System and method for controlling a forming machine for manufacturing a haircraft conductor element of an induction winding of a stator

Abstract

A system (10) for controlling a forming machine (12) for manufacturing hairpin conductor elements (17) of induction windings of stators, the system comprising a source of geometric information (18) of at least one hairpin conductor element (17) obtained with the forming machine (12) and a processing device (20) connected to the source of geometric information (18) and connectable to the forming machine (12); wherein the geometry information source (18) is configured to provide current geometry information (19) of the at least one hairpin conductor element (17), and the processing device (20) comprises: a state generator (24) configured to receive a set of information, the set of information comprises at least current geometry information (19) of the hairpin conductor element (17) provided by the geometry information source (18), reference geometry information of a reference hairpin conductor element, and shaping parameters (13) used by the shaping machine (12), and is further configured to generate state information (29) based on the received set of information; and an intelligent agent (30) configured to process the state information (29) using an artificial intelligence (AI) algorithm and autonomously generate a correction action (31) for the shaping parameter (13); the processing device (20) is caused to adjust the shaping parameter (13) based on the corrective action (31) generated by the smart agent (30).

Description

[0001] The present invention relates to a system and a method for controlling operations (bending, calendering, advancing and cutting) performed on electric wires by a forming machine for manufacturing hairpin type conductor elements, more simply hairpin conductor elements, even more simply hairpins, of the induction winding of a stator.

[0002] The system and the method according to the present invention are particularly useful and practical in the field of controlling the forming operations for producing hairpin conductor elements which constitute the induction winding of a stator of an electric machine, for example a motor or a generator, but are not exclusive.

[0003] As is known, electric motors, direct current generators, alternating current generators and transformers comprise a core of ferromagnetic material on which windings made of electric wires arranged according to a specific geometry are arranged. The circulation of an electric current in at least one winding determines, by electromagnetic induction, the circulation of an induced current in at least one other winding. Moreover, between the ferromagnetic core and the respective windings, a force interaction occurs and is able, for example, to make the rotor of an electric motor rotate with respect to the stator.

[0004] As mentioned above, the above-mentioned induction windings are made using wires of electrically conductive material, usually copper. For specific applications, the induction windings are manufactured using wire-like elements, more simply hairpin conductor elements, of electrically conductive material which are first inserted into specific slots provided in the stator, i.e. the ferromagnetic core, of the electric machine constructed and then stably coupled to each other at least at one end, usually by a welding operation.

[0005] In general, the induction windings of a stator comprise various hairpin conductor elements of different shapes. Basically, many shapes can be manufactured, which depend only on the capabilities of the forming machine.

[0006] Typical, but not limiting, examples of these hairpin conductor elements are those shaped like a fork. The fork has a pair of straight shanks which are connected to each other at one end by a bridge-like crosspiece. Typically, the shape of the fork is roughly like an upside-down U, the shape of the bridge is like a toothpick. Each shank of the fork and therefore of the hairpin conductor element has a free end for insertion into a respective slot of the stator of the electric machine. In particular, according to the desired logic of the induction winding of the electric machine, the first end of each hairpin conductor element is inserted into a respective first slot, while the second end of the same hairpin conductor element is inserted into a respective second slot.

[0007] Generally, the production of the hairpin conductor elements can be a sequence of N shaping operations (bending, calibrating, pushing and cutting). These hairpin conductor elements are produced by shaping machines adapted to perform this sequence of shaping operations on a wire made of an electrically conductive material, typically copper. The operations performed by these shaping machines are controlled by shaping instructions which comprise adjustable and modifiable shaping parameters enabling to obtain hairpin conductor elements having the necessary structural features (e.g. shape, dimensions, etc.) for their intended use.

[0008] This treatment of the wire according to the shaping instructions requires very precise operations, since the hairpin conductor elements need to be shaped to comply with very low, i.e. tight, dimensional tolerances. The purpose of the tight dimensional tolerances required for shaping the hairpins is to facilitate the insertion of these hairpins into the slots provided in the stator of an electric machine under construction, especially when there are multiple hairpins to be inserted into small spaces at the same time. Indeed, the purpose is to provide an induction winding complying with the required specifications.

[0009] Generally, in mechanical technology, the term "tolerance" denotes, in the industrial manufacturing of a part (in this case a hairpin conductor element), the allowed deviation between the ideal reference measurement defined by the design drawing and the effective measurement of the produced part; more precisely, the difference or range between the maximum and minimum allowed measurements.

[0010] However, during the shaping of the hairpin conductor elements, disturbing conditions external to the shaping machine and / or working conditions internal to the shaping machine can occur which generate drifts, causing the production of hairpin conductor elements to increasingly fall outside the tolerance interval, which must therefore be discarded or at least reworked. For example, the disturbing conditions external to the shaping machine can include variations in the mechanical properties (tensile modulus, yield stress, etc.) of the wire or rather of the corresponding material used, variations in the elastic recovery effect or "springback" and variations in the ambient temperature. For example, the working conditions internal to the shaping machine can include play and variations due to wear, and variations in the dimensions of the shaping tools (e.g. due to wear or due to tooling changes).

[0011] At present, the quality of the shaping operations, and therefore of the hairpin conductor elements of the induction winding of the stator, is periodically and on a spot-check basis judged by a human operator using equipment adapted to be magnified, such as a digital microscope, and / or measuring instruments for mechanical testing, such as a control template.

[0012] For simplicity, a hairpin conductor element can be considered to be of good quality if the measurements characterizing it fall within the above-mentioned tolerance interval with respect to the corresponding reference measurements of a reference (or master) hairpin conductor element, obviously depending on the specific type of hairpin conductor element under production.

[0013] Therefore, the aim of the quality checks performed by the human operator is to verify whether the measurement results characterizing the hairpin conductor elements fall within the tolerance interval of the corresponding reference measurement results, and therefore whether these hairpin conductor elements have the necessary structural features for their intended use.

[0014] Please note that the selection of the measurement results obtained from the hairpin conductor elements is very critical. In fact, these measurement results must uniquely characterize the hairpin conductor elements and they must be able to be distinguished from the inevitable measurement noise.

[0015] Moreover, currently, many conventional forming machines have a set of automatic control devices that detect any problems in the forming operation and / or any defects in the hairpin conductor elements. In this case, i.e. in the event of problems and / or defects, the forming machine emits an alarm signal (acoustic and / or visual) that alerts the corresponding human operator who performs the above-mentioned quality checks on the hairpin conductor elements.

[0016] If these quality checks (performed periodically and on a sample basis or as a result of the alarm signal emitted by the forming machine) give negative results, the human operator will take action on the forming parameters of the forming instructions in order to recalibrate the forming machine, adapt the operations performed by the forming machine to the interference conditions outside the forming machine and to the working conditions inside the forming machine and therefore return to obtaining hairpin conductor elements that comply with the specifications, i.e. hairpin conductor elements that have the necessary structural features for their intended use.

[0017] However, this conventional method is not without drawbacks, including the fact that the quality checks and any corrections of the forming parameters of the forming instructions are closely dependent on the experience and / or skills of the human operator who performs them. In practice, the same human operator will act in a subjective manner and therefore can make different decisions in different environments, on different days, etc.

[0018] Another drawback of this conventional method is that it requires long reaction times, which lead to the production of a large number of hairpin conductor elements that must be rejected even if the quality checks give negative results.

[0019] In particular, these reaction times depend both on the time elapsed between the production of the first hairpin having a measurement result outside the tolerance interval and the negative result of the quality check, and on the time elapsed between the negative result of the quality check and the correction of the forming parameters of the forming instructions.

[0020] As regards this second point, it is noted that there is a complex dependence between the measurement results of the hairpin conductor elements, the focus of the quality check and the shaping parameters of the shaping instructions to be set in the shaping machine. Therefore, many shaping parameters of the shaping instructions need to be changed simultaneously in order to correct even a single measurement result of the hairpin conductor elements. Basically, this correction process is very complex and cannot be completed by anyone, even by a human operator with experience and skills, who needs a lot of time to successfully complete it.

[0021] Another drawback of this conventional approach basically consists in that it leads to an iterative correction of the shaping parameters of the shaping instructions, in particular following a "trial and error" approach, which is very wasteful, in particular because it requires a long time and is difficult to perform.

[0022] Finally, automatic systems for controlling the shaping machine are currently known, i.e. automatic systems for controlling the operation performed by the shaping machine for manufacturing the hairpin conductor elements of the induction winding of the stator.

[0023] These known automatic systems are configured to monitor the hairpin conductor elements of the induction winding of the stator manufactured by the shaping machine and to modify the shaping parameters of the shaping instructions, basically without the need for human intervention, in order to recalibrate the shaping machine, to adapt the operation performed by the shaping machine to the interference conditions outside the shaping machine and to the working conditions inside the shaping machine, so as to obtain hairpin conductor elements that comply with the specifications, i.e. hairpin conductor elements having the necessary structural characteristics for their intended use.

[0024] However, these known automatic systems are not without drawbacks, one of which is the fact that their operation is based on measurement results of the hairpin conductor elements associated or associable one step or one operation at a time with the individual shaping steps, i.e. with the individual shaping operations, which are often unreliable, difficult to define and often too noisy.

[0025] As mentioned above, the production of the hairpin conductor elements can be a sequence of N shaping operations (bending, calendering, advancing and cutting).

[0026] Ideally, by isolating the effect of the nth shaping step through the corresponding measurement of the hairpin conductor element, it would seem convenient to correct the shaping parameters of the shaping instructions related to that step independently of the corrections required for the previous and subsequent shaping steps. For example, if the angle alpha_n caused by the nth shaping step could be identified precisely, and if the angle desired_alpha_n of the reference hairpin conductor element were known, a single measurement feedback of the angle alpha_n could be used to correct only a few shaping parameters of the shaping instructions only for the nth shaping step.

[0027] However, in real-world situations, the measurement of the hairpin conductor element that can actually be obtained with precision and without interference from acquisition noise of the current geometry information depends on and is therefore influenced by many or even all of the shaping steps. Controlling is therefore a very complex matter. For example, a reliable measurement is the distance between the ends of the two legs that constitute the hairpin conductor element, or simply distance AB, and this measurement depends on and is therefore influenced by all of the shaping steps.

[0028] Another drawback of these known automatic systems is that, in following the above-mentioned simplified control method, which consists in making corrections one shaping step at a time, the errors generated at each shaping step inevitably add up together, with the risk of impairing the measurements that are actually important for the hairpin conductor element, such as the AB measurement mentioned above.

[0029] Another drawback of these known automatic systems is that they do not allow to take into account variations of the working conditions inside the shaping machine.

[0030] The aim of the present invention is to overcome the limitations of the known art described above by devising a system and a method for controlling a shaping machine for manufacturing hairpin conductor elements of the induction winding of a stator that makes it possible to obtain better results and / or similar results with lower costs and higher performance levels than the conventional solutions can obtain.

[0031] Within this aim, the aim of the present invention is to conceive a system and a method for controlling a shaping machine for manufacturing hairpin conductor elements of the induction winding of a stator that makes it possible to compensate for variations of the shaping process, i.e. variations of the operation of the shaping machine over time (the shaping process and / or the operation of the shaping machine are not constant over time), caused for example by interference conditions and / or working conditions.

[0032] Another aim of the present invention is to devise a system and a method for controlling a shaping machine for manufacturing hairpin conductor elements of the induction winding of a stator that makes it possible to adapt the operation of the shaping machine, i.e. the operation in the shaping process performed by the shaping machine, to interference conditions external to the shaping machine, such as for example variations of the mechanical properties (tensile modulus, yield stress, etc.) of the wire or rather of the relevant material used, variations of the springback and of the ambient temperature.

[0033] Another aim of the present invention is to devise a system and a method for controlling a shaping machine for manufacturing hairpin conductor elements of the induction winding of a stator that makes it possible to adapt the operation of the shaping machine, i.e. the operation in the shaping process performed by the shaping machine, to working conditions internal to the shaping machine, such as for example clearances and variations due to wear.

[0034] Another object of the present invention is to conceive a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, making it possible to enable the forming process, in particular the operations performed by the forming machine, to be independent of the ability and / or condition of a human operator, thus passing from a subjective check to an objective check, which produces predictable and repeatable results.

[0035] Another object of the present invention is to design a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, making it possible to eliminate or at least minimize the reaction time after the first hairpin having a measurement outside the tolerance interval is produced.

[0036] Another object of the present invention is to design a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, making it possible to correct at least one measurement of a hairpin conductor element by simultaneously modifying a plurality of forming parameters of the forming instructions to be set in the forming machine using an artificial intelligence algorithm.

[0037] Another object of the present invention is to design a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, making it possible to comfortably keep the forming machine within the corresponding forming operating limits, which can make the operation of the forming machine unstable and not very robust.

