DESCRIPTION

Field of the invention

The present invention relates to a method of controlling a forging system.

Namely, the invention relates to a control method aimed at optimizing an incremental open-die forging system.

The invention also relates to an automated incremental open-die forging system. Description of the Prior Art

As it is widely known in the art, an incremental open-die forging system allows a hot workpiece that comes from a furnace to be forged by the control of a mampulator and a press or hammering. One or more operators cause the work pieces to be forged, which usually have an elongate shape, such as railroad axles, gripped by the manipulator to be fed and rotated under the press.

In incremental open-die forging, the hot workpiece goes into a press with so- called "cogging" dies, it's partially pressed at each action of the press and displaced by the manipulator to assume preset dimensions in the various areas being processed.

In manually controlled forging process, the workpiece was rotated and displaced by the manipulator, which is controlled by the operator aside manual dimensional measurement. Particularly, during the work cycles the operator uses mechanical and optical references for real-time selection of the actions that are most suitable for obtaining the preset dimensions in the various processed areas of the workpiece.

While the manually controlled forging process provides satisfactory results in certain respects, it still suffers from a few drawbacks.

The operators should be proficient in the forging techniques, to prepare the subsequent pressing or feeding actions in very short times, whereby the quality of the finished workpiece largely depends on the skills and expertise of each operator.

The control manual of the mampulator and the press generally leads to finished workpieces with dimensional inaccuracy.

The operator is required to find the dimensions of each portion during the working cycle in difficult situation, due to both the short time he/she is allowed, from 1 to 3 seconds, and for the available means, which often consist of optical targets. To avoid the error of creating undersized portions or areas, which might irreversibly affect the quality of the finished workpiece, the operator would safely maintain an oversize as compared with the required dimensions, e Because of undersized areas could not be recovered in later processing. Moreover the workpiece should be discarded, in which manufacturing cost would be increase.

Nevertheless, oversizing leads to metal allowances that involve an economic burden in terms of time and cost increases, due to both the use of excess material and to the later steps, e.g. a turning step, required to reduce the finished workpiece to the preset required dimensions.

A further drawback is that hot-working of workpieces and the various corrections required on such workpieces would lead to torque stresses on the material, which would eventually increase the probability of cracking.

Therefore, the workpieces obtained heterogeneous characteristics.

In an attempt to avoid these drawbacks, a prior art solution is a semi- automatically controlled forging system, which has some operation steps with fixed and automated actions. By allowing the execution of programmed repetitive action cycles during the forging process.

Although this arrangement fulfills its purpose, it still has a few drawbacks.

The repetitive action cycles are fixed and automated by programmed parameters that do not account for the characteristics and special deformations of the workpiece to be forced, which are individual and depend on a few unforeseeable parameters, e.g. changes in the raw material, workpiece shapes obtained in particular working steps, workpiece temperature, ambient temperature and other parameters.

Therefore, these semi-automatic systems still require the presence of an operator who would manually assess the working process and make real-time changes to the system during the operation steps.

Thus, semi-automatic systems also require the presence of the operator and the workpieces obtained thereby still suffer from the above mentioned drawbacks, such as the provision of excessive metal allowances to prevent the formation of undersized areas, which would cause the workpiece to be discarded In the light of the above prior art, one object of the present invention is to provide workpieces with optimized size and property consistency.

Another object is to provide an automated control method that allows the quality of the finished workpieces to be unrelated and independent skills of the operator.

A further object is to provide a control method that can be implemented through a set of functional rules.

A further object is to provide a forging process that can be carried out in a shorter time than it would be required with the presence of an operator.

Yet another object is to provide a system that can be easily manufactured with a simple and contact structure.

The technical problem underlying the present invention is that providing a control method and a system that allow the operation steps of the forging process to be adapted into the actual conditions of the workpiece to be formed, and which has such functional and structural characteristics as to obviate the drawbacks that still affect prior art methods and systems.

Summary of the invention

The given solution idea underlying the present invention is automatically controlling the manipulator and the press according to the actual dimensions of the workpiece to be forged in each operation step of the forging process.

