Adaptive ejector valve array

Abstract

A method for providing adaptive control of the flow of a plurality of ejector valves arranged in close proximity to each other in a sorting machine is described. The method comprises measuring a temporal response curve for each ejector valve, comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves, determining from the comparison a predicted flow for each ejector valve, and adapting a drive signal for each ejector valve in dependence on the predicted flow. Temporal response curves may be measured and compared in a calibration (non-sorting) mode, and real-time feedback of the measured temporal responses in a sorting mode may be used to indicate health of a valve and to further adapt the drive signal to provide conformal performance. The number of neighboring ejector valves may also be taken into account.

Claims

1. A method for providing adaptive control of the flow of a plurality of ejector valves arranged in an array in a sorting machine, the method comprising the steps of: measuring a temporal response curve for each ejector valve; comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves, wherein the set of stored temporal response curves in the library is associated with the number of neighbouring ejector valves each ejector valve has in the array; determining from the comparison a predicted flow for each ejector valve; and adapting a drive signal for each ejector valve in dependence on the predicted flow.

2. A method for providing adaptive control of the flow of a plurality of ejector valves arranged in an array in a sorting machine, the method comprising the steps of: measuring a temporal response curve for each ejector valve; comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves, wherein the step of measuring a temporal response curve for each ejector valve comprises measuring a temporal response curve for each operation of the ejector valve over a number of operations of that ejector valve, and averaging the obtained response curves for each operation to produce the measured temporal response curve; determining from the comparison a predicted flow for each ejector valve; and adapting a drive signal for each ejector valve in dependence on the predicted flow.

3. The method according to claim 2, wherein the set of stored temporal response curves in the library is associated with a flow parameter.

4. The method according to claim 3, wherein the flow parameter is associated with at least one electrical parameter.

5. The method according to claim 4, wherein the electrical parameter is (i) current associated with the ejector valve, (ii) rate of change of current associated with the ejector valve, (iii) flux or flux linkage associated with the ejector valve, or (iv) voltage associated with the ejector valve.

6. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting a time delay before the drive signal is applied and/or the length of a portion of the drive signal.

7. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting the applied voltage of the drive signal.

8. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting the height and/or length of a peak at the beginning of the drive signal.

9. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting the height and/or length of a hold period of the drive signal.

10. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting a rate of decay at the end of the drive signal.

11. The method according to claim 2, wherein the step of adapting the drive signal comprises adjusting the depth and/or length of a trough at the end of the drive signal.

12. The method according to claim 2, wherein, a dynamic temporal response curve of an ejector valve is measured during a sorting operation to determine a flow which is compared with the previously predicted flow to provide an indication of the health or condition of the ejector valve.

13. The method according to claim 12, wherein the indication of the health or condition of the ejector valve is used to further adapt the previously determined drive signal.

14. The method according to claim 13, wherein the indication of the health or condition of the ejector valve is associated with the ejector valve opening time, and the further adaptation of the drive signal comprises applying the deviation in time between the measured ejector valve opening time and a previously predicted flow represented by an associated ejector valve opening time.

15. The method according to claim 2, wherein the previously predicted flow is updated by the currently determined flow and timing of each ejector valve.

16. A method for providing adaptive control of the flow of a plurality of ejector valves arranged in an array in a sorting machine, the method comprising the steps of: measuring a temporal response curve for each ejector valve; comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves; determining from the comparison a predicted flow for each ejector valve, wherein the predicted flow for each ejector valve is determined by computing the cumulative error between the mean temporal response curve of the library set and the measured temporal response curve of the ejector valve, and selecting from the library set those temporal response curves which minimise error with respect to the measured temporal response curve to predict the flow; and adapting a drive signal for each ejector valve in dependence on the predicted flow.

17. The method according to claim 16, wherein the predicted flow is further determined by weighting the predicted flow of each of the selected library curves with respect to their respective error.

18. A method for providing adaptive control of the flow of a plurality of ejector valves arranged in an array in a sorting machine, the method comprising the steps of: measuring a temporal response curve for each ejector valve; comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves, wherein the set of stored temporal response curves in the library is associated with temperature, and the ambient temperature of an enclosure housing the ejector valves is measured and used in the selection of library curves; determining from the comparison a predicted flow for each ejector valve; and adapting a drive signal for each ejector valve in dependence on the predicted flow.

