relating to a glass surface design.

[0083] As anticipated, said initial statements comprise performance parameters required for the glass surface selected from optical, energetic, static, acoustic, usage parameters or a combination thereof.The method comprises step c) of processing the provided statements by means of the artificial intelligence algorithm so as to generate the sequence of layers representative of the correspondence between the provided statements and the training of the artificial intelligence algorithm.

[0084] Preferably, step c) provides for receiving the provided statements, analyzing them, understanding them and recognizing the peculiar characteristics of the project in order to frame it within a complex information set associated with the training of the artificial intelligence.

[0085] According to one aspect, the initial statements can comprise a proposed sequence of layers with performance parameters to be modified, or a series of layers that can be used (for example a list of products available in the warehouse).

[0086] Specifically, as anticipated, the artificial intelligence algorithm is resident in a processing unit 20 in signal communication with the user device 10. In this way, the artificial intelligence algorithm directly receives the provided statements for their relative processing.

[0087] The method comprises step d) of generating by means of the artificial intelligence algorithm a final sequence of layers representative of the correspondence between the provided statements and the training of the artificial intelligence algorithm.Advantageously, the final sequence of layers is generated as a function of the information on the project and the training provided so that it is definitive and error-free, being based on the entire technical knowledge available on the subject associated with the company and not only.

[0088] Said final sequence of layers comprises a layering position, one or more layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields and a thickness for each selected layer.

[0089] Preferably, in step c) the artificial intelligence algorithm processes the statements provided on the glass surface design by associating a sequence value to each layer and relative thickness in a layering position representative of the probability of use of said layer with relative thickness in that layering position in the glass surface design. Furthermore, in step d) the artificial intelligence algorithm generates the final sequence of layers comprising the layers with a higher sequence value as a function of the training provided by step c).

[0090] In accordance with a preferred embodiment, step c) comprises step cl) of processing the initial input statements so as to generate a first sequence of layers by using the trained artificial intelligence algorithm. Said initial information can be representative of the project in broad terms without the specific requests to be able to generate the final sequence of layers, but only a first sequence of layers. Otherwise, if the artificial intelligence algorithm recognizes that the initial statements are sufficient, step c) provides for directly generating the final sequence of layers.In case of insufficiency of the information, step c) comprises step c2) of generating by means of the trained artificial intelligence algorithm one or more additional requests on the user device 10 in signal communication with the processing unit 20. Specifically, if the artificial intelligence algorithm were to detect that the initial statements are approximate and have informational deficiencies / ambiguities, the artificial intelligence algorithm, thanks to the training, realizes that the first sequence of layers may not be correct. In fact, the additional requests require the input of one or more additional statements on the user device to resolve informational deficiencies and / or ambiguities necessary to determine a final sequence of layers.

[0091] It follows that step c) comprises step c3) of inputting the additional statements requested by means of the user device 10.

[0092] Subsequently, step c) comprises step c4) of processing the initial statements and the additional statements by means of the trained artificial intelligence algorithm to generate a second sequence of layers which is more specific than the first sequence.

[0093] Finally, step c) comprises step c5) of repeating steps c2)-c4) until a final sequence of layers representative of the correspondence between the provided statements and the training of the artificial intelligence algorithm is generated. In this way, the artificial intelligence algorithm thanks to the training analyzes and iteratively processes the initial and additional statements to converge in an optimized and rapid manner to theinformation necessary for the generation of the final sequence of layers for the relative project.

[0094] According to one aspect, the initial and additional provided statements can comprise textual statements, text and image files and design and exchange files such as, for example, XML or JSON.

[0095] The method also comprises a step e) of providing a calculation algorithm configured to perform calculations of the optical, energetic, static, acoustic, usage type or a combination thereof on the final sequence of layers generated in step d) and providing a performance estimate of the final sequence of layers based on the calculation performed.

[0096] Preferably, the calculation algorithm is of the type known to the person skilled in the art and is specialized in the calculation of performance according to regulations and certified for the relative execution of the calculations.

[0097] The method comprises a step f) of comparing the performance estimate calculated in step e) with the required performance requirements.

