GB2062896A - Fault detection in sequential operations - Google Patents

Abstract

A method of detecting faults in a sequential operation which comprises a succession of operation systems, includes continually monitoring the succeeding states of the operation and the time during which each state is maintained, by means of a transducer connected through an interface to a digital computer system, identifying by comparison with the observed sequential operation each state or a combination of states of the operation having a significant effect on the operation, together with the tolerance limits with respect to the time associated with one or more of each such state or combination of states, preparing from this identification a monitoring programme of the operation and running said programme in said computer system, said monitoring programme being adapted to compare the monitored states with those in the monitoring programme and to output an error or fault notification should disparities occur between the monitoring programme and the monitored information.

Description

SPECIFICATION Fault detection in sequential operations This invention relates to the detection of faults in sequential operation. The operations may, for example, be those of a mechanical, electrical, fluidic, pneumatic or electronic system, or any of these in combination. The invention is particularly suitable for examining complex or rapidly operating systems where there is difficulty in identifying precisely the source of a fault or malfunction in the system without removing the system from normal operation and carrying out extensive test procedures.

The problem to be solved is therefore that of finding a technique for detection of faults or malfunctions in a sequential operation which does not unduly interfere with the normal use of the system.

According to one aspect of the invention, a method of detecting faults in a sequential operation which comprises a succession of operational systems, includes continually monitoring the succeeding states of the operation and the time during which each state is maintained, by means of transducer connected through an interface to a digital computer system, identifying by comparison with the observed sequential operation each state of combination of states of the operation having a significant effect on the operation, together with the tolerance limits with respect to the time associated with one or more of each such state or combination of states, preparing from this identification a monitoring programme of the operation and running said programme in said computer system, said monitoring programme being adapted to compare the monitored states with those in the monitoring programme and to output an error or fault notification should disparities occur between the monitoring programme and the monitored information.

Additionally the method may include monitoring the time during which at least one succession of specified states or combination of states is maintained, identifying the tolerance limits with respect to the time associated with each such succession of specified states of combination of states, and including such identification in the monitoring programme of the operation.

In carrying out the method, it may be necessary to identify by comparison with the observed sequential operation, the commencement of the operation.

The monitoring programme may suitably be run in the computer system concurrently with the monitoring of the succeeding states of the operation, although in certain circumstances, exact concurrence may not be necessary.

Each state may be defined by an analogue level and/or a digital level. In most circumstances it is preferable to operate with a digital level, since this simplifies the computer system to be used. The method is particularly suitable for use with microprocessor systems and can be very simple to programme.

Certain of the state changes which are monitored will not have any significant effect on the operation being carried out. The identification of the significant and the non-significant changes of state is an important step in the method and will depend, in each case, on the nature of the process and the specific inputs which are being monitored.

The function of the interface is to translate the signals from the transducers to a suitable input level for the computer system. The interface must not load the transducer signals in any way which might affect the transducer outputs. The interface will also be required to reverse the polarity of signals from certain transducers to ensure that each signal corresponds to a positive known situation.

One embodiment of the invention will now be described by way of example only.

A rolling mill for rolling steel ingots into billets comprises: a) An entry table adapted to receive one, or a pair of, hot ingots from soaking pits in which they have been heated or maintained hot, b) a parting device which operates to permit only one of the two ingots to move forward for rolling (except when carrying out "tandem rolling" when two ingots are rolled together nose-totail), c) a weighing zone for weighing each ingot separately prior to rolling, and d) a rolling mill train which may include both reversing and continuous mills.

Hot metal detectors (radiation-sensitive cells) are positioned both before and after the ingot parting device. These detectors sence the presence or absence of an ingot at these positions.

The operation of the rolling mill is normally determined by an electronic digital controller and follows a predetermined sequence. Electrical signals from transducers situated at various locations in the equipment described above provide the controller with the information required for control. In the presently described arrangement, any analogue signals from the transducers will be converted to digital inputs suitable for the controller.

The signals from the various transducers can be listed as follows: Transducer Signal Inputs (1) Lower Parting Device Relay.

(2) Ingot Parting Device Failed.

(3) Entry Table Stationary.

(4) Raise Parting Device.

(5) Parting Device Raised.

(6) Run Entry Table at 1.5 m/s.

(7) Lower Parting Device.