19. A sorting apparatus comprising a plurality of ejector valves arranged in an array, and having at least one processing and control module adapted to perform the method of claim 2.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

(2) FIG. 1a illustrates an ejector valve array for ejecting a product from a product stream;

(3) FIG. 1b is another view of the valve array of FIG. 1a, depicting the array of ejector valves facing the product stream and ejecting product therefrom;

(4) FIG. 2 illustrates the construction of a typical ejector valve used in the valve array of FIG. 1a;

(5) FIG. 3 is an illustration of an idealised temporal response of the ejector valve of FIG. 2; and

(6) FIG. 4 is a flowchart depicting a method according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

(7) The term electrical or magnetic temporal responses includes electrical responses to signals of magnetic origin.

(8) FIG. 1a is a perspective view of an embodiment of an ejector array housing 10 having a plurality of ejector valves 12a; 12b; 12c; 12d; 12e in close proximity to each other. The ejector valves 12a; 12b; 12c; 12d; 12e are connected to a processing and control module 14 via link 16. Whilst the number of ejector valves shown (for simplicity) is a one dimensional array of five, those skilled in the art will appreciate that many more ejectors in two or three dimensional array configurations may be employed. For example, in sorting machine applications, it is not uncommon for array configurations having 64 or even 786=468 ejector valves in an array to be utilised.

(9) The housing 10, associated ejector valves 12a; 12b; 12c; 12d; 12e and processing and control module 14 and link 16 are part of a sorting machine (not shown). The sorting machine provides a material stream or product stream 20 for sorting which may comprise individual particles, pellets or units 20a; 20b; 20c (for example, rice grains or plastic pellets). The ejector valves 12a; 12b; 12c; 12d; 12e are positioned so as to intercept the product stream 20. Typically, such a product stream 20 may have a lateral width of 30 cm, across which, for example, an array of equally spaced 64 or more ejector valves 12a; 12b; 12c; 12d; 12e is arranged to enable sorting as shall now be described with the aid of FIG. 1b.

(10) FIG. 1b is a side view of the arrangement of FIG. 1a showing, for purposes of clear illustration, a single ejector valve 12a arranged to face product stream 20. The product stream 20 follows trajectory 22 into pathway or bucket or hopper 24 if the ejector valve 12a is not triggered by the processing and control module 14.

(11) FIG. 1b also shows the trajectory 30 taken by a particle 20a of product stream 20 when the ejector valve 12a is triggered to eject a fluid F (in this example air) by the processing and control module 14 providing an adapted timing signal 28. In this case, the unit or pellet 20a is directed by fluid flow F (such as an air puff) to an alternative pathway or bucket or hopper 26. Pathway, bucket or hopper 26 may designate reject material, or in other applications may designate a particular type of material that is desirably sorted from the stream 20 for alternative processing.

(12) Whilst FIG. 1b illustrates buckets or hoppers 24, 26, it will be appreciated that these are schematic only, and in an industrial automated environment the sorting streams could feed into conveyor belts or other apparatus providing pathways or flows for further suitable processing.

(13) The sorting machine comprises vision or colour or shape detection systems, as well known in the art, to provide a trigger signal to the processing and control module 14. For example, an off-colour (non-white) grain of rice may be detected by such a system slightly upstream of the ejector array, and the trigger or reject signal is subsequently input to the processing and control module 14. The processing and control module 14 is thus triggered to generate a temporal drive signal 28 to the ejector to eject the reject particle or pellet from the stream 20 to bucket or hopper 26.

(14) The temporal drive signal 28 is adapted for each individual ejector valve so as to allow for a predicted flow F for each individual ejector valve 12a; 12b; 12c; 12d; 12e.

(15) The ejector valves 12a; 12b; 12c; 12d; 12e in the array 10 preferably have nozzles directed towards particular target areas of the product stream 20, and are connected to a source of air under pressure via a manifold.

(16) Those skilled in the art will also appreciate that, in some sorting configurations and applications, the dimensions of the product being sorted (for example, rice grains or other similar sized grains) dictates the spacing and nozzle size of each ejector.

(17) In such applications the ejector nozzles 12a; 12b; 12c; 12d; 12e may be grouped themselves and the group served by a shared actuator (for example a pneumatic actuator) and connected to the actuator via plastic tubing due to space constraints.

(18) In some sorting applications, the ratio of nozzles to actuators may be 4:1.