[0038] Another object of the present invention is to design a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, making it possible to provide greater stability for the forming process, in particular the operations performed by the forming machine.

[0039] Another object of the present invention is to provide a system and a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, which are highly reliable, easy to implement and practical compared to known techniques and are economically competitive.

[0040] This and other objects, which will become more apparent hereinafter, are achieved by a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, said system comprising a source of geometric information of at least one hairpin conductor element obtained with said forming machine and a processing device operatively connected to said source of geometric information and connectable to said forming machine; wherein said source of geometric information is configured to provide current geometric information of said at least one hairpin conductor element and said processing device comprises: a state or condition generator configured to receive a set of information comprising at least the current geometric information of the hairpin conductor element provided by the geometric information source, reference geometric information of a reference hairpin conductor element and shaping parameters used in the shaping machine, and further configured to generate state or condition information based on the received set of information; and an intelligent agent configured to process the state information using artificial intelligence algorithms and autonomously generate correction actions for the shaping parameters; so that the processing device adjusts the shaping parameters based on the correction actions generated by the intelligent agent.

[0041] The object and the aim are also achieved by a method for controlling a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator, controlled by means of a geometric information source of at least one hairpin conductor element and a processing device operatively connected to the geometric information source and connectable to the shaping machine, the method comprising the steps of: providing, by means of the geometric information source, current geometric information of the at least one hairpin conductor element obtained with the shaping machine; receiving, by means of a state or condition generator of the processing device, a set of information comprising at least the current geometric information of the hairpin conductor element provided by the geometric information source, reference geometric information of a reference hairpin conductor element and shaping parameters used by the shaping machine, and further generating state or condition information based on the received set of information; and processing, by means of an intelligent agent of the processing device, the state information using artificial intelligence algorithms and autonomously generating correction actions for the shaping parameters; so that the processing device adjusts the shaping parameters based on the correction actions generated by the intelligent agent.

[0042] The present invention also relates to a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator according to claim 15.

[0043] The present invention also relates to a system for controlling a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator and a method for training the system according to claims 28 and 29, respectively.

[0044] Further characteristics and advantages of the present invention will become better apparent from the description of a preferred, but non exclusive, embodiment of a system for controlling a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator and of a method according to the present invention, illustrated by way of non-limitative example in the accompanying drawings, in which: Figure 1is a schematic block diagram of an embodiment of a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 2A and Figure 2B are two schematic block diagrams, where the first is shown in general in Figure 2A and the second is shown in detail in Figure 2B : elements and data flow of a first embodiment of a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 3A and Figure 3B are two schematic block diagrams, where the first is shown in general in Figure 3A and the second is shown in detail in Figure 3B : elements and data flow of a second embodiment of a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 4A and Figure 4B are two schematic block diagrams, where the first is shown in general in Figure 4A and the second is shown in detail in Figure 4B : elements and data flow of a third embodiment of a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 5 is a schematic flow diagram of an embodiment of a method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 6 is a schematic block diagram showing data of a digital twin of a forming machine for training an embodiment of a system for controlling the forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application, as well as input information and output information; Figure 7 is a schematic block diagram showing a logic for generating a simulated hairpin conductor element, followed by a digital twin of a forming machine for training an embodiment of a system for controlling the forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 8 is a schematic block diagram of elements and data flow of an embodiment of a system for training a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application in the absence of variability (or noise); Figure 9A and Figure 9Bis a block and graph diagram schematically showing the training of variability (or noise) in a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application; Figure 10 is a schematic block diagram of the elements and data flows of an embodiment of a system for training a system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator in the presence of variability (or noise) according to the present application; Figure 11A and Figure 11B is a perspective view of an example of a hairpin conductor element of an induction winding of a stator.

[0045] With reference to Figures 1 to 4B , the system for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application, generally indicated by reference number 10, essentially comprises a source of geometric information 18, preferably three-dimensional (3D), and a processing device 20. The source of geometric information 18 can be associated with or associable with the forming machine 12. The processing device 20 is operatively connected or connectable to the source of geometric information 18 and vice versa. The processing device 20 is operatively connected or connectable to the forming machine 12 and vice versa.

[0046] The forming machine 12 is configured to manufacture a plurality of real hairpin conductor elements 17 and therefore to provide these hairpin conductor elements in output.

[0047] The forming machine 12 comprises an electronic control unit 14, preferably of the programmable logic controller (PLC) type, and a forming device 16.

[0048] The electronic control unit 14 of the forming machine 12 is operatively connected to the forming device 16 and is adapted to process and interface with the forming device 16 and with the processing device 20.

[0049] The electronic control unit 14 of the forming machine 12 is configured to command, control and coordinate the operation of the forming device 16 according to forming instructions, as said, which comprise forming parameters 13 that enable the obtaining of hairpin conductor elements 17 having the necessary structural characteristics for their intended use.

[0050] The forming device 16 of the forming machine 12 is configured to shape an electrical wire into a real hairpin conductor element and therefore to produce real hairpin conductor elements 17. Examples of forming devices for shaping an electrical wire into a hairpin conductor element are disclosed in WO2012 / 156066 and WO2015 / 132180. The forming device 16 of the forming machine 12 is operatively connected to the electronic control unit 14.

[0051] Advantageously, the shaping device 16 of the shaping machine 12 comprises a plurality of electromechanical tools configured to perform on the electrical wire the operations of bending, calendering, advancing and cutting for producing in output a plurality of real hairpin conductor elements 17.

[0052] In an embodiment, the shaping device 16 comprises a bending unit 17A configured to perform on the electrical wire the operation of bending when producing the plurality of real hairpin conductor elements 17.

[0053] In an embodiment, the shaping device 16 comprises a calendering unit 17B configured to perform on the electrical wire the operation of calendering when producing the plurality of real hairpin conductor elements 17.

[0054] In an embodiment, the shaping device 16 comprises an advancing unit 17C configured to perform on the electrical wire the operation of advancing when producing the plurality of real hairpin conductor elements 17.

[0055] In an embodiment, the shaping device 16 comprises a cutting unit 17D configured to perform on the electrical wire the operation of cutting when producing the plurality of real hairpin conductor elements 17.

[0056] As mentioned above, the operations of the shaping device 16 of the shaping machine 12 are commanded, controlled and coordinated by the electronic control unit 14 according to the shaping instructions.

[0057] The geometric information source 18 of the system 10 according to the present application is configured to provide current geometric information 19, preferably three-dimensional (3D), about at least one hairpin conductor element 17 produced by the shaping machine 12 and thus output from the shaping machine. The geometric information source 18 is further configured to send the current geometric information 19 about the hairpin conductor element 17 to the processing device 20.

[0058] The geometric information source 18 can be of various types. Therefore, the current geometric information 19 about the hairpin conductor element 17 provided by the geometric information source 18 can be of various types. Therefore, the accuracy of the current geometric information 19 can vary.

[0059] With reference to Figure 11A and Figure 11B , the current geometric information 19 about the hairpin conductor element 17 can comprise or represent one or more measurements of said hairpin conductor element 17 selected from the group consisting of: segment AC, segment BD, segment CE, segment DE, segment CD, bending point C, bending point D, head width HeadWidth, virtual rest plane Pav, first torsion plane Pt1, first normal plane Pn1, central bending profile S XYZ, first central bending segment Sseg1, first central bending circumference Scir1, second central bending circumference Scir2, second central bending segment Sseg2, first calendering circumference C1, first calendering plane Pc1, first calendering radius Rc1, second calendering circumference C2, second calendering plane Pc2, second calendering radius Rc2, second twisting plane Pt2, second normal plane Pn2, angle between the twistings , first twisting angle , second twisting angle , central bending angle , first bending angle , second bending angle , first leg trimming height and second leg trimming height .

[0060] For example, the current geometrical information 19 about the hairpin conductor element 17 can be obtained using a measuring instrument for mechanical testing, for example a control template, and then the obtained information is provided by manual input, for example using a computer operatively connected or connectable to the processing device 20.

[0061] In an embodiment, the geometrical information source 18 is a 3D vision device 18 associated or associable with the shaping machine 12. Preferably, the 3D vision device 18 is a laser scanner sensor. In this embodiment, the current geometrical information 19 is a digital 3D reconstruction 19 of the hairpin conductor element 17.

[0062] The 3D vision device 18 of the system 10 according to the present application is configured to generate a digital three-dimensional (3D) reconstruction 19 of the shape or geometry (where the two terms are understood as synonyms) of at least one hairpin conductor element 17 produced by the shaping machine 12 and therefore output from the shaping machine. The 3D vision device 18 is also configured to send the digital 3D reconstruction 19 of the hairpin conductor element 17 to the processing device 20.

[0063] Preferably, the current geometrical information 19 (for example the digital 3D reconstruction 19) of the hairpin conductor element 17 provided by the geometrical information source 18 (for example generated by the 3D vision device 18) is a point cloud (point cloud data PCD), as shown in Figures 2A to 4B

[0064] ​In an embodiment, the 3D vision device 18 is operatively connected to or comprises a supervisor (not shown) configured to preliminarily verify the correctness and therefore the reliability of the digital 3D reconstruction 19 of the hairpin conductor element 17 generated by the 3D vision device 18, i.e. before sending it to the processing device 20. In general, the reconstruction of a hairpin conductor element 17 is correct and reliable if its digital 3D reconstruction 19 does not have anomalies that make it unusable for defining the shape or geometry of the hairpin conductor element 17.

[0065] In general, three types of anomalies can currently be found in the digital 3D reconstruction 19 of the hairpin conductor element 17, i.e. these anomalies can manifest themselves in substantially three ways.

[0066] The first type of anomaly is the lack of points in the region of the point cloud where they should be found. Since these points are not present, it is not possible to obtain 3D geometric information about that region of the digital 3D reconstruction 19 of the hairpin conductor element 17. Therefore, it is not possible to complete the information 29 about the status of at least one hairpin conductor element 17 generated by the status generator 24, which will be described in detail below.

[0067] The second type of anomaly is the presence of spikes, i.e. local regions where the detected points differ greatly from the adjacent ones. Therefore, the 3D geometric information about that region of the digital 3D reconstruction 19 of the hairpin conductor element 17 is severely distorted. Therefore, the information 29 about the status of at least one hairpin conductor element 17 generated by the status generator 24 will be incorrect, which will be described in detail below.

[0068] The third type of anomaly is represented by a point cloud that is "corrugated", "creped" due to the anomalous oscillations exhibited by the hairpin conductor element 17 during reading with the 3D vision device 18. In this case, the 3D geometric information about that region of the digital 3D reconstruction 19 of the hairpin conductor element 17 is also severely distorted. Therefore, the information 29 about the status of at least one hairpin conductor element 17 generated by the status generator 24 will be incorrect, which will be described in detail below.

[0069] Advantageously, the supervisor of the 3D vision device 18 is robust to the inevitable inaccuracies in the digital 3D reconstruction 19 of the hairpin conductor element 17 generated by the 3D vision device 18. In other words, advantageously, the supervisor of the 3D vision device 18 operates on the digital 3D reconstruction 19 of the hairpin conductor element 17 by applying a predefined tolerance interval.

[0070] Advantageously, the supervisor of the 3D vision device 18 is configured to associate a numerical value with the degree or weight of any of the above-mentioned anomalies. If this numerical value exceeds an empirically defined threshold, the digital 3D reconstruction 19 of the hairpin conductor element 17 is marked as incorrect, and therefore unreliable.

[0071] If the result of the preliminary verification performed by the supervisor of the 3D vision device 18 is positive, i.e. when the digital 3D reconstruction 19 is free of anomalies and therefore correct and reliable, the 3D vision device 18 is configured to send the digital 3D reconstruction 19 of the hairpin conductor element 17 to the processing device 20.

[0072] Conversely, if the result of the preliminary verification performed by the supervisor of the 3D vision device 18 is negative, i.e. when the digital 3D reconstruction 19 presents anomalies and therefore is incorrect and unreliable, the 3D vision device 18 is configured not to send the digital 3D reconstruction 19 of the hairpin conductor element 17. In this case, the supervisor of the 3D vision device 18 can be configured to send an anomaly signal of the digital 3D reconstruction 19, etc. to the processing device 20.

[0073] The preliminary verification performed by the supervisor of the 3D vision device 18 has the advantage of stopping immediately when it starts to propagate errors deriving from the presence of anomalies in the digital 3D reconstruction 19 of the hairpin conductor element 17. In fact, if the anomalous digital 3D reconstruction 19 of the hairpin conductor element 17 is sent to the processing device 20, this is likely to cause incorrect evaluations and / or actions of said device, with a potential negative impact on the forming machine 12 and therefore on the production of the hairpin conductor element 17.