On the base of such kind of solution idea, the technical problem is solved by an open- die forging method that comprises at least one actuator for forging a blank, and to define a workpiece with the required dimensions by an operation sequence of blank processing operation steps, characterized in that it comprises the steps as follows:

- generating, for each operation step, respective intermediate end dimensions for the blank, as required at the end of that operation step;

- iteratively repeating, for each operation step of the operation sequence, the following steps:

- detecting the blank using at least one electromagnetic infrared radiation detection means, and generating a dimensional map of the blank, by analyzing the electromagnetic radiation received by the detection means;

- estimating the dimensions of the blank, using an artificial intelligence system, as a function of parameters of the dimensional map and further preset parameters of the artificial intelligence system, to define estimated dimensions of the blank;

- determining a reference deviation map, as a difference between the estimated dimensions of the blank and the intermediate end dimensions required by the operation step;

- adapting the artificial intelligence system as an error function that is designed to reduce the deviation map and determine at least one control signal to control the at least one actuator.

Advantageously, the step of detecting the blank also allows generation of a thermal map of the blank, by analysis of the electromagnetic radiation received by said at least one detection means which allows, in certain cases, further controls to be made on the blank and possible further specific controls to be made on the actuators.

Conveniently, the method includes the determination of a chronological series of operation steps, that define the operation sequence according to the reference deviation map, thereby allowing optimization of forging times.

Also, the chronological series of operation steps is a dynamic series of operation steps, which is stored in the artificial intelligence system and is controlled at the end of each operation step, thereby particularly affording improvement of process times.

Advantageously, in the method the artificial intelligence system is equipped with training means, to carry out a set of predetermined actions for each operation step, each action of the set comprising predetermined sequences of control signals. Particularly, the training means include training maps having predetermined rules and/or algorithms of operation sequences which are adapted to define such actions to generate effective and efficient operation actions.

Conveniently, in the method the artificial intelligence system has one or more of the logics selected from the group comprising: mathematical logics, statistical logics, artificial neural networks, genetic networks, fuzzy logics; also, the electromagnetic infrared radiation detection means is obtained by connecting an array of IR cameras to a visual processing module, the visual module being adapted to analyze the signals received from the array of IR cameras and generate the dimensional map and/or the thermal map. Furthermore, in the method the artificial intelligence system is equipped with a control unit which receives the dimensional map in real time and processes the values therein using an artificial intelligence algorithm, to estimate the dimensions and further characteristics of the blank.

The problem is also solved by an open-die forging system that comprises at least one actuator for forging a blank, to define a workpiece with the required dimensions by an operation sequence of blank processing operation steps, characterized in that it comprises:

- at least one detection means, which is adapted to detect the electromagnetic infrared radiation emitted by the blank and to generate a dimensional map of the blank;

- an artificial intelligence system, which is connected to such at least one detection means and is adapted to estimate the dimensions of the blank, by analyzing the values in the dimensional map and further preset parameters in the artificial intelligence system, the artificial intelligence system comprising a control unit, which is adapted to determine a deviation map as a difference between the estimated dimensions of the blank and predetermined intermediate end dimensions, as required by each operation step of the operation sequence;

- the control unit being also adapted to provide the artificial intelligence system as an error function that is designed to reduce the deviation map and determine at least one control signal to control the at least one actuator.

Advantageously, the detection means include an array of infrared radiation cameras connected to a visual module, which is adapted to generate the dimensional map by processing the signals received from the infrared cameras, the control unit being connected to the visual module and being adapted to process the dimensional map to define the reference deviation map. This allows to control in a precise mode each step of the operative sequence.

Conveniently, the control unit comprises an artificial intelligence algorithm which is used to estimate the blank dimensions during each processing step, and is adapted to determine the at least one control signal adapted to control the at least one actuator, thereby allowing constant monitoring of the blank being processed.

Furthermore, the at least one actuator comprises a manipulator and a press and the at least one control signal comprises a first control signal and a second control signal, which are adapted to control the manipulator and the press respectively, for each operation step of the operation sequence.