(8) Parting Device Lowered.

(9) NO INPUT (disconnected).

(10) Dual Working. (This is set manually by the mill operator.) (11) Weighing Zone Empty.

(12) Ingot Nose at First Position Before Ingot Parting Device. (Determined by Hot Metal Detector.) (13) Raise Parting Device Relay.

(14) Ingot Nose at Second Position After Ingot Parting Device. (Determined by Hot Metal Detector.) (1 5) Stop Entry Tables.

(16) General Reset.

In this example, an interface is employed between the controller and a microprocessor. The interface translates the transducer signals entering the controller to a 20mA drive suitable for the microprocessor. The interface is chosen so that it only lightly loads the input signals to the controller. If no controller had been present, the interface could have been used directly with the inputs from the transducer signals (with analogue/digital conversion if necessary). In both situations, each input to the microprocessor will be digital, either logical 0 or 1, indicating that the input is in one of two conditions. Means are incorporated in the interface for, if necessary, reversing the polarity of certain input signals from the transducers to ensure that they each correspond to a 1 (or "ON") input for the statements in the previous list.

The microprocessor system incorporating the technique for detecting faults or malfunctions in the operation of the rolling mill sequence and runs the monitoring programme includes:- a) A function keyboard to control the monitoring operation and to facilitate the entry of the monitoring programme.

b) A printer and visual display to notify, display and provide hard copy (i.e. printed) records of any error or fault should they occur in the operation of the rolling mill sequence.

c) A cassette recorder to provide storage of the monitoring programme.

A TRACE function is first brought into operation in the microprocessor system and continuously operates until manually stopped. This function simply monitors the conditions of each input, and compiles a list of the input changes and times the duration of each input condition to a resolution of 1 millisecond. A hard copy of this list is printed on the printer.

When the sequence has been monitored sje"I times (this can be checked by visual comparison with the mill operation, or the mic:CrSr.essor can simply be left to operate in TRACE for, say, one hour), the TRACE function is stopped.

A typical section of a list for the example concerned is as follows: *LINE TIME 1-4 5-8 9-12 13-16 DURATION **STATE 1 12:20 1-1- --11 1-11 --1- 20.336 ) 2 12:20 1--- ---1 1--1 ---- 0.017 3 12:20 ---- -1 -1 1 -1 ---- 0.414 ) 4 12:20 ---- -1-1 1-11 ---- 1.879 5 12:20 ---- -1-1 1-1- ---- 0.453 6 12:20 -- ---1 1-1- 1 -1 - --1- 1.353 ) 7 12:20 --1- ---1 1-1- --1- 0.170 ) O 8 12:20 --1- --11 1-1- -11- 0.024 9 12:20 1-1- --11 1-1- -11- 5.840 ) 10 12:21 1-1- --11 1-1- --1- 31.536 11 12:21 1--- ---1 1-1- ---- 0.015 12 12:21 ---- ---1 1-1- ---- 3.511 13 12:21 ---- ---1 11-- ---- 2.482 14 12:21 ---- -1-1 11-- ---- 7.588 1 15 12:21 ---- ---1 11-- ---- 0.739 2 16 12;21 ---- ---1 11-1 --1- 1.341 ) 3 17 12:21 ---1 ---1 11-1 --1- 0.010 ) 18 12::21 --11 ---1 11-1 1-1- 0.607 19 12:21 --11 ---- 11-1 1-1- 3.205 ) 4 20 12:21 ---1 1--- 1111 1--- 0.010 21 12:21 ---1 11-- 1111 1--- 4.213 5 22 12:21 ---1 1--- 1-11 111- 1.342 23 12:21 ---- 1-1- 1-11 111- 0.015 ) 6 24 12:21 --1- 1-1- 1-11 -11- 0.011 25 12:21 1-1- 1-1- 1-11 -11- 0.412 7 26 12:21 1-1- --1- 1-11 -11- 3.163 8 27 12:21 1-1- --11 1-11 -11- 0.022 9 *Line numbers.

The line numbers have been added for ease of reference in the text.

**State numbers.

State numbers have been added to indicate the separate states in the operation of the rolling mill sequence.

In this list each line represents a differing input condition from that represented in the line below. Inputs 1 to 1 6 are presented in four blocks (for ease of reading) from left to right across the print-out.