[0074] The processing device 20 of the system 10 according to the present application comprises an electronic control unit 22. The electronic control unit 22 is the main functional element of the processing device 20 and, for this purpose, it is operatively connected with the other elements comprised in the processing device 20.

[0075] The electronic control unit 22 of the processing device 20 is provided with suitable capabilities for the calculation and for interfacing with the other elements of the processing device 20, and it is configured to command, control and coordinate the operation of the elements of the processing device 20 to which it is operatively connected.

[0076] The processing device 20 of the system 10 according to the present application further comprises a state generator 24 configured to generate state information 29, for example comprising information on at least one hairpin conductor element 17, on the forming machine 12, etc. In particular, the state generator 24 of the processing device 20 is configured to receive a set of information deriving from various sources, including for example the geometric information source 18 and the forming machine 12, and then generate the state information 29 based on all this information (as is or processed).

[0077] The geometric information source 18 is configured to transmit current geometric information 19 about the at least one hairpin conductor element 17 to the state generator 24 of the processing device 20.

[0078] Advantageously, the state generator 24 of the processing device 20 is robust to inevitable inaccuracies in the digital 3D reconstruction 19 of the hairpin conductor element 17 generated by the 3D vision device 18. In other words, advantageously, the state generator 24 of the processing device 20 operates on the digital 3D reconstruction 19 of the hairpin conductor element 17 by applying a predefined tolerance interval.

[0079] The set of information received by the state generator 24 of the processing device 20 can comprise one or more items of information selected from the group consisting of: the current geometric information 19 about the at least one hairpin conductor element 17 provided by the geometric information source 18, for example, the digital 3D reconstruction 19 of the shape or geometry of the at least one hairpin conductor element 17 provided by the 3D vision device 18; an anomaly signal derived from the digital 3D reconstruction 19 of the supervisor (if any) of the 3D vision device 18 (if any); the shaping parameters 13 currently used by the shaping machine 12, or rather, used in the shaping instructions executed by the shaping machine 12; the current operating data 15 provided by the shaping machine 12 (for example, speed, torque, etc.); and context or background information 27, for example, provided by a human operator using an adapted data input device, or provided by the shaping machine 12, for example, recorded in the memory unit 34.

[0080] Therefore, the state information 29 can comprise one or more items of information selected from the group consisting of: the current geometric information 19 about the at least one hairpin conductor element 17 provided by the geometric information source 18, for example, the digital 3D reconstruction 19 of the shape or geometry of the at least one hairpin conductor element 17 provided by the 3D vision device 18, or data or measurements obtained from the digital 3D reconstruction 19; an anomaly signal derived from the digital 3D reconstruction 19 of the supervisor (if any) of the 3D vision device 18 (if any); the shaping parameters 13 currently used by the shaping machine 12, or rather, used in the shaping instructions executed by the shaping machine 12; the current operating data 15 provided by the shaping machine 12 (for example, speed, torque, etc.); and The contextual or background information 27 is provided, for example, by a human operator using an adapted data input device, or by the forming machine 12, for example recorded in a memory unit 34.

[0081] The contextual or background information 27 of the status information 29 can comprise one or more items of information selected from the group consisting of: reference geometry information about a reference (or master) hairpin conductor element, for example a digital 3D reconstruction of the shape or geometry of the reference (or master) hairpin conductor element, or data or measurements relating to the reference (or master) hairpin conductor element; the type of hairpin conductor element 17; the current diameter of the wire coil powering the forming machine 12; and data about the collection basket of the hairpin conductor element 17.

[0082] Preferably, the set of information received by the state generator 24 of the processing device 20 comprises at least: current geometry information 19 about at least one hairpin conductor element 17 provided by the geometry information source 18, for example a digital 3D reconstruction 19 of the shape or geometry of at least one hairpin conductor element 17 provided by the 3D vision device 18; reference geometry information about a reference (or master) hairpin conductor element, for example a digital 3D reconstruction of the shape or geometry of the reference (or master) hairpin conductor element; and the forming parameters 13 currently used by the forming machine 12, or rather used in the forming instructions executed by the forming machine 12.

[0083] Preferably, the status information 29 comprises at least: data or measurements obtained from the digital 3D reconstruction 19 of the shape or geometry of at least one hairpin conductor element 17; data or measurements relating to the reference (or master) hairpin conductor element; and the forming parameters 13 currently used by the forming machine 12, or rather used in the forming instructions executed by the forming machine 12.

[0084] Note that the reference (or master) hairpin conductor element can be real or digital, and in the second case the element is therefore simulated.

[0085] Note that the selection of the measurements obtained from the current geometry information 19 of the hairpin conductor element 17 is very critical. In fact, these measurements must uniquely characterize the hairpin conductor element 17 and they must be able to be distinguished from the inevitable noise of the acquisition of the current geometry information 19.

[0086] With reference to Figure 2B , Figure 3B and Figure 4B , in an embodiment, the state generator 24 of the processing device 20 can comprise a meter 26 configured to obtain at least one measurement of the hairpin conductor element 17 from the current geometric information 19 of said hairpin conductor element 17, e.g. from the digital 3D reconstruction 19. Thus, in general, a feature of the hairpin conductor element 17 can be defined by at least one measurement associated with said hairpin conductor element 17.

[0087] In this embodiment, the geometric information source 18 is configured to send the current geometric information 19 of the hairpin conductor element 17 to the meter 26 of the state generator 24 of the processing device 20.

[0088] With reference again to Figure 2B , Figure 3B and Figure 4B , in an embodiment, the state generator 24 of the processing device 20 can comprise an assembler 28 configured to generate the state information 29 by combining together in a predefined order the various information obtainable from the categories listed above.

[0089] In this embodiment, the state information 29 is a set of information grouped together by the assembler 28.

[0090] In an embodiment in which the state generator 24 of the processing device 20 comprises both the meter 26 and the assembler 28, the information grouped together by the latter in order to generate the state information 29 comprises at least one measurement of the hairpin conductor element 17 obtained from the current geometric information 19 of said hairpin conductor element 17, e.g. from the digital 3D reconstruction 19.

[0091] The processing device 20 of the system 10 according to the present application further comprises an intelligent agent 30 configured to process the state information 29 generated and provided by the state generator 24 using artificial intelligence algorithms, or more simply AI algorithms, and to autonomously generate correction actions 31 of the shaping parameters 13 of the shaping instructions.

[0092] The processing device 20 is configured to send the correction actions 31 of the shaping parameters 13 of the shaping instructions to the shaping machine 12 that executes them.

[0093] In this way, based on the processing results of the state information 29 performed by the intelligent agent 30, the processing device 20 recalibrates the shaping machine 12, adjusting the shaping parameters 13 used in said shaping machine 12 based on the correction actions 31 generated by the intelligent agent 30.

[0094] Basically, the intelligent agent 30 generates a suitable correction action 31 of the shaping parameters 13 of the shaping instructions, so that the shaping machine 12 is able to produce a next hairpin conductor element 17 having a shape, geometry or measurement result as close as possible to the shape, geometry or measurement result of the reference (or master) hairpin conductor element. In other words, the intelligent agent 30 generates a suitable correction action 31 of the shaping parameters 13 of the shaping instructions, which enables to balance the errors in the measurement results of the hairpin conductor elements 17 and the constraints of the shaping machine 12.

[0095] Basically, the correction of the shaping parameters 13 of the shaping instructions performed by the shaping machine 12 enables to keep the hairpin conductor elements 17 within the limits of the tolerance interval and, therefore, to not produce or at least limit the hairpin conductor elements 17 to be discarded.

[0096] The correction action 31 generated by the intelligent agent 30 can comprise a correction value of the shaping parameters 13. The correction action 31 generated by the intelligent agent 30 can comprise a correction variation of the values of the shaping parameters 13.

[0097] Conveniently, the intelligent agent 30 always generates a correction action 31 even if the shape, geometry or measurement result of the two hairpin conductor elements (the real hairpin conductor element 17 produced by the shaping machine 12 and the reference (or master) hairpin conductor element) are substantially identical. Obviously, in this case, the correction action 31 can have a substantially minimum and negligible impact, if not actually zero, on the shaping parameters 13. In other words, in this case, the shaping parameters 13 of the shaping instructions performed by the shaping machine 12 can remain substantially unchanged.

[0098] The intelligent agent 30 of the processing device 20 is configured to receive the status information 29 from the status generator 24 of said processing device 20. The status generator 24 of the processing device 20 is configured to send the status information 29 to the intelligent agent 30 of said processing device 20.

[0099] Advantageously, in a first step, the intelligent agent 30 of the processing device 20 uses the information comprised in the status information 29 to define a first transfer function between said status information 29 (in particular the shaping parameters 13) and the shape or geometry of said hairpin conductor elements 17 (where, as mentioned above, the two terms are understood as synonyms). Advantageously, in a second step, the intelligent agent 30 of the processing device 20 defines, through the first transfer function described above, a second transfer function for generating the values of the shaping parameters 13 f() to be used to produce hairpin conductor elements 17 as close as possible to the reference (or master) hairpin conductor element.

[0100] For example, let us assume xis the shape or geometry of a reference (or master) hairpin, y1 is the shape or geometry of a hairpin hairpin_1 and z1 is a shaping parameter 13 for manufacturing hairpin_1 wherein y1 differs from x In order to correct the shaping parameter 13, the intelligent agent 30 uses a transfer function z2 = f(x, y1, z1) wherein z2 after processing results in a hairpin hairpin_2 having a shape or geometry y2 such that y2 - x tends to zero.

[0101] Basically, the intelligent agent 30 defines a first transfer function that combines z1 and y1 and then uses this first transfer function to define a second transfer function z that selects the best y for the specific case it is processing. f() z2 The reason why it is necessary to know both and

[0102] is that the combination between the shape or geometry of the hairpin and the shaping parameter 13 is not fixed but can vary for many reasons, such as the material of the wire, the wear of the shaping machine, etc. y1 z1 At least one measurement of the hairpin conductor element 17 can be obtained from the current geometric information 19 of the hairpin conductor element 17. The at least one measurement of the hairpin conductor element 17 can comprise a distance, but also a geometric element such as a segment, a plane, a point, an angle, etc.

[0103] At least one measurement of the hairpin conductor element 17 can be obtained from the digital 3D reconstruction 19 of the hairpin conductor element 17 generated by the 3D vision device 18.

[0104] In embodiments, the intelligent agent 30 of the processing device 20 can be configured to evaluate at least one measurement of the hairpin conductor element 17 produced by the shaping machine 12 and thus output from the shaping machine 12 with respect to a corresponding measurement of a reference (or master) hairpin conductor element provided as context or background information 27. Preferably, the intelligent agent 30 can be configured to evaluate whether the at least one measurement of the hairpin conductor element 17 falls outside a predefined tolerance interval with respect to the corresponding measurement of the reference (or master) hairpin conductor element.

[0105] In embodiments, the intelligent agent 30 of the processing device 20 can be configured to evaluate at least one measurement of the hairpin conductor element 17 produced by the shaping machine 12 and thus output from the shaping machine 12 with respect to a corresponding measurement of a reference (or master) hairpin conductor element provided as context or background information 27. Preferably, the intelligent agent 30 can be configured to evaluate whether the at least one measurement of the hairpin conductor element 17 falls outside a predefined tolerance interval with respect to the corresponding measurement of the reference (or master) hairpin conductor element.

[0106] ​In this embodiment, the intelligent agent 30 can be configured to autonomously correct the shaping parameters 13 of the shaping instructions based on at least one measurement of the hairpin conductor element 17. In this embodiment, the intelligent agent 30 of the processing device 20 can be configured to autonomously correct the shaping parameters 13 of the shaping instructions based on the difference of at least one measurement of the hairpin conductor element 17 with respect to a corresponding measurement of a reference (or master) hairpin conductor element.

[0107] In embodiments in which the 3D vision device 18 comprises a supervisor, when the digital 3D reconstruction 19 is not reliable, i.e. in the presence of an anomalous signal originating from the digital 3D reconstruction 19 of said supervisor, the intelligent agent 30 of the processing device 20 can be configured to not generate any correction action 31 on the shaping parameters 13 of the shaping instructions, thereby keeping the shaping parameters 13 of the shaping instructions executed by the shaping machine 12 unchanged.

[0108] Advantageously, the intelligent agent 30 is a neural network. As mentioned above, there is a complex dependency between the measurements of the hairpin conductor element, the focus of the quality check and the shaping parameters 13 of the shaping instructions to be set in the shaping machine 12. The use of a neural network makes it possible to intervene simultaneously on many shaping parameters 13 of the shaping instructions in order to correct even a single measurement of the hairpin conductor element 17.