The artificial intelligence system further comprises:

- a first module, connected to the control unit and adapted to store the operation sequence;

- a second module, connected to the control unit and comprising operational behavior and rule training maps and a third module connected to the control unit and comprising algorithms of operation sequences, the second and third modules being adapted to define successive actions of the operation steps, each action being defined by sequences of such control signals;

- a fourth module, connected to the control unit and comprising the preset parameters.

The characteristics and advantages of the present invention will appear more clearly from the following description of one embodiment, thereof given by way of indicative and non limiting example with reference to the annexed drawings, in which:

- Figure 1 shows a side view of a forging system according to the invention;

- Figure 2 is a schematic perspective view of a section of a workpiece to be forged, gripped by the jaws of a press;

- Figure 3 is a schematic top view of the system of Figure 1 , showing the visual cones of a plurality of detection means;

- Figure 4 shows three different visual detections of a workpiece to be forged; Fig. 5 shows a block diagram of the system of the present invention.

Detailed description

Figure 1 shows an incremental open-die forging system 1, according to the present invention, which comprises at least two main actuators 2 and 3 for forging a blank 4 to obtain a workpiece having the required dimensions.

In the present embodiment, the main actuators are a manipulator 2 and a press 3, which are controlled by first control signals I l and second control signals In2 respectively, to operate in an automatic and synchronized mode.

The forging system 1 may also include secondary actuators, which assist the main actuators and do not directly interfere with the forging process. In the method, the secondary actuators are controlled by further control signals.

The forging process comprises an operation sequence OS including a series of operation steps for processing the blank 4, which start from a step in which the blank 4 is initially centered and gripped, and end with a step in which it is released, as the required dimensions are obtained, with multiple working steps intervening therebetween.

Each operation step of the series includes a set of actions, each action being executed by the control of main or secondary actuators, and through the generation of first or second control signals Inl and In2 or the generation of further control signals.

The control method of the present invention uses an artificial intelligence system

10 and provides storage of the operation sequence OS and the working steps of the forging system 1 in the artificial intelligence system 10.

Therefore, the method includes storage of a set of actions for each corresponding operation step of the series.

Also, the method conveniently provides, for each working operation step of the operation sequence OS, a set of intermediate end dimensions that the blank 4 is designed to have at the end of the operation step, with possible tolerances.

For each working operation step of the operation sequence OS, the control method of the present invention provides iterative repetition of the steps of:

- detecting the blank 4 using the electromagnetic infrared radiation detection means 8, and generating a dimensional map Ml and possibly a thermal map M2 of the blank 4, by analyzing the electromagnetic radiation received by the detection means 8;

- estimating the dimensions of the blank 4, as results of a function of the artificial intelligence system 10, which uses the parameters in the dimensional map Ml and possibly the thermal map M2 that have been detected, and preset parameters PGI of the artificial intelligence system 10, to define estimated dimensions of such blank 4.

Thus, this method includes the steps of:

- determining a deviation map E as a difference between the estimated dimensions of the blank 4 and the set of preset intermediate end dimensions associated with the current operation step;

- deteraiining, by the artificial intelligence system 10, an ordered set of actions for the current processing operation step and generating the first control signal Inl and/or the second control signal In2 respectively, for controlling the manipulator 2 and/or the press 3 according to such ordered set of actions.

Therefore, the control system includes the step of:

- adapting the artificial intelligence system 10 as an error function, to reduce the intermediate deviation map E.

Conveniently, the artificial intelligence system 10 may change the ordered set of actions for the current processing operation step, and possibly the chronological series of operation steps that define the operation sequence OS of the forging system 1 according to the reference deviation map E.

Particularly, the series of operation steps is a dynamic series, i.e. a chronological list of operation steps that is stored in the artificial intelligence system 10 and is iteratively controlled by the artificial intelligence system 10 at the end of each operation step, or controlled in case of error conditions or detected emergency steps. This will considerably shorten the processing times and reduce any errors in the intermediate end dimensions of the workpiece at the end of each operation step of the forging process.

Conveniently, in the method of the present invention, the artificial intelligence system 10 is equipped with training maps STM (Sequence Training Maps), which in turn include preset rules 15 for executing preset actions for each operation step of the forging process. In particolare, le azioni di detto gruppo di azioni comprendono sequenze predefinire di primi segnali di comando Inl e secondi segnali di comando In2.