(19) A typical construction of an individual ejector valve 12a; 12b; 12c; 12d; 12e is shown in FIG. 2. The ejector valve 12a; 12b; 12c; 12d; 12e comprises a housing comprising two parts 41, 42 held together by screws 43 and defining a valve chamber 44 which receives pressurised air from a source (not shown) through an inlet port 46. The output port 48 from the chamber 44 is connected to the chamber 44 through an output duct body 50 upon the end face of which is a valve seat 51. The valve seat 51 is closed by a valve plate 52. The plate 52 is formed in a magnetizable material, and is withdrawn from the valve seat 51 to open the valve by selective activation of an electromagnetic assembly 54. The electromagnetic assembly 54 includes an electromagnet which is activated and deactivated by the application of an electrical drive signal.

(20) The temporal response typical of an ejector valve 12a; 12b; 12c; 12d; 12e of the type shown in FIG. 2 is shown in FIG. 3, in which the upper chart depicts a trigger or reject signal provided by the aforementioned vision control system.

(21) The middle chart of FIG. 3 shows a drive signal 28 subsequently provided by the processing and control module 14 via link 16 to the electromagnetic assembly 54 of the ejector valve 12a; 12b; 12c; 12d; 12e.

(22) The lower chart of FIG. 3 indicates the physical temporal response of the ejector valve 12a; 12b; 12c; 12d; 12e measured by, in this embodiment, the ejector nozzle outlet pressure of the ejector valve 12a; 12b; 12c; 12d; 12e.

(23) The trigger signal is generated by the aforementioned vision or colour or shape recognition and detection system, and input to the processing and control module 14, which then generates the drive signal 28. The drive signal 28 represents the magnitude and duration of a pulse and the ejector nozzle outlet pressure represents the magnitude of the pressure measured a distance downstream of the output port 48 of the ejector valve by a pressure transducer.

(24) There is a time delay T.sub.r between the trigger/reject signal being applied and the ejector valve 12a; 12b; 12c; 12d; 12e being substantially open (typically defined as a flow greater than 80%-95% of the steady state flow of a fully opened ejector valve 12a; 12b; 12c; 12d; 12e), typically 1-2 ms.

(25) The ejector valve opening time or rise time t.sub.r of the ejector valve is shown in FIG. 3, and represents the time between the ejector valve 12a; 12b; 12c; 12d; 12e being substantially closed (typically defined as a flow less than 20%-5% of the steady state flow of a fully opened valve 12a; 12b; 12c; 12d; 12e) and the valve being substantially open, typically 1-3 ms.

(26) There is a time delay T.sub.f between the end of the trigger/reject signal being applied and the ejector valve 12a; 12b; 12c; 12d; 12e being substantially closed as shown in the upper chart of FIG. 3.

(27) Also shown in the lower chart of FIG. 3 is the ejector valve 12a; 12b; 12c; 12d; 12e close time or fall time t.sub.f of the valve, which represents the time between the ejector valve 12a; 12b; 12c; 12d; 12e being substantially open and the ejector valve being substantially closed, typically 0.5-2 ms.

(28) In one design of ejector valve 12a; 12b; 12c; 12d; 12e, the drive signal has a first current level, here comprising 12 Vdc, pulse width modulated to provide an amp or more, to open the valve 12a; 12b; 12c; 12d; 12e, a hold period at a second, lower current level so as to maintain the valve 12a; 12b; 12c; 12d; 12e in the open position and a reverse current to close the ejector valve 12a; 12b; 12c; 12d; 12e.

(29) The signals shown in FIG. 3 are schematic and do not reflect the initial current peak or the reverse current, or voltage under or overshoots at the end of the pulses as typically found in real world operation.

(30) Accordingly, if the rise and fall times vary for different ejector valves 12a; 12b; 12c; 12d; 12e in an ejector array 10, then the effective time of opening and closing of each ejector valve 12a; 12b; 12c; 12d; 12e with respect to the application of the drive signal will be different for each valve. This difference may depend on ambient operating temperature, the age and machining tolerances of each ejector valve 12a; 12b; 12c; 12d; 12e. Furthermore, in an array 10 of ejector valves 12a; 12b; 12c; 12d; 12e, the firing (opening and closing) of an ejector valve 12b may also, have an inductive electromagnetic effect which may influence the firing of its neighbours 12a; 12c in the array 10.

(31) This means that different ejector valves 12a; 12b; 12c; 12d; 12e will have airflows rising above and falling below a trigger threshold at different times with respect to the application of the same drive signal 28. The trigger threshold is dependent on the product being sorted and is the threshold of airflow above which particles of product 20a; 20b; 20c will be deflected from the product stream 20 away from the sorting trajectory or pathway.

(32) This impacts the performance of the ejector array 10 as a whole, because acceptable particles preceding and following the particle targeted for rejection or redirection might be deflected from the stream 20 in error, as well as or in place of the targeted particle 20a.