[0098] The method comprises a step g) of generating a confirmation signal if the comparison between the performance estimate and the required performance requirements is less than control values and generating a training signal if the comparison between the performance estimate and the required performance requirements is greater than thecontrol values. In this way, the method allows to evaluate the correctness of the sequence of layers generated by the artificial intelligence algorithm or if the final sequence may not actually satisfy the performance requirements.

[0099] Preferably, the control values comprise ranges of values for each performance parameter and are modifiable to be able to reach the combination of required performance parameters and / or get as close as possible.

[0100] If a training signal has been generated, the method comprises a step h) of repeating steps d)-g) excluding the final sequence associated with the training signal and the information provided.

[0101] In accordance with a preferred embodiment, step a) comprises a step a3) of training the artificial intelligence algorithm by providing the final sequence associated with the training signal, the information provided relating to it and the performance estimates if a training signal has been generated in step g).

[0102] In this way, iteratively, the artificial intelligence algorithm is also trained as a function of the calculation of the performance estimates to generate a final sequence of layers against required performance requirements.

[0103] In accordance with a preferred embodiment, the method comprises a step i) of providing the final sequence of layers in the form of a rendering to the user device 10 together with the performance estimates, if a confirmation signal is generated in stepg). In this way, the user can have a vision of the selected sequence of layers and the possibility of providing it directly to a configurator for its relative realization.

[0104] APPLICATION EXAMPLE

[0105] The user enters in the user device as initial statements "I would need a glazing that has a solar factor of 15%, luminous transmittance of 30%, thermal transmittance equal to 1.0, and that resists a wind load equal to 2kPa".

[0106] The Al through the artificial intelligence algorithm evaluates if it has enough elements to be able to proceed and could ask (additional requests): "Do you want to evaluate a specific glass manufacturer?"

[0107] The user responds with the additional requested data (additional statements).

[0108] The Al transfers the data to the artificial intelligence algorithm that tries to convert the values into a final sequence of layers. If it does not find one, it increases the tolerance range with respect to the requested values incrementally until it reaches a desired final sequence of layers, describing it in the form of XML, JSON or other interchange format. (In case the Al does not find a suitable configuration it could ask the user "I am having difficulty finding a configuration, could you indicate the parameters in the form of a "range", i.e. < 30% or > 30% instead of exactly 30%?").Once the final sequence of layers is reached, the Al performs a calculation with the calculation algorithm according to regulations and verifies that the parameters are respected (with the performance calculations), if they are not, it inserts these results in the database to train the artificial intelligence algorithm and resumes. If instead they are, the Al sends to the user device 10 the XML or JSON file (or other exchange file) of the final sequence of layers and the results obtained so that the user can evaluate it.

[0109] Alternatively, the user can provide the Al with a sequence of layers asking for the modification of one or more performance parameters. At this point, the Al would take as input the XML and JSON of the configuration and the results and would use them to "read" the user's request as an integration of the prompt via chat in which he asks that a certain result be varied. The Al would use all these as input for a new evaluation following the steps of the method.

[0110] Obviously, a person skilled in the art may make numerous equivalent modifications to the variants described above, without thereby departing from the scope of protection defined by the appended claims.

Claims

CLAIMS1. Method (1) for the generation of a sequence of layers for a glass surface, said sequence of layers comprising, according to a layering position, layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elements and shields or a combination thereof each having its own thickness, said method being characterized in that it comprises the steps of:a) providing a machine learning generative artificial intelligence algorithm based on a conversational model, said artificial intelligence algorithm being trained by means of public and / or non-public sources, said sources being chosen from written and photographic sources found in HTML-type pages, text files and images and design and exchange files, said artificial intelligence algorithm being trained to generate a sequence of layers for a glass surface necessary to make a glass surface by selecting one or more layers, a relative thickness for each selected layer and a layering position; b) providing by input on a user device (10) one or more initial statements relating to a glass surface design, said initial statements comprising performance parameters required for the glass surface selected from optical, energetic, static, acoustic, usage parameters or a combination thereof;c) processing the statements provided by means of said artificial intelligence algorithm so as to generate said sequence of layers, said sequence of layers being representative of the correspondence between the statements provided and the training of the artificial intelligence algorithm, said artificial intelligence algorithm being resident in a processing unit (20) in signal communication with the user device (10),d) generating, by means of said artificial intelligence algorithm, a final sequence of layers representative of the correspondence between the statements provided and the training of the artificial intelligence algorithm.