The artificial intelligence system 10 further has operation sequence algorithms SA, which are defined by computational algorithms of operation steps of the forging process, and are used instead of the training maps STM. Particularly, the sequence algorithms SA are used, for instance, in simplified operation steps, e.g. the operation steps that do not require fulfillment of preset rules 15.

The preset parameters PGI are the values of the required dimensions as determined from technical drawings and other data concerning the workpiece, as well as the dimensions of the dies in use. These values are appropriately stored to be used by the artificial intelligence system 10 for making the finished workpiece.

Furthermore, in the method the artificial intelligence system 10 incorporates different types of logics such as: mathematical logics, statistical logics, artificial neural networks, genetic networks and fuzzy logics, which allow time-dependent values to be defined according to the self-training defined by the artificial intelligence system 10 itself.

In one embodiment of the method, the electromagnetic infrared radiation detection means 8 are obtained by connecting an array of IR cameras 9a-9f to a visual processing module VSM, which is adapted to generate the dimensional map Ml and possibly the thermal map M2 of the blank 4. Particularly, the IR cameras 9a-9f are arranged in mutually aligned relationship, to detect all the work positions of the blank 4, from the starting step of centering in which it is aligned with the press 3 to the release step, with the achievement of the required dimensions and, for each operation step, the corresponding set of intermediate end dimensions, within tolerance limits.

Advantageously, the method provides the use of six IR cameras 9a-9f. Considering a Cartesian reference system with axes X, Y, Z, as shown in Figure 1, in the method the six IR cameras 9a-9f are arranged to allow detection of the blank 4 in a vertical forging plane X-Z, which is substantially perpendicular to a plane P on which the manipulator 2 is supported. Possibly, in the method a few secondary IR cameras are provided, for real-time detection of the blank 4 in a plane X- Y which is substantially the horizontal forging plane.

Conveniently, the method provides for equipping the artificial intelligence system

10 with a control unit AIM which receives the dimensional maps MI and possibly the thermal maps M2, in real time, and processes the values therein using an artificial intelligence algorithm, to estimate the dimensions and further characteristics of the blank 4. The control unit AIM further uses the training maps STM to determine such features.

The starting step of centering the blank 4 will be now described by way of illustration and without limitation.

Once the control unit AIM (Artificial Intelligence Module) receives a start signal, informing that the hot blank 4 is about to come out of the furnace, a preventive initial actuator monitoring step is carried out.

Using a preset operational sequential algorithm SA, the initial monitoring step checks that the actuators are located in a safety area and, if this is not the case, it generates precise and preset sequences of first control signals Inl, second control signals In2 and possibly other control signals, to possibly place secondary actuators in these safety areas.

A secondary actuator is defined by a support device 6 interposed between the manipulator 2 and the press 3, and has a support base for supporting the blank 4 before the latter is seized by the manipulator 2.

The control unit AIM uses a further operation sequential algorithm SA to generate corresponding control signals to lift and rotate the support device 6 such that a robot that extracts the blank 4 from the furnace can lay it on the support base. Of course, the support device 6 is placed with its support base at a level that, considering a reference plane, e.g. the support plane P for the manipulator 2, depends on the nominal diameter of the blank 4.

The support device 6 is rotated such that the blank 4 on the support base is placed in an axial and central position with respect to an entrance 3a of the press 3.

The blank 4 is detected by the detection means 8, and the IR cameras 9a-9f and the visual module VSM generate a dimensional map Ml and possibly the thermal map M2 of the blank 4. Particularly, the training maps STM comprise one or more comparative tables and suitable preset rules 15 allowing the intensity of the electromagnetic radiation emitted by the hot workpiece 4 to be correlated to a color of the grayscale, with the hot blank 4 being mapped to color pixels, in order to correct the deformations of the dimensional map Ml and generate the reference dimensional map Mr.