[0109] In embodiments, the intelligent agent 30 is a fully connected neural network (FCNN) and a feedforward neural network (FFNN). In another embodiment, the intelligent agent 30 is a recurrent neural network (RNN). Advantageously, the intelligent agent 30 is trained using techniques, i.e. algorithms, of automatic reinforcement learning (RL), preferably of the actor critic type, preferably with a replay buffer.

[0110] Generally, the association between the inputs and the outputs of the intelligent agent 30 is defined by the many weights or parameters of the neurons of any neural network constituting said intelligent agent 30. Generally, there can be hundreds of thousands, if not millions, of these weights or parameters in a neural network. The update of these weights or parameters of the neural network is performed during the training of the intelligent agent 30, which will be described in detail below.

[0111] As a non-limiting example, the intelligent agent 30 can be a fully connected neural network (FCNN) and a feedforward neural network (FFNN) constituted by an input layer receiving the state information 29, three fully connected hidden layers with rectified linear unit (ReLU) activation functions, these hidden layers being respectively provided with 256, 256 and 128 nodes, and an output layer for generating the correction action 31 on the shaping parameters 13 of the shaping instructions.

[0112] Reference is made to Figure 3A andFigure 3B In an embodiment, the processing device 20 of the system 10 according to the application further comprises a state analyzer 32 configured to preliminarily evaluate the state information 29 generated by the state generator 24, i.e. before sending it to the intelligent agent 30, and then to forward the evaluated state information 29 to the intelligent agent 30 when, based on the result of the preliminary evaluation performed by said state analyzer 32, it is indicated that at least one corrective action 31 is required.

[0113] Basically, the state analyzer 32 is configured to pre-process the state information 29, applying an evaluation logic simpler than the one applied by the above-mentioned intelligent agent 30, and to skip and therefore not to ask said intelligent agent 30 when it does not need to do so.

[0114] In this embodiment, the state generator 24 of the processing device 20 is configured to send the state information 29 to the state analyzer 32 of said processing device 20 instead of the intelligent agent 30.

[0115] Advantageously, during the evaluation of the state information 29, the state analyzer 32 of the processing device 20 is further configured to compare the hairpin conductor element 17 produced by the forming machine 12 and therefore output from it with a reference (or master) hairpin conductor element provided as context or background information 27, in particular to compare the respective digital 3D reconstruction. The purpose of this comparison is to identify any discrepancy between the two hairpin conductor elements, i.e. the real hairpin conductor element 17 produced by the forming machine 12 and the reference (or master) hairpin conductor element, for example in at least one measurement, preferably applying a predefined tolerance interval, in order to detect the need to correct the forming parameters 13 of the forming instructions executed by the forming machine 12.

[0116] As mentioned above, the at least one measurement of the hairpin conductor element 17 can be obtained from the current geometric information 19 of said hairpin conductor element 17. The at least one measurement of the hairpin conductor element 17 can include distances, but also geometric elements such as segments, planes, points, angles, etc.

[0117] As mentioned above, the at least one measurement of the hairpin conductor element 17 can be obtained from the digital 3D reconstruction 19 of said hairpin conductor element 17 generated by the 3D vision device 18.

[0118] In an embodiment, the status analyzer 32 of the processing device 20 can be configured to evaluate at least one measurement of the hairpin conductor element 17 produced by the shaping machine 12 and thus output from the shaping machine with respect to a corresponding measurement of a reference (or master) hairpin conductor element provided as context or background information 27. In particular, the status analyzer 32 can be configured to compare the at least one measurement of the hairpin conductor element 17 with the corresponding measurement of the reference (or master) hairpin conductor element provided as context or background information 27. Preferably, the status analyzer 32 can be configured to evaluate whether the at least one measurement of the hairpin conductor element 17 falls outside a predefined tolerance interval with respect to the corresponding measurement of the reference (or master) hairpin conductor element.

[0119] In an embodiment, the status analyzer 32 of the processing device 20 can be configured to evaluate a trend of at least one measurement of a sequence or series of hairpin conductor elements 17 produced by the shaping machine 12 and thus output from the shaping machine. In particular, the status analyzer 32 can be configured to compare the trend of the at least one measurement of the sequence or series of hairpin conductor elements 17 with a corresponding measurement of a reference (or master) hairpin conductor element provided as context or background information 27. Preferably, the status analyzer 32 can be configured to evaluate whether the trend of the at least one measurement of the sequence or series of hairpin conductor elements 17 falls outside a predefined tolerance interval with respect to the corresponding reference (or master) measurement.

[0120] The purpose of the above evaluation is to verify whether the trend of the measurement diverges and, if necessary, to intervene before the measurement exceeds the tolerance interval and results in the production of hairpin conductor elements 17 that will be rejected.

[0121] If the result of the preliminary evaluation performed by the status analyzer 32 is positive, i.e. when at least one corrective action 31 is required, i.e. a discrepancy is identified that falls outside the tolerance interval between two hairpin conductor elements, and thus the decision made by said status analyzer 32 is positive, the status analyzer 32 of the processing device 20 is configured to forward the status information 29 of the hairpin conductor element 17 or the status information of a sequence or series of hairpin conductor elements 17, for example the last N hairpin conductor elements 17 made by the shaping machine 12, to the intelligent agent 30 of said processing device 20.

[0122] In this embodiment, the intelligent agent 30 can be configured to autonomously correct the forming parameters 13 of the forming instructions based on the trend of at least one of the measurements in the sequence or series of hairpin conductor elements 17 with respect to the corresponding measurements of the reference (or master) hairpin conductor element. Preferably, the intelligent agent 30 can be configured to correct the forming parameters 13 of the forming instructions at each iteration in order to compensate for the diverging trend of the measurements, which in this way extends the operating time of the forming machine 12 before it reaches its operating limits.

[0123] In an embodiment, the status analyzer 32 can be configured to alert a human operator of the presence of an anomalous trend of the measurements that, if not corrected, would fall outside the tolerance interval over time. In this way, the human operator can be alerted of the possible need to perform maintenance and / or intervention on the forming machine 12.

[0124] By contrast, if the result of the preliminary evaluation performed by the status analyzer 32 is negative, i.e. when no corrective action 31 is required, i.e. no difference falling outside the tolerance interval between the two hairpin conductor elements has been identified, and therefore the decision made by the status analyzer 32 is negative, the status analyzer 32 of the processing device 20 is configured not to forward the status information 29 of the hairpin conductor element 17 to the intelligent agent 30 of the processing device 20, and therefore it is configured to“skip” the intelligent agent 30.

[0125] Basically, in this case, in the absence of status information 29, the intelligent agent 30 of the processing device 20 does not perform any corrective action 31 of the forming parameters 13 of the forming instructions, thus leaving the forming parameters 13 of the forming instructions performed by the forming machine 12 unchanged.

[0126] The preliminary evaluation performed by the status analyzer 32 of the processing device 20 has the advantage of saving at least part of the computational resources of the processing device 20, in particular the resources used by the intelligent agent 30, if the result of the preliminary evaluation is negative, i.e. when no corrective action 31 is required, i.e. no difference falling outside the tolerance interval between the two hairpin conductor elements has been identified, and therefore the decision made by the status analyzer 32 is negative.

[0127] With reference to Figure 4A and Figure 4B In an embodiment, the processing device 20 of the system 10 according to the present application further comprises a corrective action analyzer 33 configured to preliminarily evaluate the corrective actions 31 of the forming parameters 13 generated by the intelligent agent 30, i.e. before they are sent to the forming machine 12, and then to forward these corrective actions 31 of the forming parameters 13 to the forming machine 12 when, based on the result of the preliminary evaluation performed by the corrective action analyzer 33, it is indicated that the corrective actions 31 comply with a predefined threshold band.

[0128] In this embodiment, the intelligent agent 30 of the processing device 20 is configured to send the corrective action 31 of the shaping parameters 13 to a corrective action analyzer 33 of said processing device 20.

[0129] If the result of the preliminary evaluation performed by the corrective action analyzer 33 is positive, i.e. when the shaping parameters 13 or the corrective variation of the shaping parameters 13 remain within a predefined threshold band, and therefore the decision made by said corrective action analyzer 33 is positive, the corrective action analyzer 33 of the processing device 20 is configured to forward the corrective action 31 of the shaping parameters 13 to the shaper 12.

[0130] Conversely, if the result of the preliminary evaluation performed by the corrective action analyzer 33 is negative, i.e. when the shaping parameters 13 or the corrective variation of the shaping parameters 13 fall outside a predefined threshold band, and therefore the decision made by said corrective action analyzer 33 is negative, the corrective action analyzer 33 of the processing device 20 is configured not to forward the corrective action 31 of the shaping parameters 13 to the shaper 12, and therefore it is configured to "skip" the shaper 12.

[0131] Basically, in this case, the shaper 12 does not receive any corrective action 31 of the shaping parameters 13 of the shaping instructions, thus keeping the shaping parameters 13 of the shaping instructions performed by said shaper 12 unchanged.

[0132] In general, the above-mentioned predefined threshold band is defined so that the shaping parameters 13 or the corrective variation of the shaping parameters 13 satisfy a suitable set of constraints given by the limits of the shaping process and / or of the shaper 12, for example the spatial occupation of the shaping tools, the correct interaction between the shaping tools and the wires, etc.

[0133] If the variation of the shaping parameters 13 is too large, the preliminary evaluation performed by the corrective action analyzer 33 of the processing device 20 has the advantage of avoiding possible damages to the shaper 12 caused by the corrective actions 31 that would bring the shaping parameters 13 of the shaping instructions outside the operating limits of said shaper 12, these corrective actions 31 often being due to errors on the part of the intelligent agent 30.

[0134] If the variation of the shaping parameters 13 is too small, the preliminary evaluation performed by the corrective action analyzer 33 of the processing device 20 has the advantage of avoiding the application of corrective actions 31 that are substantially ineffective on the shaping process and / or on the shaper 12.

[0135] In a preferred embodiment (not shown), the processing device 20 of the system 10 according to the present application also comprises the state analyzer 32 and the corrective action analyzer 33 both described above.

[0136] Advantageously, the processing device 20 of the system 10 according to the present application further comprises a memory unit 34 configured to record the corrective action 31 on the shaping parameters 13 of the shaping instructions executed by the shaping machine 12. In an embodiment, the memory unit 34 is further configured to record the current geometric information 19 of the at least one hairpin conductor element 17 provided by the geometric information source 18. In an embodiment, the memory unit 34 is further configured to record the digital 3D reconstruction 19 of the shape or geometry of the at least one hairpin conductor element 17 provided by the 3D vision device 18, or data or measurements obtained from the digital 3D reconstruction 19. In an embodiment, the memory unit 34 is further configured to record reference geometric information of a reference (or master) hairpin conductor element. In an embodiment, the memory unit 34 is further configured to record a digital 3D reconstruction of the shape or geometry of the reference (or master) hairpin conductor element, or data or measurements related to the reference (or master) hairpin conductor element. In an embodiment, the memory unit 34 is further configured to record context or background information 27, for example information provided by a human operator using an adapted data input device or by the shaping machine 12.

[0137] With reference to Figure 5 The method for controlling a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application comprises the steps described below.

[0138] First, in a step 41, current geometric information 19, preferably three-dimensional (3D) information, about at least one hairpin conductor element 17 is provided by a geometric information source 18, the hairpin conductor element being produced by the shaping machine 12 and thus output from the shaping machine. The geometric information source 18 can be associated with the shaping machine 12 or associable therewith.

[0139] As mentioned above, the geometric information source 18 can be of various types. Therefore, the current geometric information 19 about the hairpin conductor element 17 provided by the geometric information source 18 can be of various types. Therefore, the accuracy of the current geometric information 19 can vary.

[0140] For example, the current geometric information 19 about the hairpin conductor element 17 can be obtained and then provided using a measuring instrument for mechanical testing.

[0141] In an embodiment, the geometric information source 18 is a 3D vision device 18 associated with the shaping machine 12 or associable therewith. Preferably, the 3D vision device 18 is a laser scanner sensor. In this embodiment, the current geometric information 19 is a digital 3D reconstruction 19 of the hairpin conductor element 17. Preferably, the current geometric information 19 (for example the digital 3D reconstruction 19) of the hairpin conductor element 17 provided by the geometric information source 18 (for example generated by the 3D vision device 18) is a point cloud (point cloud data PCD), asFigures 2A to 4B as shown.

[0142] In an embodiment, after step 41, the method proceeds directly to the subsequent step 45. In another embodiment, after step 41, the method proceeds to the subsequent step 42, as shown. Figure 5

[0143] In this other embodiment, in step 42, the correctness and therefore the reliability of the digital 3D reconstruction 19 of the hairpin conductor element 17 generated by the 3D vision device 18 is preliminarily verified by a supervisor (not shown) operatively connected to or included in said 3D vision device 18, i.e. before sending anything to the processing device 20. As mentioned above, in general terms, the digital 3D reconstruction 19 of the hairpin conductor element 17 is correct and reliable if it does not have anomalies that make it unusable to define the shape or geometry of the hairpin conductor element 17.

