Die casting machine

Abstract

The invention provides a die casting machine capable of more accurately obtaining the amount of molten metal. A die casting machine (10) is provided with: a robot (34) having a plurality of electrically controllable joints and a plurality of links connected by the plurality of joints; a ladle (32) attached to the robot (34); a sleeve (14) that receives the molten metal; and a controller (60) configured to drive the robot (34) so that the position and orientation of the ladle (32) are three-dimensionally changed, to convey the molten metal to the sleeve (14) by using the ladle (32), and to calculate the amount of molten metal in the ladle (32) on the basis of the load of at least one of the plurality of joints.

Description

Technical Field

[0001] This specification discloses a die-casting machine that uses a robot to move the ladle. Background Technology

[0002] A die-casting machine with a molten metal supply device that delivers molten metal via a ladle has been known for some time. In this case, molten metal is drawn from the furnace via a ladle and supplied to a sleeve located on the main body of the die-casting machine. To maintain a high quality of the die-cast product, it is necessary to accurately detect the amount of molten metal supplied to the sleeve.

[0003] Therefore, techniques for measuring the amount of molten metal drawn from a ladle have been proposed in the past. For example, Patent Document 1 discloses a die-casting machine in which a ladle is moved by a ladle moving mechanism. Patent Document 1 discloses a method for calculating the amount of molten metal in a ladle based on the oil level position of the molten metal detected by an energized sensor, the oil level height of the molten metal detected by a displacement sensor, or an image obtained by photographing the molten metal in the ladle.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2021-192918 Summary of the Invention

[0007] The problem that the invention aims to solve

[0008] Patent Document 1 calculates the amount of molten metal based on the height and shape of the molten metal surface. However, the molten metal surface varies significantly depending on the ladle's orientation. For example, the height and shape of the molten metal surface change drastically depending on whether the ladle is facing upwards or tilted. Therefore, the technology in Patent Document 1 is prone to errors caused by the ladle's orientation, making it impossible to accurately calculate the amount of molten metal.

[0009] Therefore, this specification discloses a die-casting machine that can more accurately obtain the amount of molten metal.

[0010] Methods for solving problems

[0011] The die-casting machine disclosed in this specification is characterized by comprising: a robot having a plurality of electrically controllable joints and a plurality of links connected through the plurality of joints; a ladle mounted on the robot; a sleeve for receiving molten metal; and a controller configured to drive the robot in a manner that causes the position and orientation of the ladle to change in three dimensions, to transport molten metal to the sleeve using the ladle, and to calculate the amount of molten metal in the ladle based on the load of at least one of the plurality of joints.

[0012] In this case, the robot may also have multiple motors respectively disposed on the multiple joints and move the corresponding joints, and the controller is configured to calculate the amount of molten metal based on the change in load when the robot changes from a first posture to a second posture, wherein the load is the output torque or current value of the motor.

[0013] Alternatively, both the first and second postures can be the postures of the robot when it is temporarily stationary during the process of conveying molten metal from the furnace to the sleeve.

[0014] Alternatively, the first posture can be a metering posture in which the remaining molten metal in the extracted molten metal is returned to the furnace, and the second posture can be a casting standby posture near the sleeve, ready for casting.

[0015] Alternatively, the controller may be configured to adjust the injection action that extrudes the molten metal from the sleeve into the mold based on the calculated amount of molten metal.

[0016] Other die-casting machines disclosed in this specification are characterized by comprising: a ladle for drawing molten metal and pouring it into a sleeve; an output data collection unit for collecting the trajectory of the ladle; an input data collection unit for collecting pouring quality and pouring conditions; and a controller configured to function as an AI by learning, at least based on the trajectory collected by the output data collection unit and the pouring quality and pouring conditions collected by the input data collection unit, in order to output a trajectory corresponding to a target pouring quality, and to control the drive of the robot in such a way that the trajectory of the robot or the ladle becomes the trajectory output from the AI.

[0017] Invention Effects

[0018] The die-casting machine disclosed in this specification allows for more accurate measurement of the amount of molten metal. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the structure of a die-casting machine.

[0020] Figure 2 This is a diagram illustrating the injection process.