The artificial intelligence system 10 and namely the control unit AIM receives the reference dimensional map Mr and estimates the dimensions of the blank 4. Particularly, the control unit AIM uses preset parameters PGI and preset rules 15, included in the training maps STM, and estimates the actual dimensions of the blank 4 irrespective of any recognized cold elements, such as surface scales or else. This is because scales are cold elements and are identified as black spots in the thermal map M2, and hence recognized by the control unit AIM.

Particularly, the control unit AIM by processing the thermal map M2 allows recognition and identification of a surface scale from an end-of-section scale 7. Thus, while the surface scale has to be deemed as part of the blank 4, the end-of-section scale 7 generates a hollow area and shall be deemed as a portion to be removed. Figure 4 schematically shows the end-of-section scale 7 as the darkest portion to the right of the blank 4.

Once the dimensions of the blank 4 have been estimated, the artificial intelligence system 10 with the preset parameters PGI defines the gripping level and generates the first control signals Inl to control the arm of the manipulator 2 such that the blank 4 may be clamped at a gripping point. Particularly, the gripping point is required to match the median point of the blank 4.

The blank 4 is lifted and is detected again by the IR cameras 9a-9f, with its dimensions being estimated again by the artificial intelligence system 10 and the control unit AIM, and analyzed against an additional preset rule 15. Particularly, this additional preset rule 15 are designed to check whether the jaws of the arm 2a of the manipulators 2 are properly closed on the blank 4 with no interlocking arrangement. More particularly, the additional preset rule 15 is designed to determine whether the blank 4 is properly gripped, in line with the arm 2a of the manipulator 2, according to the angle of the lifted blank 4.

If a difference is found between the gripping point and the median point, the value of this dimensional difference is determined by means of the deviation map E and the artificial intelligence system 10 is adapted as an error function to reduce the amount of this difference. Particularly, the control unit AIM uses an appropriate preset rule 15 to generate further control signals Inl to move the manipulator 2 and straighten the jaws of the arm 2a by an iteration of movements that tend to correct the anomaly, e.g. with the method of successive approximation.

During the operation step OS such as rough machining of the blank 4, the control unit AIM analyzes the dimensions and temperature of the blank 4 based on continuous detections by the IR cameras 9a-9f, and autonomously decides the process parameters, such as the pressing strokes of the press 3, as well as the amount of material to be pressed, to optimize the forging process times.

The present invention also relates to a forging system 1 for implementing the above described control method, in which cooperating details and parts having the same structure and the same operation will be designated by the same numerals and acronyms.

The system 1 comprises main actuators, i.e. devices that directly interfere with the forging process and secondary actuators, which assist the main actuators and do not directly interfere with the forging process.

The main actuators are a manipulator 2 and a press 3 that are controlled and moved in automated and synchronized fashion and are adapted to forge a blank 4 that comes from a furnace. The secondary actuators include: loading systems, unloading systems, motorized placement devices 6 and automation systems for changing dies on the press 3.

The manipulator 2 is a kind of robot that can grasp and move the blank 4 in any direction, even under the press 3.

The manipulator 2 comprises a moving arm 2a having a gripper whose jaws are designed to grip the blank 1 that has mainly, but without limitation, a substantially cylindrical elongate shape with a longitudinal axis A-A.

Reference will be made hereinbelow to a Cartesian reference system with axes X, Y and Z, with a plane X-Z being referred to as a main or vertical forging plane, and a plane X-Y being referred to as a secondary or horizontal forging plane.

The axis A-A lies on the horizontal plane X-Y which is substantially parallel to support plane P of the manipulator 2. The horizontal plane X-Y is disposed at a reference level D relative to the support plane P.

In a starting step of centering, the blank 4 is engaged by the manipulator 2 and in later positioning steps it is rotated about the axis A-A, translated in the axial direction A-A and inclined to form angles a between the axis A-A and the horizontal plane X-Y and angles β between the axis A-A and the vertical plane X-Z, which are automatically obtained according to the dimensions that have been assumed. The manipulator 2 is also designed to carry out various rotation steps, in which the blank 4 is rotated by 90°, with the horizontal plane X-Y exchanging each time with the vertical plane X-Z.

