CZ36943U1 - Device for checking surface defects of forgings - Google Patents

Description

Current state of the art

Despite the general use of penetration technologies, inspection of surface defects of forgings is almost exclusively carried out by human visual inspection. If the defect has not yet been detected by the manufacturer of the forgings, unwanted losses occur during the subsequent processing of de facto unsaleable scrap. Alternatively, there is a risk of a complaint from the final customer. In addition, machining tools and machines are unnecessarily worn out, in some cases the tools are destroyed - due to unexpected disproportion of the semi-finished product. Therefore, visual inspection by the operator of the machine tool is used in the work shops before machining, where the result depends on the subjective judgment of the operator.

As far as automated systems for the detection of the forging process and defects in forgings are concerned, they are currently known exclusively in connection with the taken thermal images. Thus, for example, Chinese patent application CN 113641155 A describes a high temperature forging detection control system that includes an information acquisition module, a control module, and an auxiliary function module, wherein the control module is based on an embedded industrial personal computer and includes a Halcon image processing module, a positioning module, detection module, temperature detection module, size detection module, defect detection module and quality evaluation module. The quality evaluation module evaluates the quality of the forging according to the detection data of the position detection module, temperature detection module, size detection module, and defect detection module and produces a detection report; the information acquisition module includes a camera and an infrared temperature sensor, acquires image information, and sends image information to the control module; and the auxiliary function module includes a camera control module, a point cloud analysis module, a motion control module, and a detection data and detection report query module. According to the high-temperature forging detection control system, accurate detection of various parameters of the detected forging is realized in real time in a high-temperature environment.

The solution according to Chinese patent application CN 115035084 A, which relates to a method and device for monitoring forging defects with relevant electronic equipment and a storage medium, is of a similar type. The method includes the steps of: obtaining the original image of the forging and the thermal image of the forging in the process of forging, detecting the defects of the thermal image through the neural network processing module, generating the heat map of the forging with defect marking, fusion of the original image and the heat map with defect marking, generating the target image of defects, performing orthogonal frequency division coding modulation on the target defect image, generating monitoring image data, and sending the monitoring image data to the monitoring terminal. The monitoring image data is used for the monitoring terminal to determine the forging defect in the forging process. Thus, the monitoring terminal can clearly recognize the defects of the forging in the forging process, and the accuracy of the recognition of the defects of the forging can be improved by using the intelligent monitoring system.

As is clear from the above description of known solutions, this is the detection of defects in forgings based on their identification from thermal images taken during the forging process. They are very complicated systems, the application of which in practice and especially the interpretation of their outputs for the unequivocal detection of surface defects of final forgings, such as missing material, errors after trimming or punching (burr, splice, pushed needle), displacement, deflection, folded material - rivets or transpositions, can be quite problematic.

- 1 CZ 36943 U1

In practice, there are also laser applications for the evaluation of surface defects. This technology can perfectly map the surface of the examined forging and compare it with a 3D model from the CAD environment. The use of this method is more time-consuming due to longer scanning of the surface of forgings with complex shapes. Also, the cost of investing in a 3D scanner is about three times higher than the technical solution proposed below. In order for the surface of the evaluated forging to be scanned correctly, the 3D laser camera or the component must be positioned by a 6-axis manipulator (robot), in the presented technical solution it is only a 3-axis manipulator. Overall, the proposed solution brings a more productive and economical output when evaluating forging defects.

The essence of the technical solution

The device for checking the surface defects of forgings according to the presented technical solution contributes to a large extent to the elimination of the shortcomings of known solutions. It includes an input transfer point for the manipulator with places for storing and leveling the forgings, a rotating arm of the manipulator, rails for the manipulator to travel, a camera inspection cell and output transfer points with a magazine for defective forgings and a magazine for perfect forgings.