[0021] Figure 3 This is a diagram illustrating an example of calculating the amount of molten metal.

[0022] Figure 4 This is a diagram illustrating an example of the plunger's movement curve.

[0023] Figure 5 This is a diagram of a typical neural network model.

[0024] Explanation of reference numerals in the attached figures

[0025] 10 Die-casting machine, 12 Equipment body, 14 Sleeve, 15 Variable cavity, 16 Gating gate, 18 Piston, 20 Mold, 22 Chamber space, 24 Furnace, 30 Molten metal supply device, 32 Ladle, 34 Robot, 36 Linkage, 40 Basic joint, 42 Root joint, 44 Intermediate joint, 46 Wrist joint, 50 Optical camera, 52 Thermal imaging camera, 54 Disc, 56 Weight sensor, 60 Controller, 62 Processor, 64 Memory, 66 Timer, 70 Model, 72 Input layer, 74 Intermediate layer, 76 Output layer, 78, 80, 82 Nodes, 84 Edge, K1 First interval, K2 Second interval, K3 Third interval. Detailed Implementation

[0026] The structure of the die-casting machine 10 will be described below with reference to the attached drawings. Figure 1 This is a schematic diagram showing the structure of the die-casting machine 10. The die-casting machine 10 is a device for performing die-casting processes. For example, the die-casting machine 10 is a device that integrally molds large parts used in vehicles or aircraft, and is referred to as a mega-class die-casting machine or giga-class die-casting machine.

[0027] The die-casting machine 10 includes a main body 12, a furnace 24, a molten metal supply device 30, and a controller 60. For example... Figure 2 As shown, the main body 12 of the equipment has a mold 20 forming a chamber space 22 and a sleeve 14 communicating with the mold 20. After molten metal is injected into the chamber space 22 through the sleeve 14, the molten metal is cooled and solidified, thereby manufacturing a die-cast product. A detailed description of the structure of the main body 12 of the equipment is omitted.

[0028] The furnace 24 heats the material metal (such as aluminum) of the product until it melts. The molten material metal, i.e., molten metal, is stored in the furnace 24. The furnace 24 is located separately from the main body of the equipment 12.

[0029] The molten metal supply device 30 supplies molten metal stored in the furnace 24 to the sleeve 14 of the equipment body 12. The molten metal supply device 30 includes a ladle 32 for accumulating molten metal and a robot 34 for transporting the ladle 32. The robot 34 is a multi-joint robot with multiple links 36 connected by joints. Each joint is equipped with an electric motor as a power source, and the movement of each joint is electrically controlled by the electric motor. Each electric motor is equipped with a sensor (e.g., an encoder) for detecting its driving force. Furthermore, as the joints move, the posture of the robot 34 changes, thereby altering the position and posture of the ladle 32 mounted on the robot 34.

[0030] exist Figure 1 In this example, robot 34 has a base joint 40 capable of rotating about a vertical axis and four joints 42, 44, and 46 capable of rotating about mutually parallel axes. Hereinafter, the most basal joint among the four joints 42, 44, and 46 will be referred to as the "root joint 42," the most foremost joint as the "wrist joint 46," and the joint located between the root joint 42 and the wrist joint 46 as the "intermediate joint 44." The ladle 32 is mounted to the foremost side of robot 34 via the wrist joint 46.

[0031] The controller 60, described later, calculates the posture of the robot 34 based on the actuation values ​​of each of the multiple joints, and then calculates the position of the ladle 32. Additionally, the controller 60 obtains the load values ​​of each of the multiple joints. For example, the controller 60 can obtain the current value or the output torque of the motor installed at the joint as the load value of the joint. Alternatively, a torque sensor that detects the torque acting on the joint can be provided, and the detection value of the torque sensor can be obtained as the load value of the joint. Based on the load value of the joint, the controller 60 calculates the amount of molten metal drawn by the ladle 32, which will be described later.

[0032] Robot 34 uses ladle 32 to draw molten metal from furnace 24. Then, robot 34 transports ladle 32 to the vicinity of pouring port 16 of sleeve 14 and pours the metal from ladle 32 into pouring port 16. It should be noted that an optical camera 50 and a thermal imaging camera 52 are positioned near sleeve 14 to capture images of the area around sleeve 14. Controller 60 evaluates the pouring quality based on the images captured by the optical camera 50 and thermal imaging camera 52, which will be described later.