[0144] As mentioned above, currently three types of anomalies can be found in the digital 3D reconstruction 19 of the hairpin conductor element 17, i.e. these anomalies can essentially manifest in the three ways mentioned above.

[0145] If the result of the preliminary verification performed by the supervisor of the 3D vision device 18 in step 42 is positive, i.e. when the digital 3D reconstruction 19 is free of anomalies and therefore correct and reliable, in step 43 the digital 3D reconstruction 19 of the hairpin conductor element 17 is sent by the 3D vision device 18 to the processing device 20 and then the method proceeds to the subsequent step 45.

[0146] Conversely, if the result of the preliminary verification performed by the supervisor of the 3D vision device 18 in step 42 is negative, i.e. when the digital 3D reconstruction 19 presents anomalies and therefore is incorrect and unreliable, in step 44 the digital 3D reconstruction 19 of the hairpin conductor element 17 is not sent by the 3D vision device 18 and the method returns to the start, in particular to the previous step 41, thus "skipping" steps 43, 45, 46, 47 and 48. In this case, the anomaly signal of the digital 3D reconstruction 19, etc. can be sent by the supervisor of the 3D vision device 18 to the processing device 20 and then the method returns to the start, in particular to the previous step 41, thus "skipping" steps 43, 45, 46, 47 and 48.

[0147] ​In step 45, the status information 29 is generated by a status generator 24 comprised in the processing device 20, e.g. comprising information about at least one hairpin conductor element 17 on the shaping machine 12, etc. In particular, a set of information originating from various sources, including e.g. the geometry information source 18 and the shaping machine 12, is received and then the status information 29 is generated by the status generator 24 of the processing device 20 based on the various information (as is or processed).

[0148] As mentioned above, preferably, the status information 29 comprises at least: data or measurements obtained from the digital 3D reconstruction 19 of the shape or geometry of at least one hairpin conductor element 17; data or measurements of a reference (or master) hairpin conductor element; and the shaping parameters 13 currently used by the shaping machine 12.

[0149] In an embodiment, still in step 45, at least one measurement of a hairpin conductor element 17 is obtained by a gauge 26 comprised in the status generator 24 of the processing device 20 from the current geometry information 19 about said hairpin conductor element 17, e.g. from the digital 3D reconstruction 19. Thus, typically, a feature of a hairpin conductor element 17 can be defined by at least one measurement associated with said hairpin conductor element 17.

[0150] In an embodiment, still in step 45, the status information 29 is generated by an assembler 28 comprised in the status generator 24 of the processing device 20 by grouping together in a predefined order the various information obtained from the above listed categories. In this embodiment, the status information 29 is a set of information combined by the assembler 28.

[0151] In an embodiment, after step 45, the method proceeds directly to the subsequent step 47. In another embodiment, as Figure 5 shown, after step 45, the method proceeds to the subsequent step 46.

[0152] In this other embodiment, in step 46, the status information 29 is preliminarily evaluated, i.e. assessed before it is sent to the intelligent agent 30, and then, when based on the results of the performed preliminary evaluation it is indicated that at least one corrective action 31 is required, the assessed status information 29 is forwarded to the intelligent agent 30 by a status analyzer 32 comprised in the processing device 20.

[0153] Advantageously, during the evaluation of the status information 29 performed in step 46, the hairpin conductor element 17 produced by the forming machine 12 and thus output from the same is compared, by the status analyzer 32 of the processing device 20, with the reference (or master) hairpin conductor element provided as contextual or background information 27, in particular between the respective digital 3D reconstructions. As mentioned above, the purpose of this comparison is to identify any discrepancy between the two hairpin conductor elements, i.e. the real hairpin conductor element 17 produced by the forming machine 12 and the reference (or master) hairpin conductor element, for example in at least one measurement, preferably applying a predefined tolerance interval, in order to detect the need for a corrective action 31 on the forming parameters 13 of the forming instructions executed by the forming machine 12.

[0154] If the result of the preliminary evaluation performed by the status analyzer 32 in step 46 is positive, i.e. when a corrective action 31 is needed, i.e. a discrepancy is identified that falls outside the tolerance interval between the two hairpin conductor elements, and therefore the decision made by the status analyzer 32 is positive, in step 47 the status information 29 of the hairpin conductor element 17 or of a sequence or series of hairpin conductor elements 17, for example the last N hairpin conductor elements 17 made by the forming machine 12, is forwarded by the status analyzer 32 of the processing device 20 to the intelligent agent 30 of the processing device 20, and then the method proceeds to the subsequent step 48.

[0155] Conversely, if the result of the preliminary evaluation performed by the status analyzer 32 in step 46 is negative, i.e. when a corrective action 31 is not needed, i.e. a discrepancy is not identified that falls outside the tolerance interval between the two hairpin conductor elements, and therefore the decision made by the status analyzer 32 is negative, the status information 29 of the hairpin conductor element 17 is not forwarded by the status analyzer 32 of the processing device 20 to the intelligent agent 30 of the processing device 20, and then the method returns to the start, in particular to the previous step 41, thus "skipping" steps 47 and 48.

[0156] Basically, in this case, step 48 is not performed, and therefore the intelligent agent 30 of the processing device 20 does not perform any corrective action 31 on the forming parameters 13 of the forming instructions, leaving the forming parameters 13 of the forming instructions executed by the forming machine 12 unchanged, in the absence of status information 29.

[0157] Finally, in step 48 the status information 29 is processed, in particular evaluated, and then the method autonomously generates, by the intelligent agent 30 included in the processing device 20, a corrective action 31 that corrects the forming parameters 13 of the forming instructions.

[0158] In this way, based on the result of the processing of the state information 29 performed by the intelligent agent 30, the processing device 20 re-calibrates the forming machine 12, adjusting the forming parameters 13 used in said forming machine 12 based on the corrective action 31 generated by the intelligent agent 30.

[0159] As mentioned above, the intelligent agent 30 generates a suitable corrective action 31 of the forming parameters 13 of the forming instructions, so that the forming machine 12 is able to produce the next hairpin conductor element 17 having a shape, geometry or measurements as close as possible to the shape, geometry or measurements of the reference (or master) hairpin conductor element, preferably within a predefined tolerance interval. In other words, the intelligent agent 30 generates a suitable corrective action 31 of the forming parameters 13 of the forming instructions, so that it is possible to balance the errors in the measurements of the hairpin conductor element 17 and the constraints of the forming machine 12.

[0160] As mentioned above, correcting the forming parameters 13 of the forming instructions performed by the forming machine 12 makes it possible to keep the hairpin conductor element 17 within the limits of the tolerance interval and, therefore, not to produce hairpin conductor elements 17 that will be discarded, or at least to limit them.

[0161] Conveniently, the intelligent agent 30 always generates a corrective action 31, even if the shape, geometry or measurements of the two hairpin conductor elements (the real hairpin conductor element 17 produced by the forming machine 12 and the reference (or master) hairpin conductor element) are substantially identical. Obviously, in this case, the corrective action 31 can have a substantially minimal and negligible effect, if not actually zero, on the forming parameters 13. In other words, in this case, the forming parameters 13 of the forming instructions performed by the forming machine 12 can remain substantially unchanged.

[0162] Please note that, in the embodiments comprising step 46, in which it is found that there is no discrepancy outside the tolerance interval between the two hairpin conductor elements (the real hairpin conductor element 17 produced by the forming machine 12 and the reference (or master) hairpin conductor element), and as mentioned above, in the case of finding this lack, the method returns to the start, in particular to the previous step 41, therefore "skipping" steps 47 and 48.

[0163] In the embodiments comprising steps 42 and 44, i.e. when the digital 3D reconstruction 19 is not reliable, i.e. in the presence of an anomalous signal of the digital 3D reconstruction 19 originating from said supervisor, the intelligent agent 30 of the processing device 20 does not generate any corrective action 31 of the forming parameters 13 of the forming instructions, therefore leaving the forming parameters 13 of the forming instructions performed by the forming machine 12 unchanged.

[0164] In an embodiment, the correction actions 31 of the shaping parameters 13 defined by the intelligent agent 30 in step 48 are preliminarily evaluated, i.e. before sending them to the shaping machine 12, and then, when the results based on the performed preliminary evaluation indicate that said correction actions 31 of the shaping parameters 13 are compliant with a predefined threshold band, these correction actions 31 of the shaping parameters 13 are forwarded to the shaping machine 12 by means of a correction action analyzer 33 comprised in the processing device 20.

[0165] If the results of the preliminary evaluation performed by the correction action analyzer 33 in the above-mentioned step are positive, i.e. when the shaping parameters 13 or the correction variations of the shaping parameters 13 remain within the predefined threshold band, and therefore the decision made by said correction action analyzer 33 is positive, the correction actions 31 of the shaping parameters 13 are forwarded to the shaping machine 12 by means of the correction action analyzer 33 of the processing device 20, and then the method returns to the start, in particular to the preceding step 41.

[0166] Conversely, if the results of the preliminary evaluation performed by the correction action analyzer 33 in the above-mentioned step are negative, i.e. when the shaping parameters 13 or the correction variations of the shaping parameters 13 fall outside the predefined threshold band, and therefore the decision made by said correction action analyzer 33 is negative, the correction actions 31 of the shaping parameters 13 are not forwarded to the shaping machine 12 by means of the correction action analyzer 33 of the processing device 20, and then the method returns to the start, in particular to the preceding step 41.

[0167] As mentioned above, in this case the shaping machine 12 does not receive any correction action 31 of the shaping parameters 13 of the shaping instructions, thus keeping the shaping parameters 13 of the shaping instructions performed by said shaping machine 12 unchanged.

[0168] As mentioned above, the above-mentioned predefined threshold band is defined so that the shaping parameters 13 or the correction variations of the shaping parameters 13 satisfy a suitable set of constraints given by the limits of the shaping process and / or of the shaping machine 12, for example the spatial occupation of the shaping tools, the correct interaction between the shaping tools and the electrical wires, etc.

[0169] With reference to Figures 6 to 10 Advantageously, the training of the system 10 for controlling the shaping machine for manufacturing the hairpin conductor elements of the induction winding of the stator, in particular the training of the intelligent agent 30 of the processing device 20, according to the present application, is performed by a training agent 76 associated with at least one digital twin 60 of the shaping machine 12, which will be described in detail below, wherein the digital twin 60 is configured to generate at least one plurality of simulated hairpin conductor elements 62, which can be used by the training agent 76 as data for training the intelligent agent 30.

[0170] Basically, in training the intelligent agent 30 of the processing device 20, the real forming machine 12 described above is replaced by a digital twin 60 of the forming machine.

[0171] Advantageously, the training data comprises a plurality of simulated forming parameters used by the digital forming machine implemented in the digital twin 60, which are different from each other and simulate real forming parameters that the real forming machine 12 can adopt, as will be described below with reference to Figure 9B the detailed description.

[0172] Advantageously, the training of the intelligent agent 30 of the processing device 20 is performed by the training agent 76 through training data generated by a plurality of digital twins 60 of the respective forming machines, each different from each other. This makes it possible to directly increase, in proportion to the number of digital twins 60 of the forming machines, the variability (or noise) of the data provided as input to the training agent 76 for training the intelligent agent 30.

[0173] As is known in the field of artificial intelligence, the greater the variability of the training data, the higher the quality of the training and the better the performance (in terms of effectiveness, efficiency, speed, general applicability, robustness, etc.) of the trained system.

[0174] In the present application, the above-mentioned variability (or noise) refers to the various operating conditions in which the forming machine can work. As mentioned above, the interference conditions external to the forming machine can include variations in the electrical wires or, more precisely, in the mechanical properties (tensile modulus, yield stress, etc.) of the corresponding material used, the elastic recovery effect or “springback” and variations in the environmental temperature. As mentioned above, the operating conditions internal to the forming machine can include clearances and variations due to wear, as well as variations in the dimensions of the forming tool (for example due to wear or due to changes in the tool).

[0175] Using the digital twin 60, it is possible to make the intelligent agent 30 able to interface with many and varied different operating conditions by simulating them and, therefore, exploring them through the training process, and to learn a strategy for correcting the forming parameters 13 of the forming instructions that is general, robust and applicable to all the operating conditions simulated and, therefore, explored through the training process.

[0176] Please note that the digital forming machine implemented in the digital twin 60 can (and usually is) different from the real forming machine 12 in terms of features, parameters, etc. In other words, usually, the digital forming machine implemented in the digital twin 60 is not an exact copy of the real forming machine 12. However, the forming process performed by the digital forming machine implemented in the digital twin 60 must in any case be similar to the forming process performed by the real forming machine 12.