The press 3 moves in the direction Z perpendicular to the horizontal plane X-Y and deforms the blank 4 by an open die 5, which is substantially composed of two half- shells located on each side of the horizontal plane X-Y. Once the blank 4 is positioned, it is pressed under the press 3, with as many pressing cycles as required to obtain the required dimensions.

The system 1 comprises detection means 8 for detecting the electromagnetic infrared radiation emitted by the blank 4.

In one embodiment, the detection means 8 are defined by an array of infrared or IR cameras 9a-9f, connected to a visual processing module VSM, which is adapted to receive the signals transmitted by the infrared cameras 9a-9f and to generate the dimensional map Ml and possibly a thermal map M2 of the blank 4.

The IR cameras 9a-9f are substantially aligned along a directrix C-C which is located at the side of the manipulator 2 and the press 3. The array of IR cameras 9a-9f allows detection of the blank 4 during the entire forging process: from the initial step in which the blank 4 is centered to a final release step.

In the illustrated embodiment, there are six IR cameras 9a-9f in the array, in such an arrangement as to allow detection of the vertical plane X-Z.

Figure 3 is a schematic top view of the manipulator 2, the press 3 and the array of IR cameras 9a9f, which are arranged along a directrix C-C substantially parallel to the directrix X-X.

The IR cameras 9a-9f may be also arranged along multiple directrices Ci-Ci, and the distance dl from the directrix X-X may be also changed for logistical reasons.

In a preferred arrangement, along the directrix C-C there is a sequence of:

- a first IR camera 9a located at a motorized support device 6, which is designed to detect the blank 4 during the starting step of centering, the support device 6 being adapted to support the blank 4 during this step;

- a second IR camera 9b located proximate but external to the press 3, for detecting the starting step of centering at the entrance opening 3 a of the press 3;

- a third IR camera 9c located proximate to the initial end of the open die 5 for measuring the diameter and position X of the workpiece that enters the die;

- a fourth IR camera 9d located at the center of the open die 5 for measuring the amount of material in the die;

- a fifth IR camera 9e located proximate to the final end of the open die 5 for measuring the diameter and position X of the workpiece that exits the die;

- a sixth IR camera 9f located proximate but external to the press 3, for detecting the final step of release of the finished workpiece.

Each IR camera 9a-9f of the array is placed at a distance from the previous and next cameras, to detect the entire work area. Particularly, the second IR camera 9b, the third IR camera 9c, the fourth IR camera 9d and the fifth IR camera 9d are in such positions that the field of view of each of them partially overlaps the field of view of the adjacent IR cameras 9a- 9f at the directrix X-X, as shown in Figure 3.

The number and position of the IR cameras 9a-9f, as well as the position of the direction C-C and the level thereof relative to the support plane P may change according to the type of system and shall be intended without limitation to the present invention.

The visual module VSM is adapted to process the intensities of the radiations received from the IR cameras 9a-9f and to map the blank 4 to color pixels in a grayscale, to define the dimensional map Ml and possibly the thermal map M2 of pixels.

The detection means 8 are connected to an artificial intelligence system 10 which comprises different types of logics, such as: mathematical logics, statistical logics, artificial neural networks, genetic networks and fuzzy logics.

The artificial intelligence system 10 comprises a control unit ATM which is connected to the visual module VSM. Particularly, the control unit AIM is adapted to receive the dimensional map Ml and possibly the thermal map M2 and to estimate the dimensions of the blank 4 in real time, by computer means, using a series of algorithms and appropriate preset rules 15.

In one embodiment, the control unit AIM further comprises a plurality of algorithms, which are adapted to synthesize the shape of the blank 4 during the next processing operation steps of the forging process.

Particularly, the control unit AIM is adapted to receive successive dimensional maps Ml and possibly thermal maps M2 and to synthesize them in a few basic measurements, such as actual diameters and amounts of forged material, for optimization of the forging process. At least one algorithm in the control unit AIM is further adapted to control the displacements imparted to the manipulator 2 and the press 3, to carry out the operation steps of the operation sequence OS. Particularly, the control unit AIM generates appropriate first control signals Inl and second control signals In2 respectively, adapted to move the manipulator 2 and the press 3.