[0033] The molten metal poured into the sleeve 14 passes through the plunger 18 (see reference). Figure 2 The injection is made into chamber space 22. The forward and backward movement of plunger 18 is controlled, for example, by a hydraulic circuit (not shown).

[0034] The controller 60 controls the drive of the die-casting machine 10. Physically, the controller 60 is a computer with a processor 62 and a memory 64. It should be noted that... Figure 1 In the illustration, the controller 60 is shown as a single computer, but the controller 60 can also be configured as a combination of multiple physically separate computers.

[0035] The controller 60 controls the drive of various parts of the die-casting machine 10 based on instructions from the operator and detection values ​​from sensors installed at various locations within the die-casting machine 10. For example, the controller 60 controls the movement of movable parts of the die-casting machine 10 (e.g., mold 20, plunger 18, and robot 34), or controls the temperature of the furnace 24 and mold 20. Furthermore, in this example, the controller 60 also calculates the amount of molten metal drawn by the ladle 32 based on the load on the joints of the robot 34. The calculation of this amount of molten metal will be explained below.

[0036] First, the rationale for calculating the amount of molten metal will be explained. For example... Figure 2 As shown, molten metal supplied to sleeve 14 is injected into chamber space 22 via plunger 18. During casting, plunger 18 retracts to the base end of sleeve 14. At this time, the volume Vc of the space surrounded by sleeve 14 and plunger 18 (hereinafter referred to as "variable cavity 15") is larger than the volume Vm of the cast molten metal. Therefore, immediately after casting, as... Figure 2 As shown in the figure enclosed by an ellipse, air exists above the surface of the molten metal.

[0037] Subsequently, the plunger 18 advances forward, and the variable cavity 15 gradually decreases in size. Furthermore, if the volume Vc of the variable cavity 15 is approximately equal to the volume Vm of the molten metal, the variable cavity 15 is filled with molten metal. Then, the plunger 18 advances further, and the molten metal is extruded from the variable cavity 15 into the chamber space 22. During the period before the molten metal has covered most of the chamber space 22, the pressure acting on the molten metal is relatively low. Then, after the molten metal has covered most of the chamber space 22, the plunger 18 is advanced further, thereby applying pressure to the molten metal inside the chamber space 22, causing the molten metal to spread to the ends of the chamber space 22. It should be noted that, in order to distribute the molten metal to the details, the pressure inside the chamber space 22 can be reduced before injection.

[0038] That is, the advance of the plunger 18 can be roughly divided into three sections: a first section K1 where the volume Vc of the variable cavity 15 is larger than the volume Vm of the molten metal; a second section K2 where Vc becomes less than Vm and the molten metal is squeezed into the chamber space 22; and a third section K3 where a larger pressure is required to squeeze out the molten metal. In the first section K1, to prevent fluctuations in the molten metal and, consequently, to prevent air entrapment, the plunger 18 is required to move at a low speed. By preventing air entrapment, the generation of porosity in the casting is reduced. In the second section K2, to forcefully squeeze out the molten metal, the plunger 18 is required to move at a high speed. By moving the plunger 18 at a high speed, the molten metal is filled in a short time, thereby reducing the deviation in the solidification period of the molten metal. Furthermore, in the third section K3, to apply pressure, the plunger 18 is required to gradually decelerate.

[0039] In other words, to obtain high-quality die-cast products, it is necessary to switch the movement speed of the plunger 18 at the boundary positions of the three zones K1, K2, and K3. However, the boundary positions of the three zones K1, K2, and K3 vary depending on the amount of molten metal supplied from the ladle 32. Therefore, in order to switch the movement speed of the plunger 18 at the appropriate timing, it is necessary to accurately control the amount of molten metal supplied to the sleeve 14.

[0040] Therefore, in this example, when handling the ladle 32, the load on the joints of the robot 34 is obtained, and the amount of molten metal is measured based on this load. That is, the load applied to the joints of the robot 34 varies according to the weight of the ladle 32 held by the robot 34. However, the load applied to the joints is not proportional to the weight of the ladle 32, and also varies according to the posture of the robot 34, etc.

