The present invention relates to abrasive machining and in particular, to improvements in or relating to abrasive machining processes.

Abrasive machining processes make use of a water jet or other jet in which a grit of abrasive material is entrained, such as garnet. The jet containing the entrained abrasive material is directed at a workpiece. Using a high velocity jet and an appropriate abrasive material allows material to be removed from the workpiece in a controlled manner. Abrasive machining processes can be used to provide cutting, milling, drilling, turning, shot-peening and other processes for workpieces, with particular benefits when applied to materials which are difficult to cut by conventional means, such as modern ceramics, composites and alloys, diamond, concrete and other materials.

The nature of these abrasive machining processes creates a hostile environment around the workpiece, including high-velocity water droplets (or other fluid deflecting off the workpiece), loose abrasive material and debris removed from the workpiece. Accordingly, difficulties have been experienced in providing accurate monitoring and control of the machining process.

Examples of the present invention provide control apparatus for controlling an abrasive machining process, comprising:

a first input for receiving, in use, sensor signals derived from a workpiece being machined;

a second input for receiving, in use, metrology signals representing the surface of the workpiece;

a first module operable to use one or more evaluation algorithms to process sensor signals received at the first input to evaluate the workpiece surface shape achieved, and to compare the evaluated shape with the shape required, and to cause modification of the instructions to the machining process in response to differences between the evaluated shape and the shape required; and

a second module operable after a machining phase is finished, to receive metrology signals at the second input and to compare the measured shape represented by the metrology signals with the shape required, and to cause modification of an evaluation algorithm in response to differences between the measured shape and the shape required.

The second module may cause modification of an evaluation algorithm by modifying the relationship contained in the algorithm between the depth of abrasion created by the machining process, and the input energy delivered from the machining process. The modification of the algorithm may be by modifying the relationship contained in an algorithm between the depth of abrasion created by the machining process, and the speed of the workpiece relative to the work position at which the machining process creates abrasion. The machining process may be operable to create multiple passes across a workpiece, the second module being operable to modify the evaluation algorithm between passes.

The abrasive machining process may be controlled by instructions derived from a definition of the required shape and a model of the effect of the machining process on the workpiece to be machined to the required shape. The second module may be operable to modify the model of the effect of the machining process in response to differences between the measured shape and the shape required. The machining process may be operable to create multiple passes across the workpiece, the second module being operable to modify the model between passes.

The sensor signals may represent energy delivered to the workpiece being machined. The first module may provide real-time feedback to the machining process. The metrology signals may be used also as the sensor signals. The first and/or second modules may receive, in use, a signal representing the speed of the workpiece relative to the work position at which the machining process creates abrasion. The first module may cause modification by causing a change in the relative speed of the workpiece and the work position.

Examples of the present invention will now be described in more detail, by way of example only, and with reference to the accompanying drawings, in which:

Fig. 1 is a simple schematic diagram of an abrasive machining apparatus for operation in accordance with an example of the invention being described;

Fig. 2 is a general block diagram of the apparatus of Fig. 1 and control apparatus for controlling the machining apparatus; and Fig. 3 is another general block diagram illustrating another manner of controlling the machining apparatus.

Background Fig. 1 illustrates an abrasive machining apparatus 10. In this example, the apparatus performs a milling process on a workpiece 12. The apparatus 10 has a nozzle 14 supplied at 16 with water in which an abrasive grit is entrained. This results in an abrasive jet 18 leaving the nozzle 14 to impact on the workpiece 12 at a work position 20, at which abrasion of the workpiece 12 is created. Arrangements are provided (not shown) to create movement of the nozzle 14 relative to the workpiece 12, allowing the nozzle 14 and the work position 20 to track across the surface of the workpiece 12. This movement is implemented by a CNC (computer numerical control) machine (not shown in Fig. 1 ). The apparatus 10 is controlled by control apparatus (not shown in Fig. 1 ). Control apparatus

Fig. 2 is a block diagram of the apparatus of Fig. 1 , augmented by showing the control apparatus 22 and other features. The milling apparatus 10 provides a water jet milling process 24, as described above in relation to Fig. 1 . Movement of the nozzle 14 and work position 20 relative to the workpiece 12 is controlled by a CNC machine 26. The CNC machine 26 is programmed at 28 from CAD/CAM (computer-aided design/computer-aided manufacture) software 30 by means of an NC (numerical control) program.