[0177] The use of the digital twin 60 of the shaping machine has the advantage that the training of the virtualized system 10, in particular of the intelligent agent 30 of the processing device 20, is avoided, with consequent savings of resources, such as raw materials (wires, electricity), time, money, etc., for shaping a sufficient number of real hairpin conductor elements 17 using the real machine 12 to generate sufficient data (sufficient in terms of quantity and variability) for the training as described above.

[0178] Please note that the training of the intelligent agent 30 requires, preferably, training data of hundreds of thousands of hairpin conductor elements. In a preferred embodiment, the training of the intelligent agent 30 can be performed using training data comprising a combination of data about real hairpin conductor elements 17 and data about simulated hairpin conductor elements 62, in particular when this training is performed by means of an automatic technique (i.e. algorithm) of the actor-critic type of automatic reinforcement learning (RL) with replay buffer. In another embodiment, the training of the intelligent agent 30 can be performed using training data comprising only data about simulated hairpin conductor elements 62. In another embodiment, the training of the intelligent agent 30 can be performed using training data comprising only data about real hairpin conductor elements 17.

[0179] With reference to Figure 6 , the digital twin 60 is a digital model of a digital shaping machine of hairpin conductor elements of an induction winding of a stator. The digital twin 60 is configured by a plurality of parameters 13, 61 provided in input to said digital twin 60.

[0180] Please note that the parameters 13, 61, in particular the other parameters 61, of the digital shaping machine implemented in the digital twin 60 can (and typically are) different from the parameters of the real shaping machine 12. In other words, typically the parameters 13, 61, in particular the other parameters 61, of the digital shaping machine implemented in the digital twin 60 are not an exact copy of the parameters of the real shaping machine 12. However, since the shaping process performed by the digital shaping machine implemented in the digital twin 60 must in any case be similar to the shaping process performed by the real shaping machine 12, the parameters 13, 61 of the digital shaping machine implemented in the digital twin 60 must be similar to the parameters of the real shaping machine 12.

[0181] These parameters 13, 61 of the digital twin 60 configured shaping machine comprise shaping parameters 13 used in shaping instructions to be executed by said digital twin 60.

[0182] These parameters 13, 61 of the digital twin 60 configured shaping machine further comprise other parameters 61.

[0183] In particular, these further parameters 61 can comprise one or more parameters selected from the group consisting of: thickness of the electrical wire, mechanical properties of the electrical wire, geometry of the shaping machine (for example in CAD computer-aided design format), kinematics of the shaping machine, zero points of the shaping machine, interaction between tools of the shaping machine, physical constraints of the shaping process, real data, and torque dispensed or current drawn by the motors of the tools of the shaping machine.

[0184] By means of the digital twin 60, starting from the shaping parameters 13, one or more elements selected from the group consisting of: a plurality of simulated hairpin conductor elements 62, respective simulated three-dimensional (3D) measurements 63 of said simulated hairpin conductor elements 62, respective simulated three-dimensional (3D) geometry information 64 of said simulated hairpin conductor elements 62, and a plurality of data and information 65 related to the shaping process of the plurality of simulated hairpin conductor elements 62 can be obtained. The combination between the shaping parameters 13 and these elements 62, 63, 64, 65 is parameterized by the further parameters 61.

[0185] As mentioned above, the digital twin 60 is configured to generate a plurality of simulated hairpin conductor elements 62 which can be used by the training agent 76 as data for training the intelligent agent 30. In particular, the simulated hairpin conductor elements 62, or rather the respective shape or geometry, can be used directly as training data and / or indirectly for obtaining, generating, etc. training data.

[0186] The digital twin 60 of the shaping machine is configured to generate, in particular to virtually create and thus provide as output, at least one plurality (or rather a sequence, since they are generated one after the other) of simulated hairpin conductor elements 62. These simulated hairpin conductor elements 62 are hairpin conductor elements obtainable based on the parameters 13, 61 with which the digital twin 60 of the shaping machine is configured. In particular, the digital twin 60 is configured to generate the shape or geometry (for example in CAD computer-aided design format) of each simulated hairpin conductor element 62 output from said digital twin 60 and to provide this shape or geometry as output.

[0187] The shape or geometry of the simulated hairpin conductor elements 62 (wherein, as mentioned above, these two terms are understood as synonyms) can be included in the data used for training the intelligent agent 30. In other words, the training data provided as input to the training agent 76 can comprise the shape or geometry of the simulated hairpin conductor elements 62.

[0188] The digital twin 60 can also be configured to obtain, from the shape or geometry of the simulated hairpin conductor elements 62 described above, respective simulated three-dimensional (3D) measurements 63 of said simulated hairpin conductor elements 62.

[0189] A plurality or a sequence of simulated 3D measurements 63 of the shape or geometry of the simulated hairpin conductor element 62 can be included in the data used to train the intelligent agent 30. In other words, the training data provided as input to the training agent 76 can include a plurality or a sequence of simulated 3D measurements 63 of the shape or geometry of the simulated hairpin conductor element 62.

[0190] The digital twin 60 can also be configured to generate simulated geometric information 64 of the simulated hairpin conductor element 62 from the shape or geometry of the simulated hairpin conductor element 62 described above, for example a simulated digital three-dimensional (3D) reconstruction 64.

[0191] Basically, the simulated digital 3D reconstruction 64 is a simulation of the digital 3D reconstruction 19 of the shape or geometry of the hairpin conductor element 17 generated by the 3D vision device 18, the hairpin conductor element 17 being produced by the shaping machine 12 and thus output from the shaping machine.

[0192] Preferably, the simulated geometric information 64 of the simulated hairpin conductor element 62, for example the simulated digital 3D reconstruction 64, is a point cloud (point cloud data PCD).

[0193] A plurality or a sequence of simulated digital 3D reconstructions 64 of the shape or geometry of the simulated hairpin conductor element 62 can be included in the data used to train the intelligent agent 30. In other words, the training data provided as input to the training agent 76 can include a plurality or a sequence of simulated digital 3D reconstructions 64 of the shape or geometry of the simulated hairpin conductor element 62.

[0194] The digital twin 60 can also be configured to generate as output a plurality of data and information 65 about the shaping process of the plurality or of each of the simulated hairpin conductor elements 62.

[0195] In particular, the plurality of data and information 65 can include one or more data items and information items selected from the group consisting of: position and spatial occupancy of the tools (i.e. the volume or space occupied) during the simulated shaping, position of the wires during the simulated shaping, interaction between the tools during the simulated shaping, desired position for engagement during the simulated shaping, and undesired intersection between the tools and the wires during the simulated shaping.

[0196] The plurality of data items and information items 65 are data items and information items that are available with the parameters 13, 61 of the digital twin 60 configured shaping machine.

[0197] A plurality of data items and information items 65 related to the forming process of the plurality or sequence of simulated hairpin conductor elements 62 or each of them can be included in the data used to train the intelligent agent 30. In other words, the training data provided as input to the training agent 76 can include the plurality of data items and information items 65 related to the forming process of the plurality or sequence of simulated hairpin conductor elements 62 or each of them.

[0198] With reference to Figure 7 In an embodiment, the simulated hairpin conductor element 62 is developed by the digital twin 60 of the forming machine starting from the parameters 13, 61 configuring said digital twin 60. In other words, these parameters 13, 61 are considered as the basic source of information for the generation of the simulated hairpin conductor element 62 by the digital twin 60 of the forming machine.

[0199] As mentioned above, these parameters 13, 61 include the forming parameters 13 used in the forming instructions to be executed by said digital twin 60 and can further include other parameters 61.

[0200] In this embodiment, starting from the parameters 13, 61, and in particular from the above-mentioned forming parameters 13 of the forming instructions, the digital twin 60 is configured to obtain an ordered sequence of groups of forming parameters 67 i 、67 i+1 、67 N . Each group of forming parameters 67 i 、67 i+1 、67 N relates to, defines and enables the generation of a respective segment 69 i 、69 i+1 、69 N of the simulated hairpin conductor element 62. Subsequently, the digital twin 60 is configured to combine the ordered sequence of segments 69 i 、69 i+1 、69 N so as to completely produce the simulated hairpin conductor element 62.

[0201] The logic of the above-mentioned process is associated with the step-by-step nature of the forming process of any hairpin conductor element.

[0202] Each group of forming parameters 67 i 、67 i+1 、67 N is a subset of the forming parameters 13 of the forming instructions. Each group of forming parameters 67 i 、67 i+1 、67 N relates to the geometry of a respective segment 69 i 、69 i+1 、69 N of the simulated hairpin conductor element 62.

[0203] The various segments 69 of the simulated hairpin conductor element 62 i , 69 i+1 , 69 N The combination of the segments 69 i may be defined by a respective set of shaping parameters 67 i but can then be further modified by the presence of another segment 69 i+1 following it. Similarly, the segment 69 i+1 may be defined by a respective set of shaping parameters 67 i+1 but can then be further modified by the presence of another segment 69 i preceding it.

[0204] For example, a segment 69 i may be subjected to calendering and the subsequent segment 69 i+1 may be subjected to bending. In this case, the bending can be imposed on the segment of wire that has already been calendered, so this will affect the effect obtained on the segment 69 i+1 and previously obtained on the segment 69 i .

[0205] Optionally, starting from the parameters 13, 61, in particular from the above-mentioned shaping parameters 13 of the shaping instructions, the digital twin 60 is also configured to obtain, in addition to the ordered sequence of shaping parameter sets 67 i , 67 i+1 , 67 N a set of auxiliary parameters 66 related to the intermediate motion of the "virtual" tool of the digital twin 60 of the shaping machine.

[0206] Each set of auxiliary parameters 66 is a subset of the shaping parameters 13 of the shaping instructions. This set of auxiliary parameters 66 does not affect the geometry of the segments 69 i , 69 i+1 , 69 N of the simulated hairpin conductor element 62 but does affect the process of shaping the hairpin conductor element.

[0207] With reference to Figure 8 , in embodiments without variability (or noise), the system 10 for training a shaping machine for shaping hairpin conductor elements of an induction winding of a stator, according to the present application, in particular for training the corresponding intelligent agent 30 of the processing device 20, comprises the digital twin 60 of the above-mentioned shaping machine.

[0208] Please note that this embodiment of the training system according to the invention, lacking variability (or noise), is difficult to use in practice because it is not very universally applicable and robust. In fact, as described above, the training agent 76 is preferably shown, and in particular, an automated algorithm of the reinforcement learning (RL) type is shown for training the intelligent agent 30 under various operating conditions in which the forming machine can operate, so as to extrapolate a strategy for correcting the forming parameters 13 for the forming instructions from those conditions during the training process, a strategy that can be generalized across multiple operating conditions of the forming machine.

[0209] As described above, the digital twin 60 is defined and configured by a plurality of parameters 13, 61 provided in the input to the digital twin 60.

[0210] As described above, the digital twin 60 is configured to generate at least one, a plurality of, or a sequence of simulated hairpin conductor elements 62, and thus provides these simulated hairpin conductor elements as output. These simulated hairpin conductor elements 62 are hairpin conductor elements obtainable based on parameters 13, 61 of the digital twin 60 configured to form the machine. Specifically, the digital twin 60 is configured to generate geometric information (e.g., in a CAD computer-aided design format) for each simulated hairpin conductor element 62 output from the digital twin 60, and thus provides this geometric information as output.

[0211] As described above, the digital twin 60 can also be configured to obtain a corresponding simulated three-dimensional (3D) measurement result 63 of the simulated hairpin conductor element 62 from the shape or geometry of the simulated hairpin conductor element 62.

[0212] As described above, the digital twin 60 can also be configured to generate analog geometric information 64 of the analog hairpin conductor element 62 based on the shape or geometry of the analog hairpin conductor element 62, such as analog digital three-dimensional (3D) reconstruction 64.

[0213] Preferably, the analog geometric information 64 of the analog hairpin conductor element 62 (e.g., analog digital 3D reconstruction 64) is a point cloud (point cloud data PCD).

[0214] As described above, the digital twin 60 can also be configured to generate multiple data and information items 65 as outputs, which relate to the forming process of multiple or a sequence of simulated hairpin conductor elements 62 or each of them. These multiple data and information items 65 are data and information items obtainable using parameters 13, 61 of the digital twin 60 configured with the forming machine.