The column marked "duration" indicates the time in seconds during which the operation remained with a particular input condition. The two conditions for each input (the "OFF" and the "ON" conditions) are represented respectively by - or 1.

The list set out above would repeat itself several times, commencing with the top line, although for every repetition the time associated with each input condition will be different from that shown in the list section above due to variations in the operation of the rolling mill sequence.

The plant engineer or technician now compares his print-out list with his observation of the sequence of operation of the rolling mill. He recognises that line 14 of the print-out corresponds with the usual starting point of the sequence (i.e. input 6 shows entry table running at 1.5 m/sec; input 11 shows weighing zone loaded, and the other inputs are consistent). At line 1 5 the entry table has ceased running at 1.5 m/sec and at line 16 the ingot nose is at the first position before the ingot parting device and the entry table has stopped. Each of these steps are recognisable as having a significant effect on the operation of the rolling mill sequence, and are therefore selected to indicate to the monitoring programme that a change of state has occurred in the operation of the rolling mill sequence.

At line 17, input 4 only has changed with respect to line 16. The means that the parting device has been raised (duration 10 milliseconds) but that no other changes have occurred. This is identified as a non-significant change of input condition on the state of the operation of the rolling mill sequence. It will be appreciated that the identification of significant and nonsignificant changes of input condition on the states of the rolling mill operation will depend upon the details of the operation being monitored, and the effect of each state of the sequence (known to the plant engineer) on the overall operation of the rolling mill sequence.

Each change of state is identified by examination of the operation of the rolling mill sequence in this manner, and the tolerance with respect to time determined for each significant state or combination of states by examining the times for repeated sequences. The tolerance will usually be a maximum and minimum time over which the same state or combination of states should exist.

In this example it was found that starting from the top of the list of states, the first significant change occurred from line 1 3 to line 14, the second from line 14 to line 15, the third from line 15 to 16, the fourth from line 17 to 18, the fifth from line 20 to line 21, the sixth from line 21 to line 22, the seventh from line 24 to line 25, the eighth from line 25 to line 26 and the ninth from line 26 to line 27. From line 1 to line 1 3 there were no state changes which significantly affected the operation. The significant states, commencing with State 0, have been added in the right hand column of the list set out above.

The Plant engineer is now in a position to write a programme taking into account the maximum and minimum times for the duration of each state or combination of states. The programme format is as follows:

STATE 0 WHEN INPUT 6,8=1 ,4,5,13=O GOTO STATE 1 1 STATE 1 WHEN INPUT 6=0 GOTO STATE 2 STATE 2 WHEN INPUT 12,15=1 ,6=OGOTOSTATE3 STATE 3 WHEN INPUT 3,4,13=1 6=0 GOTO STATE 4 STATE 4 WHEN INPUT 5,6 = 1 , 8=0 GOTO STATE 5 1 STATE 5 WHEN INPUT 14,15=1 ,6=OGOTOSTATE6 STATE 6 WHEN INPUT 1,3,7 = 1 6=0 GOTO STATE 7 STATE 7 WHEN INPUT 1,3,7 = 1 , 5,6 = 0 GOTO STATE 8 STATE 8 WHEN INPUT 1,3,7,8 = 1 , 5,6 = 0 GOTO STATE 9 STATE 9 1WHEN INPUT 14=0 GOTO STATE 0 Each state is defined in sequence in the monitoring programme.For example state 0 can be defined as: STATE O IF INPUT 2,16 = 1 PRINT EVENT LOG WHEN INPUT 6,8=1 4,5,13=0 GOTO STATE 1 This means that when input 6 and 8 are 1 (or "ON") and inputs 4 and 5 and 1 3 are 0 (or "OFF"), then state 0 has ended and the monitoring programme should proceed to state 1. This permits the monitoring programme to follow the rolling mill sequence as it starts the entry table running at 1 m/s indicated by input 6 but with checks that the parting device is lowered (i.e.

input 8) and is not being raised (i.e. inputs 4,5, and 1 3). In addition it inputs 2 and 1 6 are 1 (or "ON") then a list of the most recent input changes and states is printed in an EVENT LOG on the printer, this is an error condition resulting in a general reset (i.e. input 1 6) as a result of parting device failure (i.e. input 2) and the historic list can be used by the plant engineer to investigate the cause of such failure.