(33) It has been found that the temporal response in terms of ejection pressure of an ejector valve 12a; 12b; 12c; 12d; 12e (approximately a few millimeters from the output port 48) of the type of valve shown in FIG. 2 is a function of the gap between the valve plate 52 and the valve seat 51 (the valve gap).

(34) In addition, this valve gap can be predicted by current and voltage changes with respect to time (temporal response) in the electromagnetic assembly 54 of that ejector valve 12a; 12b; 12c; 12d; 12e. Hence, the drive signal 28 can be adapted for each ejector valve based on the predicted temporal response, with real-time operation updating the drive signal 28 to further adapt.

(35) In one embodiment the adaptation to the drive signal provided by the processing and control module 14 comprises measuring, in a calibration phase, a temporal response curve for each ejector valve prior to operating the machine in a sorting operation, comparing each response curve with a library set of previously stored response curves and determining a predicted flow to provide an adapted timing or drive signal to that ejector valve.

(36) In order to provide calibration in this embodiment, an electrical drive signal 28 is applied to each of the electromagnetic assemblies 54 of the ejector valves 12a; 12b; 12c; 12d; 12e. The same signal may be applied to each assembly 54. The applied drive signal 28 may, for example, be suitable for causing the electromagnetic assemblies 54 of each ejector valve 12a; 12b; 12c; 12d; 12e to cycle through firing (i.e. opening and closing).

(37) Then, in order to calibrate the valve assemblies in the array with respect to each other, for the rise time of the valves, the following calibration measurements can be undertaken.

(38) a. The time taken for the current I in each electromagnetic assembly 54 to rise to a given threshold in response to the application of a drive signal 28 is measured. This may occur at the start of the current rise in the assembly 54 and the threshold may then be set so that it is reached before the valve plate 52 has moved off the valve seat 51. This ensures that the EMF effect of the movement of the valve plate 52 is removed from the measurement. The current in each coil can be measured by a current meter in the respective electromagnetic assembly 54 and the measured signal fed back to the processing and control module 14;

(39) b. The time at which the rate of change of current with respect to time dl/dt reaches a particular value in response to the application of the drive signal 28. The rate of change of current dl/dt in each coil 56 can be measured by a current meter or sensor in the respective electromagnetic assemblies 54 and the measured signal fed back to the processing and control module 14;

(40) c. The current in the electromagnetic assemblies 54, I, or the rate of change of current I in the electromagnetic assemblies 54 with respect to time, dl/dt, at a predetermined time delay after the application of a drive signal; or

(41) d. The value of current I in the electromagnetic assemblies 54 when the flux linkage associated with the core of each electromagnetic assembly 54 reaches a certain threshold or the value of the flux linkage when the current I reaches a certain threshold. The flux linkage associated with the coil 56 of the electromagnetic assembly 54 can be estimated by integration of the voltage applied to the coil 56, less any voltage drops which take account of the resistive losses in the coil 56. The accuracy of the estimation of magnetic flux in the core (and hence flux linkage associated with the main drive coil 56) can be further improved by the addition of a second coil, magnetically coupled to the main drive coil 56, but not used to carry drive current. The voltage induced in this second coil is an accurate measure of the rate of change of flux in the magnetic circuit. Integration of this induced voltage and scaling to adjust for numbers of turns gives a very accurate measure of flux linkage in the electromagnetic assembly 54. The flux linkage to current relationship will be a measure of valve gap or valve plate position. The measured data can also be compared to stored data to produce the required estimation of valve gap.

(42) Hence a set of temporal response curves is obtained which is stored to form a library set of stored curves.

(43) In response to the measurements above, the drive signal to each of the ejector valves 12a; 12b; 12c; 12d; 12e can be adapted so as to correlate the response of the ejector valves 12a; 12b; 12c; 12d; 12e with respect to each other when the sorting apparatus is operational.

(44) In another embodiment the library set of stored temporal response curves is collected for different valve types and configurations, and subsequently stored.

(45) In one embodiment the drive signal 28 is applied to the ejector valves 12a; 12b; 12c; 12d; 12e after a primary time delay after detection of a particle 20a to be ejected. The adaptation to the drive signal for each ejector valve 12a; 12b; 12c; 12d; 12e can be achieved by altering this primary time delay to take into account the predicted temporal response of the ejector valve.