The essence of the technical solution lies in the fact that the arm of the manipulator contains rotating jaws for gripping the forging and is connected to the kinematic elements of the manipulator to move the forging into the camera control cell and to the kinematic elements to position the forging in the camera control cell by moving along the horizontal axis and rotating the forging around this axis . This makes it possible to take multiple sets of camera images for each forging. The camera inspection cell is equipped with at least 6 cameras - a camera for a plan view of the forging, a camera for a side view of the forging, a camera for an elevational image of the forging, a camera for the projection of a line laser on the surface of the forging, the optical axis of which lies in the horizontal plane given by the optical axes of the cameras for the side view and elevational image, and with these axes form an angle of 40° to 50°, while the laser plane is perpendicular to the optical axis of the plan view camera, and also cameras for isometric images of the forging, whose optical axes form an angle of 40° to 50° with the horizontal plane. The distances of all cameras from the center of the forging are 300 to 500 mm. The camera control cell is equipped with an upper flat light, a side backlight, a rear backlight and a lower line light for its lighting. Light switching elements, laser switching elements, switching elements and camera outputs are connected to an evaluation computer that contains a module for evaluating camera images. The result of the evaluation is then information on whether the forging is defect-free or defective.

The module for evaluating camera images is preferably an evaluation module based on a deep neural network.

Clarification of drawings

The attached drawings serve to clarify the nature of the technical solution in more detail, where it represents:

Fig. 1 overall diagram of the device for checking surface defects of forgings; and Fig. 2 is a detail of the internal arrangement of the camera inspection cell.

An example of the implementation of a technical solution

The device for checking the surface defects of forgings in an exemplary embodiment (see the diagram in Fig. 1) includes an input transfer point 4 for the manipulator with places 5 for storing and leveling the forgings, a rotating arm 1 of the manipulator, a rail 3 for the movement of the manipulator, a camera control cell 2 and an output transfer points with magazine 6 (MARS pallet) for defective forgings and magazine 7 (MARS pallet) for perfect forgings.

- 2 CZ 36943 U1

Arm 1 of the manipulator contains rotating jaws for gripping the forging 2.1 and is connected to the kinematic elements of the manipulator to move the forging 2.1 to the cell 2 of the camera control and to the kinematic elements to position the forging 2.1 in the cell 2 of the camera control by shifting along the horizontal axis and rotating the forging 2.1 around this axis.

Cell 2 camera control (see Fig. 2) is equipped with 6 cameras 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f.

Particulary speaking about

- camera 2.3a of the plan view of the forging 2.1,

- camera 2.3b of the side view image of the forging 2.1,

- camera 2.3c of the outline image of the forging 2.1,

- the camera 2.3d of the projection of the line laser 2.5 onto the surface of the forging 2.1, whose optical axis lies in the horizontal plane given by the optical axes of the cameras 2.3b of the side-view image and 2.3c of the elevation image and forms an angle of 40° to 50° with these axes, while the laser plane is perpendicular to the optical axis of the camera 2.3c of the outline image and beyond

- two cameras 2.3e, 2.3f of isometric images of the forging 2.1, whose optical axes form an angle of 40° to 50° with the horizontal plane.

The distances of all cameras 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f from the center of the forging are 300 to 500 mm.

Constant light conditions inside the cell 2 of the camera inspection are ensured by light sources. The light sources are powerful industrial LED lights designed for machine vision, ensuring homogeneous illumination of the part. To illuminate the forging, cell 2 of the camera inspection is equipped with an upper flat light 2.4a, a side backlight 2.4b, a rear backlight 2.4c and a lower line light 2.4d.

Light switching elements 2.4a, 2.4b, 2.4c, 2.4d, laser switching elements 2.5, switching elements and camera outputs 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3fj are connected to an evaluation computer which includes a module to evaluate images of cameras 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f.