The NC program for the CNC machine 26 is created by the CAD/CAM software 30 with reference to a model or algorithm 32 to which the software 30 has access. The model 32 includes a model of the manner in which the workpiece 12 will be abraded by operation of the process 24. In particular, the model 32 takes into account the energy being delivered to the workpiece 12 from the nozzle 14 by the jet 18, and also takes into account the speed of movement of the work position 20 across the workpiece 12. The speed of movement is related to the time for which any point on the workpiece 12 is exposed to the jet 18. As this time increases (as the speed of movement reduces), the depth of abrasion created by the jet 18 will increase.

Sensors

Returning now to Fig. 1 , the apparatus 10 has three sensors, as follows. A pressure sensor 34 measures the supply pressure of water to the jet 18. An AE (acoustic emission) sensor 36 measures transient elastic waves generated within the structure of the nozzle 14, during use. The sensor 36 may be based around a ceramic piezoelectric transducer, for example. Other technologies for AE sensors could be used. Various mechanisms may be involved in the generation of the AE signals, such as impacts between the entrained abrasive particles and the structures of the nozzle 14. The AE signals from the sensor 36 provide a measure of the energy of the jet 18. Another sensor 38 is coupled with the workpiece 12 to derive a sensor signal from the workpiece being machined. The sensor 38 is another AE sensor, in this example, to detect transient elastic waves generated within the workpiece 12, as the workpiece 12 is being abraded by the process 24. The AE signals from the transducer 38 provide a measure of the energy delivered to the workpiece 12, by the jet 18.

The control apparatus 22 has a first input 40 from the workpiece sensor 38. Thus, the first input 40 receives, in use, sensor signals derived from the workpiece being machined. In this example, the sensor signals represent energy delivered to the workpiece 12 which is being machined. A second input 42 receives, in use, metrology signals representing the surface of the workpiece 12. The metrology signals 44 are derived from a measurement process 46 which can measure the surface of the workpiece 12 by an optical scanning or other measurement process.

The control apparatus 22 also has a third input 48 which is a signal representing the speed of the workpiece 12 relative to the work position 20 at which the milling process 24 creates abrasion. The control apparatus 22 also has a fourth input 49 representing the shape required for the workpiece 12, after machining.

First module

Within the control apparatus 22, there is a first module 50 which is operable to provide real-time feedback to the milling process 24, at 52. The term "real-time feedback" is used herein to refer to feedback which occurs while the milling process 24 continues. The first module 50 provides the feedback 52 from a decision process 53 by using an evaluation algorithm or model 54, to which the first module 50 has access.

The first module 50 also has a monitoring process 56 which uses another algorithm 57 to process the signals received at the first input 40 from the workpiece sensor 38. The algorithm 57 contains information about the expected relationship between the energy delivered to the workpiece 12 (represented by the AE sensor signal received at 40) and the expected abrasion of the workpiece 12. The amount of abrasion achieved will increase with the time of exposure to the jet 18, as noted above. Accordingly, the algorithm 57 also takes into account the signal at 48, representing the speed of the nozzle 14 relative to the workpiece 12. Using these inputs and information, the monitoring process 56 of the first module 50 uses the model 57 as an evaluation algorithm to process signals, particularly the sensor signals received at the first input 40, to evaluate the workpiece surface shape achieved by the process 24. This evaluation involves an estimation or prediction of the shape achieved, based on the information contained in the algorithm 57. Thus, the shape achieved is evaluated, rather than measured.