[0041] Therefore, in this example, a predetermined posture is determined in advance, and the weight of the ladle 32 and the molten metal stored in the ladle 32 is calculated based on the load generated under that predetermined posture. More specifically, the controller 60 calculates the amount of molten metal based on the change in joint load when changing from a predetermined first posture P1 to a second posture P2. The steps used in this calculation are not particularly limited. Figure 3 This is a diagram illustrating an example of calculating the amount of molten metal.

[0042] Figure 3 The upper section represents the robot 34 in the predefined first posture P1, and the lower section represents the robot 34 in the second posture P2. It should be noted that, for the sake of simplicity, the center of gravity height of the ladle 32 in the second posture P2 is set to be the same as that of the ladle 32 in the first posture P1.

[0043] First, consider the robot 34 in its first posture P1. When the object (i.e., the ladle 32 that has already drawn molten metal) is mounted on the robot 34 in its first posture P1, the torque acting on the root joint 42 is set to T1. Furthermore, the torque acting on the root joint 42 when the object is not mounted on the end of the robot 34 in its first posture P1 is set to M1. In addition, the horizontal distance from the object's center of gravity G to the root joint 42 is set to D1. In this case, the weight W of the object can be expressed by the following equation (1).

[0044] T1=M1+W×D1(1)

[0045] Similarly, when robot 34 is in the second posture P2, the torque of root joint 42 when there is no object is set to M2, the torque of root joint 42 when there is an object is set to T2, and the horizontal distance between the center of gravity G of the object and root joint 42 is set to D2. In this case, the weight W of the object can be represented by the following equation (2).

[0046] T2=M2+W×D2(2)

[0047] Based on equations (1) and (2), we can further obtain the following equations (3) and (4).

[0048] T2-T1=M2-M1+W×(D2-D1)(3)

[0049] W=((T2-M2)-(T1-M1)) / (D2-D1)(4)

[0050] Here, M1 and M2 are obtained in advance through experiments, etc. In addition, D1 and D2 are also measured in advance. T1 and T2 can actually be obtained by measuring the output torque of the motor of the root joint 42 or the applied current during the process of moving the robot 34 from the first posture P1 to the second posture P2 after the molten metal is drawn by the ladle 32. Furthermore, by substituting these values ​​into equation (4), the weight of the ladle 32 and the molten metal inside the ladle 32 can be obtained.

[0051] This calculation process is one example, and it can be modified as appropriate. For example, the first posture P1 and the second posture P2 can be changed accordingly. Therefore, both the first posture P1 and the second posture P2 can be set as postures that the robot 34 can adopt during the die casting process. In addition, in order to reduce the calculation error of the amount of molten metal, both the first posture P1 and the second posture P2 can be set as postures in which the robot 34 is temporarily stationary during the process of conveying molten metal from the furnace 24 to the sleeve 14. For example, normally, after the molten metal is drawn by the ladle 32, in order to return the remaining molten metal in the ladle 32 to the furnace 24, the robot 34 temporarily stops in a metering posture with the ladle 32 tilted. This metering posture can also be set as the first posture P1. In addition, normally, the robot 34 adopts a pouring standby posture near the sleeve 14 until it can be poured. This pouring standby posture can also be set as the second posture P2. By setting the above structure, it is not necessary for the robot 34 to adopt a special posture for calculating the amount of molten metal, thus preventing an increase in the cycle time during the die casting process. It should be noted that, in this case, the center of gravity height of the ladle 32 in the second posture P2 may be different from that in the first posture P1. In this case, the above equation (4) is corrected in a way that absorbs the difference in center of gravity height.

[0052] Furthermore, while the above description only utilizes the torque of root joint 42, the torque of other joints can also be used to calculate the weight. Also, in the above description, the amount of molten metal is calculated only once based on the load difference between two postures. However, more postures can be used, and the amount of molten metal can be calculated multiple times. By performing statistical processing (e.g., median, average, outlier exclusion, etc.) on these multiple calculated values ​​based on the molten metal amounts, the accuracy of the molten metal amount calculation can be further improved.