Image data from cameras 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f are transferred to the evaluation computer, where they are processed using deep neural network methods. In practice, it is often a problem to obtain a sufficient number of samples with all types of defects, but it is not a problem to obtain enough samples of flawless forgings. For this reason, forging defect detection is implemented as anomaly detection and localization. Before testing a particular type of forging for the first time, a set of flawless pieces of that forging must be submitted to the designed facility. Camera images of these pieces are used to create a model of the perfect forging, which is stored in a central evaluation computer. A pre-trained neural network is used to create a model of the forging. In the model creation phase, corresponding activation maps from the pre-trained neural network are assigned to each slice of the image. From a set of images of flawless pieces, the characteristics of the normal distribution of this class are calculated. In the test phase, the distance from this normal distribution is then calculated for each slice of the image. A greater distance indicates an area with an anomaly. Forgings with significant anomalies are evaluated as defective.

The rotating arm 1 of the manipulator takes the inspected forging 2.1 from the input transfer point 4 and places it in the camera inspection cell 2, where images are taken and evaluated. Depending on the complexity of the forging, it is possible to take several series of images that differ in the position of the forging. The manipulator allows you to change the position of the forging by moving it along the horizontal axis and rotating it around this axis. The outputs of cameras 2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3fj are connected to the evaluation

- 3 CZ 36943 U1 to a computer which, based on the images taken, determines whether the forging is in order (flawless forging) or is defective. This information is transmitted to the control unit of the entire device. The manipulator then places the forging in the appropriate magazine 7 (MARS pallet) for perfect forgings or in magazine 6 (MARS pallet) for defective forgings.

When changing the type of forging being evaluated, it is sufficient to replace the rotating jaws of the arm 1 of the manipulator for gripping the forging 2.1 and to switch the software configuration of the evaluation computer and the control unit. The device enables the evaluation of forgings up to 25 cm x 25 cm x 10 cm.

Offset measurement is performed by measuring the profiles of the forging in cross sections. Scanning the projection of laser beams onto the surface of the forging using HD cameras allows creating profiles with a resolution of 0.05 mm in the transverse axis.

Industrial applicability

The technical solution can be used in automated engineering processes, where it can fulfill the function of input control before the following machining operations. It is also used in the role of output control in operations based on die forging.

Claims (2)

PROTECTION CLAIMS 1. Device for checking the surface defects of forgings, containing an input transfer point (4) for the manipulator with places (5) for storing and leveling the forgings, a rotating arm (1) of the manipulator, rails (3) for the travel of the manipulator, a cell (2) for camera inspection and output transfer points with a magazine (6) for defective forgings and a magazine (7) for perfect forgings, characterized in that the arm (1) of the manipulator contains rotating jaws for gripping the forging (2.1) and is connected to the kinematic elements of the manipulator to move the forging (2.1) into the camera inspection cell (2) and by kinematic elements to position the forging (2.1) in the camera inspection cell (2) by moving it along the horizontal axis and rotating the forging (2.1) around this axis, and that the camera inspection cell (2) is fitted at least 6 cameras (2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f), i.e. camera (2.3a) of the plan image of the forging (2.1), camera (2.3b) of the side image of the forging (2.1), camera (2.3c) of the front-view image of the forging (2.1), with the camera (2.3d) projecting the line laser (2.5) onto the surface of the forging (2.1), whose optical axis lies in the horizontal plane given by the optical axes of the cameras (2.3b) of the side-view image and (2.3c) ) of the plan image and forms an angle of 40° to 50° with these axes, while the laser plane is perpendicular to the optical axis of the camera (2.3c) of the plan image, and further by the cameras (2.3e, 2.3f) of the isometric images of the forging (2.1), whose optical axes form an angle of 40° to 50° with the horizontal plane, with the distances of all cameras (2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f) from the center of the forging being 300 to 500 mm, while cell (2) camera control is equipped with an upper flat light (2.4a), side backlight (2.4b), rear backlight (2.4c) and lower line light (2.4d) for its illumination, and while the light switching elements (2.4a, 2.4b, 2.4 c, 2.4d), laser switching elements (2.5), switching elements and camera outputs (2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f) are connected to an evaluation computer that contains a module for evaluating camera images (2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f). 2. Device according to claim 1, characterized in that the module for evaluating camera images (2.3a, 2.3b, 2.3c, 2.3d, 2.3e, 2.3f) is an evaluation module based on a deep neural network.