Furthermore, at least one algorithm is adapted to assess various classes of information, deriving from:

- operation sequences OS, including a series of operation steps of the forging process, appropriately stored in a first module 12 connected to the control unit AIM;

- training maps STM (Sequence Training Maps), i.e. maps for training sequences of operational behaviors, such behaviors including physical rules allowing transformation of IR radiation maps into thermal maps, dimensional transformations, material deformation transformations according to the pressure that is or may be exerted by the press 3 or other similar transformations. These operational behaviors may also include sampling deduced from actual experiences of operators. The training maps STM are stored in a second module 13 connected to the control unit AIM and are adapted to define the next actions of the operation steps. The second module 13 also comprises preset rules 15, which are also used to define the next operation actions;

- operation sequence algorithms SA, which are computational algorithms of operation step actions, which are stored in a third module 14 connected to the control unit AIM and are used to define the next actions of the operation steps with a lower complexity and/or to replace the training maps STM;

- preset parameters PGI (Product General Information), i.e. the values of the required dimensions of the finished workpiece, as appropriately determined from technical drawings, that are suitably stored in the artificial intelligence system 10, from further finished workpiece manufacturing data, as well as from the dimensions of the dies in use. The preset parameters PGI are placed in a fourth module 1 1, which is connected to the control unit AIM.

The forging system 1 may further comprise additional sensors 16 and/or additional auxiliary devices or actuators 17, for detecting ambient conditions during the forging process. Particularly, the additional sensors 16, e.g. pressure sensors, are aiding sensors that sense the pressure of the press 3 in view of optimizing the forging process.

Certain sensors 16, e.g. thermal sensors, may be used in the initial step to calibrate the thermal maps M2 of the IR cameras 9a-9f. The system may further include optical sensors 16 for detecting, for instance, further levels in the placement of the open die 5 of the press 3.

Concerning the operation of the forging system 1, the IR cameras 9a-9f allow successive detections of the blank 4, whereas the visual module VSM allows later definition of corresponding dimensional maps Ml and possibly thermal maps M2.

The dimensional maps Ml and the thermal maps M2 are processed by the artificial intelligence system 10 and particularly by the control unit AIM, which allows estimation of the actual dimensions of the blank 4, using the preset parameters PGI and the preset rules 15 included in the training maps STM. The control unit AIM can also account for the thermal lenses generated by the hot blank 4.

The control unit AIM compares the estimated actual dimensions and the preset required dimensions associated with each operation step of the operation sequence OS, based on the preset parameters PGI in the fourth module 1 1 and creates a deviation map E as a difference between these dimensions.

Therefore, according to the reference deviation map E, the control unit AIM generates first control signals Inl and second control signals In2, which are sent to the manipulator 2 and the press 3 respectively, in order to cause the predetermined actions of each operation step of the operation sequence OS. These predetermined actions are generated to reduce the values in the deviation map E within the limits of tolerances defined in each operation step OS.

In one embodiment, each IR camera 9a-9f is adapted to take about 30 pictures per second, and the system AIM is adapted to detect more than 100 dimensions of the blank 4 for each picture. Some of the dimensions so detected are essential and are used to optimize the forging process, by reducing or eliminating any undesired and complex shape that the blank 4 may assume during each processing operation step.

Advantageously, according to the present invention, the control method and the forging system allows to obtain the prefixed objects by providing full automation of the forging process, which may be also followed by unskilled personnel. Indeed, since the method and system of the invention allow fully autonomous detection and determination of actual dimensions, the personnel shall only provide supervision in response to any error function generated by the system, and their action shall be only required under emergency conditions.

The throughput of the forging system of the present invention is considerably increased, as the times for analysis and estimation of the blank dimensions are of the order of one tenth of a second, i.e. much shorter than the times required by manual detection, i.e. a few seconds. Also, the cyclic operation processes that may be required during certain processing steps are carried out in synchronized fashion, which will optimize the execution times.

Furthermore, there will be a much smaller number of dimensional errors, as compared with manual or semi-automatic forging systems.

Obviously, a technician of the field, aiming at meeting incidental and specific needs, will bring several modification and alternatives to the above configuration, all within the scope of protection of the invention as defined in the following claims.