[0053] Next, the adjustment of the injection action based on the thus obtained molten metal amount will be explained. As mentioned above, in order to obtain a high-quality die-cast product, it is necessary to switch the speed of the plunger 18 at the boundaries of the first interval K1, the second interval K2, and the third interval K3. Furthermore, the boundary positions of these intervals vary according to the amount of molten metal. Therefore, the controller 60 adjusts the movement curve of the plunger 18 according to the calculated amount of molten metal.

[0054] Figure 4 This is a diagram illustrating an example of the movement curve of plunger 18. Figure 4 In the diagram, the horizontal axis represents the position of plunger 18, and the vertical axis represents the speed of plunger 18. Furthermore, the solid line L1 represents the movement curve when the first molten metal quantity W1 is reached, and the dashed line L2 represents the movement curve when the second molten metal quantity W2 is greater than the first molten metal quantity W1. It should be noted that... Figure 4In the middle, line L2 is slightly offset upwards compared to line L1, but this is to avoid the lines overlapping. In fact, lines L1 and L2 are only different in the timing of their speed changes; they are the same waveform.

[0055] from Figure 4 It can be seen that in this example, when the amount of molten metal is large, the speed switching timing is advanced compared to when it is small. On the other hand, even when the amount of molten metal is different, the shape of the speed waveform of the plunger 18 itself does not change. This is because even if the amount of molten metal changes, as long as the mold 20 (and consequently the chamber space 22) does not change, the distance between the second interval K2 and the third interval K3 does not change.

[0056] To achieve this movement curve control, the controller 60 pre-stores the speed waveform of the plunger 18 for each mold 20. Additionally, the controller 60 stores switching information that correlates the speed switching timing with the amount of molten metal. During each injection operation, the controller 60 compares the calculated amount of molten metal with this switching information to determine the corresponding speed switching timing. Then, the controller 60 controls the movement of the plunger 18 by switching the movement speed at the determined timing. This allows for a higher maintenance of the quality of the die-cast products.

[0057] Next, the control of the trajectory of the ladle 32 during the pouring of molten metal will be explained. Even if the amount of molten metal drawn by the ladle 32 can be accurately detected, a significant amount of molten metal may spill during the pouring process from the ladle 32 to the sleeve 14. In this case, the amount of molten metal supplied to the sleeve 14 will naturally deviate from the intended amount, making it impossible to switch the speed of the plunger 18 at the appropriate timing. Furthermore, if a portion of the molten metal splashes excessively during pouring, the amount of air entrapment into the molten metal increases. Therefore, the pouring quality into the sleeve 14 becomes important in order to maintain the quality of the die-cast product.

[0058] In this example, to maintain a high level of casting quality, the controller 60 generates the movement trajectory of the ladle 32. The controller 60 functions as a pre-learned AI by outputting a trajectory corresponding to the target casting quality. That is, the controller 60 stores a program related to the learned model in its memory 64, and the processor 62 inputs the target casting quality and casting conditions into the learned model, outputting a target trajectory that achieves the target casting quality.

[0059] Here, the learned model can be constructed, for example, using a convolutional neural network. This learned model is learned based on pre-collected information about casting quality, casting conditions, and the trajectory of the ladle 32. The trajectory of the ladle 32 can also be calculated, for example, based on the motion of multiple joints of the robot 34. Alternatively, one or more optical cameras 50 can be installed in the die-casting machine 10 to capture images of the ladle 32 or the robot 34, and the trajectory of the ladle 32 can be calculated based on the images captured by these optical cameras 50. In this case, the sensors measuring the motion of the joints and the optical cameras 50 function as output data collection units that collect trajectory data.

[0060] The pouring quality includes, for example, at least one of the following: the amount of molten metal scattered around the sleeve 14, the time until pouring is completed, and the temperature distribution around the ladle 32. The amount of molten metal scattered can also be calculated based on an image captured by the optical camera 50. Alternatively, a disk 54 for receiving scattered molten metal and a weight sensor 56 for measuring the weight of the disk 54 can be provided on the underside of the sleeve 14. The amount of molten metal scattered can also be obtained based on changes in the detection value of the weight sensor 56. The temperature distribution around the ladle 32 can be obtained, for example, by analyzing a thermal image captured by the thermal imaging camera 52. The time until pouring is completed can be measured by a timer 66 provided on the controller 60. Pouring conditions include, for example, the amount of molten metal in the ladle 32. As described above, the amount of molten metal can be determined based on the load on the joints of the robot 34. Furthermore, the timer 66, optical camera 50, thermal imaging camera 52, weight sensor 56, etc., which collect this information, function as input data collection units for collecting pouring quality and pouring conditions.