[0215] The system for training the system 10, in particular the corresponding intelligent agent 30 for training the processing device 20, according to the present application further comprises a state and reward generator 74 configured to generate both state information 29, e.g. comprising information about the simulated hairpin conductor element 62 on the digitally shaped machine implemented in the digital twin 60, and reward information 75. In particular, the state and reward generator 74 of the training system is configured to receive a set of information originating from various sources, including e.g. the digital twin 60, and then generate the state information 29 based on all this information (as is or processed). This state information 29 is accompanied by reward information 75. This reward information 75 is computed by the state and reward generator 74 based on the training data described above.

[0216] The reward is a numerical value defining the quality of the current situation of the shaping process of the lens assembly learned by the training agent 76 of the training system (to be described in detail below) and, thus, by the intelligent agent 30 of the system 10 for controlling the shaping machine.

[0217] In embodiments, the value of the reward information 75 can be defined by one or more elements selected from the group consisting of: a difference between the shape, geometry or measurements of the simulated hairpin conductor element 62 obtained and the shape, geometry or measurements of a reference (or master) hairpin conductor element; one or more constraints of the shaping process; and a shaping parameter 13 used in the shaping instructions to be executed.

[0218] The set of information received by the state and reward generator 74 of the training system can comprise one or more items of information selected from the group consisting of: a shape or geometry of the simulated hairpin conductor element 62 provided by the digital twin 60; a simulated 3D measurement 63 of the simulated hairpin conductor element 62 provided by the digital twin 60; simulated geometry information 64 of the simulated hairpin conductor element 62, e.g. a simulated digital 3D reconstruction 64, provided by the digital twin 60; a plurality of data items and information items 65 provided by the digital twin 60, relating to the shaping process of a plurality or a sequence of simulated hairpin conductor elements 62 or each of them; a shaping parameter 13 currently used in the shaping instructions to be executed by the digital twin 60; operational constraints (e.g. speed, torque, etc.) on the shaping provided by the digital twin 60; and context or background information 27, e.g. provided by a human operator using an adapted data input device, or provided by the shaping machine 12, e.g. recorded in a memory unit 34.

[0219] Therefore, the status information 29 about the simulated hairpin conductor element 62 can comprise one or more items of information selected from the group consisting of: a shape or geometry of the simulated hairpin conductor element 62 provided by the digital twin 60; a simulated 3D measurement 63 of the simulated hairpin conductor element 62 provided by the digital twin 60; a simulated digital 3D reconstruction 64 of the simulated hairpin conductor element 62 provided by the digital twin 60; a plurality of data items and information items 65 provided by the digital twin 60, which relate to a plurality or a sequence of simulated hairpin conductor elements 62 or to the shaping process of each of them; a shaping parameter 13 currently used in the shaping instructions to be executed by the digital twin 60; an operating constraint (e.g. speed, torque, etc.) to the shaping provided by the digital twin 60; and context or background information 27, for example provided by a human operator using an adapted data input device, or provided by the shaping machine 12, for example recorded in a memory unit 34.

[0220] As mentioned above, the context or background information 27 of the status information 29 can comprise one or more items of information selected from the group consisting of: reference geometry information about a reference (or master) hairpin conductor element, for example a digital 3D reconstruction of a shape or geometry of the reference (or master) hairpin conductor element, or data or measurements related to the reference (or master) hairpin conductor element; a type of the hairpin conductor element 17; a current diameter of a wire coil powering the shaping machine; and data about a collection basket of the hairpin conductor element 17.

[0221] As mentioned above, the status information 29 is accompanied by reward information 75.

[0222] The system for training the system 10, in particular the corresponding intelligent agent 30 of the training processing device 20, according to the present application, further comprises a training agent 76 associated with said intelligent agent 30, configured to process the state information 29 and the rewards 75 generated and provided by the state and reward generator 74, using an artificial intelligence algorithm (in short, AI algorithm), and to define and execute self-configuration or self-adjustment actions that will affect the shaping parameters 13 of the shaping instructions, (basically, the training agent 76 executes these actions on itself and, therefore, on the intelligent agent 30 that said training agent 76 is training), in order to recalibrate the digital twin 60 based on the corrective actions 31 generated by the intelligent agent 30. In short, the training agent 76 is configured to train the intelligent agent 30.

[0223] The purpose of the training agent 76 is to maximize the rewards 75, therefore the above-mentioned self-configuration or self-adjustment actions are defined with the purpose of maximizing the rewards 75. In other words, the training agent 76 defines and executes the above-mentioned self-configuration or self-adjustment actions in order to maximize at least one objective function (or cost function) based on the rewards 75, that is to say, on the rewards 75 obtained in each episode.

[0224] In an embodiment, the training agent 76 is a fully connected neural network (FCNN) and a feedforward neural network (FFNN). In another embodiment, the training agent 76 is a recurrent neural network (RNN). Advantageously, the training agent 76 implements the techniques, i.e. algorithms, of automatic reinforcement learning (RL), preferably of the actor critic type, preferably with a replay buffer. In other words, advantageously, the training agent 76 trains the intelligent agent 30 using the techniques, i.e. algorithms, of automatic reinforcement learning (RL), preferably of the actor critic type, preferably with a replay buffer.

[0225] As mentioned above, the greater the variability of the training data, the better the quality of the training and the better the performance (in terms of effectiveness, efficiency, speed, general applicability, robustness, etc.) of the training system.

[0226] Reference Figure 9A In the training of the system for controlling the shaping machine of the hairpin conductor elements of the induction winding of the stator, according to the present application, it is possible to increase the variability (or noise) of the training data by defining and configuring a plurality of digital twins 60, each different from each other.

[0227] Advantageously, starting from the shaping parameters 13 used in the shaping instructions to be executed, and from the starting or base digital twin 60 configured to generate the corresponding plurality of simulated hairpin conductor elements 62, other digital twins 60i, 60j configured to generate a corresponding plurality of simulated hairpin conductor elements 62i, 62j can be defined, in particular by varying the corresponding other parameters 61 and thus the corresponding shaping features i, j.

[0228] Basically, among all the possible values of the other parameters 61, a set can be identified which can be referred to as a starting or base set and which defines the starting or base digital twin 60. Then, said other parameters 61 are varied while still keeping them in the neighborhood of the other starting parameters 61 in order to define the other digital twins 60i, 60j.

[0229] In other words, given the starting or base set of other parameters 61, the digital shaper implemented in the starting or base digital twin 60 simulates a specific shaping process. Obviously, the digital shaper implemented in the digital twin 60 can be used to generate, in particular to virtually create, simulated hairpin conductor elements 62 having many different shapes, with the aim of training the intelligent agent 30 with respect to all these shapes, however, which depend on the specific conditions of the shaping process.

[0230] Advantageously, in order to increase the general applicability of the training of the intelligent agent 30, the set of other parameters 61 is varied, thus creating variability (or noise). Therefore, a plurality of possible scenarios of the shaping process can be explored, in which the digital twins 60i, 60j receive the same machine parameters 13 and the corresponding other varied parameters 61 as input.

[0231] Leaving the other parameters 61 unchanged results in training an intelligent agent 30 able to effectively correct the shaping process only in specific conditions. By contrast, expediently, the other parameters 61 are varied during the training process, thus creating variability (or noise), so that the training agent 76 interacts with all possible scenarios of the shaping process and trains the intelligent agent 30 to be able to effectively correct the shaping process in many conditions. In this way, a more general and robust control system 10 is obtained.

[0232] With reference to Figure 9B , the plurality of digital twins 82 on which the training is based (each digital twin being different from each other) extends along a portion of the domain of the existing shaper 81. Within the plurality of digital twins 82, and thus within the domain of the existing shaper 81, both the starting or base digital twin 83 and the real shaper 84 are located.

[0233] As mentioned above, the shaping process performed by the shaper implemented in the digital twin 82, i.e. the simulated shaping process, must in any case be similar to the shaping process performed by the real shaper 84.

[0234] Figure 9B is a graphical representation of the importance of the similarity of the shaping process, i.e. the shaping process performed by the shaping machines of the plurality of digital twins 82 (i.e. the simulated shaping processes observed during the training process) is similar to the shaping process performed by the real shaping machine 84 and thus close together along the domain of the existing shaping machines 81. Similarly, Figure 9B is a graphical representation of the importance of the closeness between the starting or base digital twin 83 and the real shaping machine 84. At the same time, Figure 9B is a graphical representation of the importance of the range of the plurality of digital twins 82, which is wide enough to include variations of the shaping process of the real shaping machine 84 and thus movements along the domain of the existing shaping machines 81.

[0235] The ability of the intelligent agent 30 to adapt itself or rather the shaping parameters 13 of the shaping instructions to variations of the shaping process performed by the real shaping machine is obtained by interacting the training agent 76 with the plurality of digital twins 82 having variable (varying with each training epoch) parameters 13, 61, in particular the other parameters 61, as previously described, the training agent 76 can train the intelligent agent 30 using an actor-critic type of automated reinforcement learning algorithm, thus simulating the shaping process performed by the shaping machines of the plurality of digital twins 82.

[0236] Basically, through the training process, the intelligent agent 30 can be taught a correction strategy for the shaping parameters 13 of the shaping instructions, which is scalable and thus adaptable to the plurality of digital twins 82.

[0237] Note that as the range of the plurality of digital twins 82 increases, the general applicability and robustness of the correction strategy for the shaping parameters 13 of the shaping instructions learned by the intelligent agent 30 during the training process also increases, but there is a risk of loss of efficiency in terms of correction performance (e.g. three correction steps are needed instead of two to correct a hairpin conductor element).

[0238] With reference to Figure 10 , in the embodiment with variability (or noise), the system for training a system 10 for controlling a shaping machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application is substantially the same as the system shown in and described with reference to Figure 8 , except for the fact that the system comprises a plurality of digital twins 60 of the shaping machine, each of which is different from each other. Figure 8

[0239] ​In practice, it has been found that the present application achieves the intended aim and objects in a wholly satisfactory manner. In particular, the system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator, conceived as such, make it possible to overcome the qualitative limits of the known art, since they make it possible to obtain better results and / or similar results to those obtained by the conventional solutions, at lower costs and with higher performance levels.

[0240] The system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are advantageous in that they make it possible to compensate for variations in the forming process of the forming machine, i.e. variations in the forming process and / or in the operation of the forming machine over time, which are caused, for example, by interference conditions and / or working conditions.

[0241] The system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are further advantageous in that they make the operation of the forming machine, i.e. the operation in the forming process performed by it, adaptable to interference conditions external to the forming machine, such as, for example, variations in the mechanical properties (tensile modulus, yield stress, etc.) of the electrical wire or, more precisely, of the relative material used, variations in the rebound and in the ambient temperature.

[0242] The system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are further advantageous in that they make it possible to adapt the operation of the forming machine, i.e. the operation in the forming process performed by it, to working conditions internal to the forming machine, such as, for example, clearances and variations due to wear.

[0243] For example, the system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are able to deal with variations in the forming process that can exist between different forming machines of the same model, or wear of the tools of the forming machine that changes the operation of the latter over time, or variability of the material that constitutes the electrical wire for manufacturing the hairpin conductor elements.

[0244] The system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are further advantageous in that they make the forming process, and in particular the operation performed by the forming machine, independent of the ability and / or the conditions of the human operator, thus passing from a subjective check to an objective check, which produces predictable and repeatable results.

[0245] The system and the method for controlling a forming machine for manufacturing hairpin conductor elements of an induction winding of a stator according to the present application are further advantageous in that they make it possible to eliminate or at least minimize the reaction time after the first hairpin with a measurement outside the tolerance interval is produced.

[0246] Another advantage of the system and method for controlling a forming machine for making hairpin conductor elements of an induction winding of a stator according to the present application is that they make it possible to correct at least one measurement of the hairpin conductor elements by simultaneously modifying a plurality of forming parameters of the forming instructions to be set in the forming machine using artificial intelligence algorithms.

[0247] Another advantage of the system and method for controlling a forming machine for making hairpin conductor elements of an induction winding of a stator according to the present application is that they make it possible to comfortably keep the forming machine within the corresponding forming operating limits, which can make its operation unstable and not very robust.

[0248] Another advantage of the system and method for controlling a forming machine for making hairpin conductor elements of an induction winding of a stator according to the present application is that they make it possible to provide greater stability for the forming process, in particular for the operations performed by the forming machine.

[0249] The application thus conceived is susceptible to numerous modifications and variants, all falling within the scope of the following claims.

[0250] Unless otherwise indicated, the various embodiments described above can be combined to provide additional and / or alternative embodiments. Moreover, this description covers combinations of variants and preferred embodiments not explicitly described.

[0251] Furthermore, all the details can be replaced by other, technically equivalent elements.

[0252] In practice, the materials used, so long as they are compatible with the specific use, as well as the contingent dimensions and shapes, can be any according to the requirements and the state of the art.

[0253] In general, the scope of the claims is not limited to the preferred embodiments described above, but is given by the following claims, which are to be interpreted in the light of the above teaching.