In a second example state I can be defined S STATE 1 IF TIME IN STATE < 7.0 PRINT EVENT LOG IF TIME IN STATE > 8.0 PRINT EVENT LOG WHEN INPUT 6 = O GOTO STATE 2 As described for state 0 above this state begins when the entry table runs at 1.5. m/s. In this example state 1 ends when input 6 is O (or "OFF") and the monitoring programme should proceed to state 2, but if the transition occurs before 7 seconds or after 8 seconds in this example an error or fault is indicated and a list is printed out of the most recent input changes in an EVENT LOG on the printer.Therefore because state 1 begins when the entry table runs at 1.5 m/s (i.e. input 6 is 1) and ends when the entry table is not running at 1.5 m/s (i.e. input 6 is 0) the time that the entry table is running is checked by the tolerance of + 0.5 s on 7.5 s as a significant part of the operation of rolling mill sequence.

In a third example states 6, 7 and 8 can be defined as: STATE 6 WHEN INPUT 1,3,7 = 1 , 6 = O GOTO STATE 7 STATE 7 START TIMER 1 INPUT 1,3,7 = 1 , 5,6, = O GOT STATE 8 STATE 8 IF TIMER 1 < 3.0 PRINT EVENT LOG IF TIMER 1 > 4.0 PRINT EVENT LOG INPUT 1,3,7,8 = 1 , 5,6 = O G OTO STATE 9 In this example in state 6 when the entry table is stationary (i.e. input 3 is 1 and input 6 is 0) the lowering of the parting device begins (i.e. input 1 and 7 are 1) and the monitoring duty passes to state 7 where a timer is started (START TIMER 1), to time the lowering of the parting device, when the parting device moves away from the raised position (i.e.input 5 is 0) the monitoring duty passes to state 8 and when the parting device is fully lowered (i.e. input 8 is 1) the monitoring duty passes to state 9, but if the lowering of the parting device is completed in less than 3 seconds or more than 4 seconds then as a result of the checks on timer 1, a list of the most recent input changes is printed as an EVENT LOG on the printer.

Therefore TIMER 1 has enabled a check on the lowering of the parting device by the tolerance of + 0.5 s on 3.5 s on a combination of states. These timers may be used at several places in the monitoring programme to time the significant combinations of states in the operation of the rolling mill sequence.

The monitoring programme is inserted in the microprocessor system using the function keyboard which has single keys for each function. When the monitoring programme is fully developed it can be run concurrently with the monitoring of the succeeding conditions of the inputs in the operation of the rolling mill sequence.

Errors indicated by the monitoring programme may indicate malfunctioning of the mechanical, hydraulic or electrical parts of the rolling mill or of the transducers, but in whichever case the plant engineer will be alerted by the print-out of the list and will be able to see where the error in occurring.

In this particular example a portable form of microprocessor system is applied to monitor operations on an as-required basis, but it may for example be installed permanently on equipment as an automatic fault-finding and diagnostic device.

Claims (6)

1. A method of detecting faults in a sequential operation which comprises a succession of operational systems, including continually monitoring the succeeding states of the operation and the time during which each state is maintained, by means of a transducer connected through an interface to a digital computer system, identifying by comparison with the observed sequential operation each state or a combination of states of the operation having a significant effect on the operation, together with the tolerance limits with respect to the time associated with one or more of each such state of combination of states, preparing from this identification a monitoring programme of the operation and running said programme in said computer system, said monitoring programme being adapted to compare the monitored states with those in the monitoring programme and to output an error or fault notification should disparities occur between the monitoring programme and the monitored information.

2. A method according Claim 1 additionally monitoring the time during which at least one succession of specified states or combination of states is maintained, identifying the tolerance limits with respect to the time associated with each such succession of specified states or combination of states, and including such identification in the monitoring programme of the operation.

3. A method according to Claim 1 or 2 wherein the monitoring programme is run in the computer system concurrently with the monitoring of the succeeding states of the operation.

4. A method according to any one of the preceding claims wherein each state is defined by a digital level.

5. A method according to any one of the preceding claims wherein an output of an error or fault notification by the monitoring programme is arranged to actuate the production of a physical record of a specified succession of monitored states preceding said error or fault notification.

6. A method of detecting faults in a sequential operation substantially as hereinbefore described.