In Fig. 2, this evaluation is schematically indicated by the monitoring process 56 within the first module 50, taking the inputs 40, 48 to provide a signal at 58 to a comparison circuit 60. The signal at 58 represents the evaluated shape of the machined workpiece surface as derived from the inputs 40, 48. The fourth input 49 represents another input to the first module 50. The comparison circuit 60 receives the input 49, representing the required workpiece shape. The evaluated shape is compared with the shape required, represented at the fourth input 49.

When the apparatus is in use, the decision process 53 receives the output 61 of the comparison circuit 60, representing any difference between the evaluated shape derived by the monitoring process 56, and the required workpiece shape represented at 49. When a difference exists, the decision process 53 uses the evaluation algorithm 54 to calculate a correction signal, which is provided to the CNC machine 26 at 52. In this example, the correction signal 52 represents a required change in the speed of the nozzle 14 relative to the workpiece 12. Thus, if the comparison circuit 60 indicates that the workpiece 12 has been insufficiently abraded, the machine 26 will be instructed to slow down the movement of the nozzle 14 relative to the workpiece 12. Conversely, if the comparison circuit 60 indicates that the workpiece 12 has been excessively abraded, the machine 26 will be instructed to speed up the movement of the nozzle 14 relative to the workpiece 12. The first module 50 continuously receives the inputs 40, 48 and 49, while the process 24 is underway, allowing the first module 50 to provide closed loop feedback to the machine 26. This is expected to provide improved accuracy in the abraded shape achieved in the workpiece 12. The accuracy achieved will depend, in part, on the accuracy of the model represented by the evaluation algorithm 54. That is, the more accurately the algorithm 54 represents the effect of any change in the speed of the nozzle 14, the more accurately the feedback at 52 will control the finished shape of the workpiece 12. Second module

The apparatus 10 also includes a second module 64. Unlike the first module 50, the second module 64 is not used to provide real-time feedback during the process 24. The second module 64 is used after a phase of the process 24 has finished. The second module 64 receives the metrology signals 44 at the input 42. The second module 64 also receives the fourth input 49, representing the required finished shape of the workpiece 12. The metrology signals 44 represent measurements of the surface shape actually achieved in the workpiece 12. Consequently, the second module 64 is able to compare the measured shape represented by the metrology signals 44 with the required finished shape of the workpiece 12. This contrasts with the operation of the first module 50, which compares the required finished shape with an evaluated shape achieved, rather than a measured shape achieved. If any difference is found between the measured shape and the shape required, it is appropriate to conclude that there is an error or inaccuracy in the evaluation algorithm 54 or 57, because the process 24 has not achieved the required finished shape, even with the real-time feedback provided by the first module 50, based around the algorithms 54 and 57. Accordingly, when a difference is found between the measured shape and the shape required, the second module 64 sends instructions at 66 and 67 to the first module 50. The instruction 66 causes a modification of the evaluation algorithms 54, and the instruction 67 causes a modification of the evaluation algorithm 57. For example, the comparison achieved by the second module 64 may indicate that the workpiece 12 is being abraded more or less quickly than is assumed by the algorithm 54. In this case, the second module 64 instructs the first module 50 to modify the algorithm 54 to change the parameters relating to abrasion rates. In one example, the algorithm 54 may be modified by modifying the relationship contained in the algorithm between the depth of abrasion created, and the speed of the workpiece 12 relative to the work position 20 at which the process 24 creates abrasion. In another example, the second module 64 may cause modification of the evaluation algorithm 57 by modifying the relationship contained in the algorithm between the depth of abrasion created in the workpiece 12, and the input energy delivered from the process 24 (represented by the input 40 from the AE sensor 38).

Once the algorithms 54, 57 have been modified, if required, a further phase of milling may begin. This further phase will again be controlled by the real-time feedback provided by the first module 50. However, the modified algorithms 54, 57 will now be used, which can be expected to provide increased accuracy in the finished shape achieved for the workpiece 12.