[0254] The disclosures in the Italian Patent Application No. 102023000008631 from which this application claims priority are incorporated herein by reference.

[0255] Where technical features mentioned in any claim are followed by reference signs, those reference signs are included for the sole purpose of increasing the intelligibility of the claims and accordingly such reference signs do not have any limiting effect on the interpretation of each element identified by way of that reference sign.

Claims

1. A system (10) for controlling a forming machine (12) for manufacturing hairpin conductor elements (17) of an induction winding of a stator, the system comprising a source (18) of geometric information of at least one hairpin conductor element (17) obtained with the forming machine (12) and a processing device (20) operatively connected to the source (18) of geometric information and connectable to the forming machine (12); wherein the source (18) of geometric information being configured to provide current geometric information (19) of the at least one hairpin conductor element (17), and the processing device (20) comprising: a state generator (24) configured to receive a set of information comprising at least the current geometric information (19) of the hairpin conductor element (17) provided by the source (18) of geometric information, reference geometric information of a reference hairpin conductor element, and forming parameters (13) used by the forming machine (12), and further configured to generate state information (29) based on the received set of information; and an intelligent agent (30) configured to process the state information (29) using artificial intelligence (AI) algorithms and to autonomously generate a corrective action (31) for the forming parameters (13); so that the processing device (20) adjusts the forming parameters (13) based on the corrective action (31) generated by the intelligent agent (30).

2. The system (10) of claim 1, wherein, the source (18) of geometric information is a 3D vision device (18), and wherein the current geometric information (19) is a digital 3D reconstruction (19) of the hairpin conductor element (17).

3. The system (10) of claim 2, wherein, the digital 3D reconstruction (19) of the hairpin conductor element (17) generated by the 3D vision device (18) is a point cloud.

4. The system (10) according to claim 2 or 3, wherein, the 3D vision device (18) is operatively connected to a supervisor configured to preliminarily verify the reliability of the digital 3D reconstruction (19) of the hairpin conductor element (17), the 3D vision device (18) being configured to send the digital 3D reconstruction (19) to the processing device (20) when the digital 3D reconstruction (19) is reliable.

5. The system (10) of claim 4, wherein, the supervisor of the 3D vision device (18) is further configured to send an anomaly signal of the digital 3D reconstruction (19) to the processing device (20) when the digital 3D reconstruction (19) is not reliable.

6. The system (10) according to claim 4 or 5, wherein the intelligent agent (30) of the processing device (20) is configured to keep the forming parameters (13) of the forming machine (12) unchanged when the digital 3D reconstruction (19) is not reliable.

7. The system (10) according to any one of the preceding claims, wherein, the state generator (24) of the processing device (20) comprises an instrument (26) configured to obtain at least one measurement of the hairpin conductor element (17) from the current geometric information (19) of the hairpin conductor element (17).

8. The system (10) according to any one of the preceding claims, wherein, The processing device (20) further comprises a state analyzer (32) configured to preliminarily evaluate the state information (29) and to forward the evaluated state information (29) to the intelligent agent (30) when at least one corrective action (31) is required.

9. The system (10) of claim 8, wherein, The state analyzer (32) of the processing device (20) is further configured to compare the current geometric information (19) of the hairpin conductor element (17) with the reference geometric information of the reference hairpin conductor element to identify any discrepancy.

10. The system (10) according to any one of the preceding claims, wherein, The processing device (20) further comprises a corrective action analyzer (33) configured to preliminarily evaluate the corrective action (31) of the shaping parameters (13) and to forward the corrective action (31) to the shaper (12) when the corrective action (31) complies with a predefined threshold band.

11. The system (10) according to any one of the preceding claims, wherein, The intelligent agent (30) of the processing device (20) is trained by means of a training agent (76) associated with at least one digital twin (60) of the shaper (12), the digital twin (60) being a digital model of a digital shaper of hairpin conductor elements of an induction winding of a stator, the digital twin (60) being configured to generate at least a plurality of simulated hairpin conductor elements (62) that can be used as training data for the training agent (76).

12. The system (10) of claim 11, wherein, The training data comprise a plurality of simulated shaping parameters used by the digital shaper of the digital twin (60), the simulated shaping parameters being different from each other and simulating real shaping parameters that can be adopted by the real shaper (12).

13. The system (10) according to claim 11 or 12, wherein, The digital twin (60) is further configured to obtain simulated 3D measurements (63) of each simulated hairpin conductor element (62), the training data comprising the simulated 3D measurements (63).

14. The system (10) according to any one of claims 11 to 13, wherein, The digital twin (60) is further configured to generate simulated digital 3D reconstructions (64) of each simulated hairpin conductor element (62), the training data comprising the simulated digital 3D reconstructions (64).

15. A shaper (12) for manufacturing hairpin conductor elements (17) of an induction winding of a stator, wherein the shaper (12) comprises shaping means (16) configured to shape an electrical wire into a hairpin conductor element (17), and a control system (10) according to any one of claims 1 to 14.

16. A method for controlling a shaper (12) for manufacturing hairpin conductor elements (17) of an induction winding of a stator, the method being controlled by means of a geometric information source (18) of at least one hairpin conductor element (17) and a processing device (20) operatively connected to the geometric information source (18) and connectable to the shaper (12), the method comprising the steps of: providing (41), by means of the geometric information source (18), current geometric information (19) of the at least one hairpin conductor element (17) obtained with the shaper (12); generating (45), by a state generator (24) of the processing device (20), state information (29) based on the received set of information; and processing (48), by an intelligent agent (30) of the processing device (20), the state information (29) using artificial intelligence (AI) algorithms and autonomously generating (48) a corrective action (31) for the shaping parameters (13); so that the processing device (20) adjusts the shaping parameters (13) based on the corrective action (31) generated by the intelligent agent (30).

17. The method of claim 16, wherein, The geometric information source (18) is a 3D vision device (18) and wherein the current geometric information (19) is a digital 3D reconstruction (19) of the hairpin conductor element (17).

18. The method of claim 17, wherein, The digital 3D reconstruction (19) of the hairpin conductor element (17) generated by the 3D vision device (18) at the providing step (41) is a point cloud.

19. The method according to claim 17 or 18, further comprising the steps of: preliminarily verifying (42), by a supervisor comprised in the 3D vision device (18), the reliability of the digital 3D reconstruction (19) of the hairpin conductor element (17); and when the digital 3D reconstruction (19) is reliable, sending (43), by the 3D vision device (18), the digital 3D reconstruction (19) to the processing device (20).

20. The method of claim 19, further comprising the step of: when the digital 3D reconstruction (19) is not reliable, sending (44), by the supervisor of the 3D vision device (18), an anomaly signal of the digital 3D reconstruction (19) to the processing device (20).

21. The method according to claim 19 or 20, further comprising the step of keeping the shaping parameters (13) unchanged when the digital 3D reconstruction (19) is not reliable.

22. The method of any one of claims 16 to 21, wherein, The steps of receiving (45) the set of information and generating (45) the state information (29) comprise the step of obtaining, by a gauge (26) comprised in the state generator (24) of the processing device (20), at least one measurement of the hairpin conductor element (17) from the current geometric information (19) of the hairpin conductor element (17).

23. The method according to any one of claims 16 to 22, further comprising the step of: preliminarily evaluating (46) the state information (29) and, when at least one corrective action (31) is needed, forwarding (46) the evaluated state information (29) to the intelligent agent (30) by a state analyzer (32) of the processing device (20).

24. The method of claim 23, wherein, The step of preliminary evaluation (46) of the state information (29) and decision (46) whether to forward the state information (29) to the intelligent agent (30) comprises the step of comparing, by the state analyzer (32) of the processing device (20), the current geometric information (19) of the hairpin conductor element (17) with the reference geometric information of the reference hairpin conductor element to identify any discrepancy.

25. The method according to any one of claims 16 to 24, further comprising the step of preliminary evaluation of the corrective action (31) of the shaping parameters (13) and forwarding, by a corrective action analyzer (33) of the processing device (20), the corrective action (31) to the shaper (12) when the corrective action (31) complies with a predefined threshold band.

26. The method of any one of claims 16 to 25, further comprising the step of: The intelligent agent (30) of the processing device (20) is trained by a training agent (76) associated with at least one digital twin (60) of the shaper (12), the digital twin (60) being a digital model of a digital shaper of hairpin conductor elements of an induction winding of a stator, the digital twin (60) being configured to generate at least a plurality of simulated hairpin conductor elements (62) that can be used as training data for the training agent (76).

27. The method of claim 26, wherein, The training data comprises a plurality of simulated shaping parameters used by the digital shaper of the digital twin (60), the simulated shaping parameters being different from each other and simulating real shaping parameters that can be employed by the real shaper (12).

28. A system for training a system (10) for controlling a shaper (12) for manufacturing hairpin conductor elements (17) of an induction winding of a stator according to any one of claims 1 to 14, the system for training comprising: at least one digital twin (60) of the shaper (12) configured to generate at least a plurality of simulated hairpin conductor elements (62) and to provide simulated geometric information (64) of each of the simulated hairpin conductor elements (62); a state and reward generator (74) configured to receive a set of information comprising at least the simulated geometric information (64) of the simulated hairpin conductor elements (62), reference geometric information of a reference hairpin conductor element, and shaping parameters (13) used by the digital twin (60), and further configured to generate state information (29) and reward information (75) based on the received information; and a training agent (76) associated with the intelligent agent (30) of the control system (10) and configured to use an artificial intelligence (AI) algorithm, process the state information (29) and the reward information (75), and define and execute self-regulating actions of the shaping parameters (13) in order to recalibrate the digital twin (60) based on corrective actions (31) generated by the intelligent agent (30). The step of preliminary evaluation (46) of the state information (29) and decision (46) whether to forward the state information (29) to the intelligent agent (30) comprises the step of comparing, by the state analyzer (32) of the processing device (20), the current geometric information (19) of the hairpin conductor element (17) with the reference geometric information of the reference hairpin conductor element to identify any discrepancy.

25. The method according to any one of claims 16 to 24, further comprising the step of preliminary evaluation of the corrective action (31) of the shaping parameters (13) and forwarding, by a corrective action analyzer (33) of the processing device (20), the corrective action (31) to the shaper (12) when the corrective action (31) complies with a predefined threshold band. The intelligent agent (30) of the processing device (20) is trained by a training agent (76) associated with at least one digital twin (60) of the shaper (12), the digital twin (60) being a digital model of a digital shaper of hairpin conductor elements of an induction winding of a stator, the digital twin (60) being configured to generate at least a plurality of simulated hairpin conductor elements (62) that can be used as training data for the training agent (76). The training data comprises a plurality of simulated shaping parameters used by the digital shaper of the digital twin (60), the simulated shaping parameters being different from each other and simulating real shaping parameters that can be employed by the real shaper (12).

28. A system for training a system (10) for controlling a shaper (12) for manufacturing hairpin conductor elements (17) of an induction winding of a stator according to any one of claims 1 to 14, the system for training comprising: at least one digital twin (60) of the shaper (12) configured to generate at least a plurality of simulated hairpin conductor elements (62) and to provide simulated geometric information (64) of each of the simulated hairpin conductor elements (62); a state and reward generator (74) configured to receive a set of information comprising at least the simulated geometric information (64) of the simulated hairpin conductor elements (62), reference geometric information of a reference hairpin conductor element, and shaping parameters (13) used by the digital twin (60), and further configured to generate state information (29) and reward information (75) based on the received information; and a training agent (76) associated with the intelligent agent (30) of the control system (10) and configured to use an artificial intelligence (AI) algorithm, process the state information (29) and the reward information (75), and define and execute self-regulating actions of the shaping parameters (13) in order to recalibrate the digital twin (60) based on corrective actions (31) generated by the intelligent agent (30).

29. A method for training a system (10) for controlling a forming machine (12) for manufacturing hairpin conductor elements (17) of the induction winding of a stator according to any one of claims 1 to 14, comprising the steps of: generating, by means of at least one digital twin (60) of the forming machine (12), at least a plurality of simulated hairpin conductor elements (62) and providing simulation geometry information (64) of each of the simulated hairpin conductor elements (62); receiving, by means of a state and reward generator (74), a set of information comprising at least the simulation geometry information (64) of the simulated hairpin conductor elements (62), reference geometry information of reference hairpin conductor elements, and forming parameters (13) used in the digital twin (60), and further generating state information (29) and reward information (75) based on the received information; and processing, by means of a training agent (76) associated with the intelligent agent (30) of the control system (10), the state information (29) and the reward information (75) using an artificial intelligence (AI) algorithm and defining and executing self-regulating actions of the forming parameters (13) in order to recalibrate the digital twin (60) based on corrective actions (31) generated by the intelligent agent (30).

Images