(46) In another embodiment the selection of the set of previously stored response curves is associated with the number of neighbouring ejector valves each ejector valve 12a; 12b; 12c; 12d; 12e has in the array. For example, with reference to FIG. 1a, ejector valves 12a and 12e each have only one neighbour (ejector valve 12b and ejector valve 12d) respectively, whereas ejector valves 12b, 12c and 12d each have two neighbouring ejector valves.

(47) In this embodiment it was found that the temporal response curve of an ejector valve 12a; 12b; 12c; 12d; 12e depends on the number of nearest neighbours that ejector valve 12a; 12b; 12c; 12d; 12e has, which in turn depends on the configuration of the ejector valves 12a; 12b; 12c; 12d; 12e in the housing array 10.

(48) The amount of electromagnetic noise present, and hence reproducible firing of a valve varies and depends upon, in an array, the number of its nearest neighbours.

(49) Hence, in this embodiment, for each ejector valve 12a; 12b; 12c; 12d; 12e, the selection step of previously stored temporal response curves is dependent on the number of nearest neighbours that each valve has.

(50) Typically, the temporal response of each valve is measured in a non-sorting operation or calibration phase as described above. This response is then compared with the appropriate, i.e. neighbour specific previously stored library of temporal response curves.

(51) Any difference is taken into account by processing and control module 14, which, in sorting operation, provides an appropriate driving or timing signal for each ejector valve 12a; 12b; 12c; 12d; 12e. Hence, the influence of valves in close proximity to each other, and their configuration in an array, is predicted and adapted for.

(52) Advantageously, various array configurations may be tested, and library sets stored, to enable calibration of future configurations as per customer desires or wishes.

(53) In another embodiment, the measured temporal response curve for each ejector valve is determined by averaging the measured response curves over a number of operations of that ejector valve. For example, the valve may be operated 5 or 10 times to produce 5 or 10 air fluid outputs (puffs), the temporal response measured for each operation or puff and the average of each operation used in the comparison with the previously stored library set of temporal response curves.

(54) In a further embodiment the predicted flow to provide the adapted timing signal to an ejector valve when sorting is determined by computing the cumulative error between the mean temporal response curve of the library set and the measured temporal response curve of the ejector valve, and selecting from the library set those curves with minimised cumulative error with respect to the measured valve temporal response curve to predict the flow.

(55) In another embodiment the predicted flow is computed by weighting the flow of the library curves by the inverse of their cumulative errors.

(56) FIG. 4 illustrates method steps according to an embodiment of the present invention.

(57) FIG. 4, at the left-hand side, illustrates steps 60, 62, 64, 66 in an embodiment when operating the apparatus in a static or calibration mode.

(58) A temporal response curve is measured 60 for each ejector valve under test [M(dt)]. The measured temporal response curve for each ejector valve is then compared 62 with a library set of stored temporal response curves [C(dt).fwdarw.Lib(dt)]. A predicted flow is then determined 64 for each ejector valve [D(F)], and a drive signal for each ejector valve is adapted 66 in dependence on the predicted flow [A(ds)].

(59) FIG. 4, at the right-hand side, illustrates steps in an embodiment when operating the apparatus in a sorting or dynamic mode.

(60) Whilst sorting, a real-time or dynamic temporal response curve is passively measured 70 for an ejector valve [M.sub.dyn(dt)]. Temperature data may also be optionally measured as part of the dynamic data. This dynamic data is then compared 72 with the previously measured temporal response curves for that ejector valve [C(dt)] to determine deviations in predicted flow. These differences are then fed back 74 to the static mode dataset, which then adjusts the previously determined adapted drive signal 66, if required.

(61) In another embodiment the comparison at step 72 is used to indicate a health parameter [H] at step 76, and this indication may also be fed back 74a to further adapt or modify any previously determined timings, or used to signify such a deviation that the valve should be replaced or serviced.

(62) In the above methods and apparatus, adaptive control of the flow of a plurality of ejector valves arranged in an array in a sorting machine is described. The method comprises measuring a temporal response curve for each ejector valve, comparing the measured temporal response curve for each ejector valve to a library set of previously stored temporal response curves, determining from the comparison a predicted flow for each ejector valve, and adapting a drive signal for each ejector valve in dependence on the predicted flow. Temporal response curves may be measured and compared in a calibration (non-sorting) mode, and real-time feedback of the measured temporal responses in a sorting mode may be used to indicate health or condition of a valve and to further adapt the drive signal to provide conformal performance. The number of neighbouring ejector valves may also be taken into account.

(63) Finally, it will be understood that the present invention has been described in its preferred embodiments and can be modified in many different ways without departing from the scope of the invention as defined by the appended claims.