The second module 64 also operates, in use, to modify the model 32 contained within the CAD/Cam software 30. As has been noted above, the model 32 includes a model of the manner in which the workpiece 12 will be abraded by operation of the process 24. If the second module 64 finds any difference between the measured shape and the shape required, it is appropriate to conclude that there is an error or inaccuracy in the model 32, because the process 24 has not achieved the required finished shape when operating in accordance with the instructions derived from the model 32 by the software 30. Accordingly, when differences are found between the measured shape and the shape required, the second module 64 sends an instruction at 68 to the software 30, or directly to the model 32. The instruction 68 causes a modification of the model 32. Operation of the apparatus

When the apparatus 10 is in operation in the manner described above, real-time feedback is provided to the machine 26. This is expected to improve the accuracy of the process 24. However, the feedback from the first module 50 is based, in part, on the output of the workpiece sensor 38 and on the algorithm 57. Consequently, the accuracy of the feedback, and thus the overall accuracy of the process 24, is dependent on the accuracy of assumptions built into the algorithm 57 for interpreting the output of the workpiece sensor 38. Nevertheless, we envisage that the provision of this real-time feedback to the process 24 will result in an improvement in the accuracy of the process 24. In particular, we envisage that using an AE sensor 38 will allow more effective feedback to be created than would be possible with many other types of sensor. For example, optical sensors may be unable to provide a meaningful real-time output in the mechanically and optically harsh conditions to which abrasive machining processes give rise. Thus, while the system described is able to adapt itself in real-time by the operation of the first module 50, the system is also able to use the second module 64 periodically (after machining phases of the process 24) to provide information based on the metrology signals which measure the actual shape achieved in the workpiece 12, and therefore allow periodic recalibration of the model 32 and the algorithms 54, 57. This is expected to improve further the accuracy of the process 24.

Alternative arrangement

Fig. 3 illustrates an alternative arrangement which contains some modifications over the arrangement illustrated in Fig. 2. In Fig. 3, the operation of the second module 64 is as described above in relation to Fig. 2, but the operation of the first module differs, and accordingly, the first module is given the reference 70 in Fig. 3.

The first module 70 operates only between operating phases of the process 24 (i.e. offline). In this example, the metrology signals are also used as the sensor signals to the first module 70. Thus, the first module 70 receives the metrology signals 44, which are derived from the workpiece being machined. The first module 70 uses these for comparison with the fourth input 49, by the comparison circuit 60. In the arrangement of Fig. 3, a variation indicated by the comparison circuit 60 is used by the first module 70, with reference to the evaluation algorithm 54, to create an instruction at 72 which represents a correction to be applied to the NC program 28 for the CNC machine 26. When the process 24 commences another phase, the CNC machine 26 will operate in accordance with the corrected program 28. This allows the program 28 to be modified by progressive iterations, resulting in errors in the final shape of the workpiece 12 being progressively reduced.

Concluding remarks

Many variations and modifications can be made to the apparatus described above, without departing from the scope of the invention. In particular, many different techniques and technologies can be used for implementing the various process steps which have been described. Some steps may be implemented by bespoke hardware or by dedicated or general purpose devices controlled by software or firmware. The algorithms 54, 57 have been illustrated and described as separate entities within separate processes. Other alternatives are possible. The whole of the functionality of the first module 50 may be integrated, particularly by means of appropriate software. It is envisaged that the control apparatus described above may be fitted retrospectively to an existing abrasive machining process, by the provision of appropriate sensors to provide the inputs described above. The methods described can be implemented as software which, when installed on a computer system, is operable to perform the method.

Whilst endeavouring in the foregoing specification to draw attention to those features of the invention believed to be of particular importance it should be understood that the Applicant claims protection in respect of any patentable feature or combination of features hereinbefore referred to and/or shown in the drawings whether or not particular emphasis has been placed thereon.