[0061] Figure 5 This is a diagram of a typical neural network model (model 70). (For example...) Figure 5 As shown, the neural network model 70 has an input layer 72, an intermediate layer 74, and an output layer 76. It should be noted that the intermediate layer 74 is not limited to one; it can also have multiple layers. Each layer 72, 74, and 76 has one or more nodes 78, 80, and 82, respectively, and nodes in different layers are connected by edges 84. It should be noted that... Figure 5 The diagram only shows a portion of the edges 84, but in reality, multiple edges 84 extend from a node 78, 80. Each edge 84 is assigned a weight coefficient Wi. Furthermore, a bias b is assigned to node 80 in the intermediate layer 74. In the neural network, the values ​​of nodes 78 and 80 are repeatedly weighted and summed, with the bias b applied to the resulting value, and the final value is used as the output. In this example, [the diagram also uses...] Figure 5The neural network is as shown. During pre-learning, the pouring quality and pouring conditions are set for the input node 78, and the bias b and weight coefficient Wi are adjusted so that the final output approximates the trajectory of the ladle 32 when that pouring quality is achieved. The controller 60 repeatedly adjusts the bias b and weight coefficient Wi under multiple conditions, i.e., learns.

[0062] During die casting, the controller 60 inputs the target pouring quality and pouring conditions into the learned model 70, obtaining the trajectory of the ladle 32 to achieve the target pouring quality. The controller 60 then controls the robot 34 to move the ladle 32 along the obtained trajectory. By using the learned model 70 to calculate the trajectory of the ladle 32, the pouring quality can be further improved, thus enhancing the quality of the die-cast product. It should be noted that during the die casting process, the trajectory of the ladle 32, pouring quality, and pouring conditions can also be collected, and the learned model 70 can be further trained based on the obtained data. This allows for the calculation of an appropriate trajectory with higher accuracy.

[0063] It should be noted that the structures described so far are merely examples, and other structures can be appropriately modified as long as the structure of technical solution 1 is present. For example, in the description so far, the controller 60 calculates the amount of molten metal and the trajectory of the ladle 32. However, if the controller 60 calculates the amount of molten metal, it may not need to calculate the trajectory of the ladle 32. Furthermore, the calculation method is not particularly limited as long as the amount of molten metal is calculated based on the load of at least one of the joints of the robot 34.

Claims

1. A die-casting machine, characterized in that, have: A robot having multiple electrically controllable joints and multiple links connected through said multiple joints; A pouring ladle, installed on the robot; Sleeve, for receiving molten metal; as well as Controller The controller is configured as follows: The robot is driven to change the position and orientation of the ladle in three dimensions, and the ladle is used to transport molten metal to the sleeve. Furthermore, the amount of molten metal in the ladle is calculated based on the load of at least one of the plurality of joints.

2. The die-casting machine according to claim 1, characterized in that, The robot has multiple electric motors, which are respectively located at the multiple joints and cause the corresponding joints to move. The controller is configured to calculate the amount of molten metal based on the change in load as the robot changes from a first posture to a second posture. The load is the output torque or current value of the motor.

3. The die-casting machine according to claim 2, characterized in that, Both the first posture and the second posture are postures when the robot is temporarily stationary during the process of conveying molten metal from the furnace to the sleeve.

4. The die-casting machine according to claim 3, characterized in that, The first posture is a metering posture in which the remaining molten metal in the extracted molten metal is returned to the furnace. The second posture is a pouring standby posture near the sleeve, ready to be poured.

5. The die-casting machine according to claim 1, characterized in that, The controller is configured to adjust the injection action that extrudes the molten metal from the sleeve into the die based on the calculated amount of molten metal.

Images