2. The method (1) according to claim 1, comprising the step of: wherein said step a) comprises a step al) of preliminarily training the artificial intelligence algorithm by providing HTML-type pages, text files and images and public and / or non-public design and exchange files having technical information relating to the design of a glass surface, said technical information provided comprising technical sheets, calculations, sequences of layers for glass surfaces associated with related parameters of realized and / or calculated performance and notions relating to technical assistance for the design and implementation of a glass surface.

3. The method (1) according to claim 1 or 2, wherein step a) comprises a step a2) of training the artificial intelligence algorithm by providing additional implementation statements once the glass surface has been installed and technical support has been requested.

4. The method (1) according to any one of claims 1 to 3, wherein in step c) the artificial intelligence algorithm processes the provided statements on the glass surface design by associating a sequence value to each layer and relative thickness in a layering position representative of the probability of use of said layer with relative thickness in that layering position in the glass surface design and in step d) the artificial intelligencealgorithm generates the final sequence of layers comprising the layers with higher sequence value as a function of the training provided by step c).

5. The method (1) according to any one of claims 1 to 4, wherein step c) comprises the steps ofcl) processing the initial input statements so as to generate a first sequence of layers by using the trained artificial intelligence algorithm;c2) generating, by means of the trained artificial intelligence algorithm, one or more additional requests on the user device (10) in signal communication with the processing unit (20), said additional requests requiring the input of one or more additional statements on the user device (10) to resolve informational deficiencies and / or ambiguities necessary to determine a final sequence of layers;c3) inputting the additional statements requested by means of the user device (10); c4) processing the initial statements and the additional statements by means of the trained artificial intelligence algorithm to generate a second sequence of layers which is more specific than the first sequence;c5) repeating steps c2) -c4) until a final sequence of layers representative of the correspondence between the statements provided and the training of the artificial intelligence algorithm is generated.

6. The method (1) according to any one of claims 2 to 5, further comprising the steps subsequent to step d) ofe) providing a calculation algorithm configured to perform calculations of the optical, energetic, static, acoustic, usage type or a combination thereof on the final sequenceof layers generated in step d) and providing a performance estimate of the final sequence of layers based on the calculation performed; said calculation algorithm being resident in the processing unit (20) in signal communication with the user device (10)f) comparing the performance estimate with the performance requirements;g) generating a confirmation signal if the comparison between the performance estimate and the performance requirements is less than control values and generating a training signal if the comparison between the performance estimate and the performance requirements is greater than the control values;h) repeating steps d)-g) if a training signal has been generated excluding the final sequence associated with the training signal and the information provided.

7. The method (1) according to claim 6, wherein step a) comprises a step a3) of training the artificial intelligence algorithm by providing the final sequence associated with the training signal, the information provided relating to it and the performance estimates if a training signal has been generated in step g).

8. The method (1) according to claim 6 or 7, further comprises step i) of providing the final sequence of layers in the form of a rendering to the user device (10) together with the performance estimates, if a confirmation signal is generated in step h).

9. Computer system for implementing a method for the generation of sequence of layers for a glass surface, comprising, according to a layering position, layers selected from glasses, coatings, plastic interlayers, gaps with air and / or argon, opaque elementsand shields or a combination thereof each having its own thickness, said system comprising a user device (10), a processing unit (20) in signal communication with the user device (10) through a link suitable to support an Internet communication protocol, a database (40) in signal communication with the processing unit (20) through a link suitable to support an Internet communication protocol, a machine learning generative artificial intelligence algorithm based on a conversational model resident in said processing unit (20), said artificial intelligence algorithm being interrogable through said user device (10), said system being characterized in that it is configured to carry out the steps of the method according to the preceding claims from 1 to 8.

Images