GB2139767A

GB2139767A - Fault diagnosis in an industrial operation

Info

Publication number: GB2139767A
Application number: GB08309925A
Authority: GB
Inventors: James Arthur Donoghue; Peter Thomas Bowes; Brian William Hollocks
Original assignee: British Steel Corp
Current assignee: British Steel Corp
Priority date: 1983-04-13
Filing date: 1983-04-13
Publication date: 1984-11-14
Also published as: GB8309925D0

Abstract

A method of diagnosing faults in an industrial operation comprises the steps of compiling characteristics of faults possible in the operation together with characteristics of tests which will enable discrimination between the specified faults, compiling selection rules, taking into account fault and test characteristics, for the operational sequence of tests such that a logical order of application of selected tests is obtained and an operational decision tree formed, and operating the decision tree upon a fault occurring in the industrial operation so as to enable the fault to be properly diagnosed.

Description

SPECIFICATION Improvements in or relating to fault diagnosis in an industrial operation This invention relates to fault diagnosis in an industrial operation and more particularly the invention relates to the detection and identification of faults in relatively complex industrial manufacturing plant.

In such plant it is highly desirable that the operators thereof should be able to ascertain rapidly, and therefore at minimum cost so far as operational time of the plant is concerned, the nature and cause, as well as the remedy, of any fault that may arise in operation.

It is an object of the present invention to provide an arrangement whereby such faults can be identified and diagnosed and their remedy rapidly determined.

According to the invention there is provided a method of diagnosing faults in an industrial operation comprising the steps of compiling characteristics of faults possible in the operation together with characteristics of tests which will enable discrimination between the specified faults, compiling selection rules, taking into account fault and test characteristics, for the operational sequence of tests such that a logical order of application of selected tests is obtained and an operational decision tree formed, and operating the decision tree upon a fault occuring in the industrial operation so as to enable the fault to be properly diagnosed.

The fault and test characteristics may be compiled in a computer; likewise the selection rules for the operational sequence of tests may be compiled in the computer so that an operational decision tree is formed therein.

The fault and test characteristics compiled may include, amongst other things, weighting of faults according to their importance, fault frequency probabiltiy, and test difficulty and cost, so as to enable in the operational decision tree, the more significant faults to be diagnosed quickly using the less expensive tests.

As will be appreciated the method herein defined enables the detection and diagnosis of faults in a complex operational and/or control system.

The computer used with the method may be a micro-computer. It may incorporate in permanent memory form the operational system for a decision tree and fault and test data. Data from a tape or floppy disc, for example, may be fed into the computer to comprise the decision tree and the fault and test data. Such data can be generated by a skilled engineer familiar with the industrial operation concerned, or may be derived automatically from a matrix of fault and test characteristics. In addition, or alternatively, data may be fed into the computer by a skilled engineer familiar with the industrial operation concerned.

The data and decision tree may be so formed that a skilled engineer may provide names and descriptions of the faults and tests, such that a relatively unskilled plant operative can use the computer, carry out tests prescribed thereby, and diagnose the fault.

In some cases an interface can be provided between the industrial plant concerned and the computer such that operational signals can be used in the fault diagnosis operation of the computer.

The computer may be portable and battery operated or may be desk mounted and mains operated.

An example of a systematic method compilation of fault and test characteristics using a computer is given herebelow.

Fault In formation Each fault is identified by a FAULT NUMBER, starting from 1, up to a practical maximum determined by the size of the computer, maybe up to 60 for a portable form of computer. Each fault is given a "Fault name" which consists of alphanumeric text and punctuation (up to 380 characters in length for the portable form computer). The FAULT NAME is displayed on the computer screen as a number of lines with scope for the display of various messages such as: a) A description of the fault, so that a craftsman can go and deal with it himself b) A specific repair instruction c) A cross-reference to drawings, technical manual or other information that may be of assistance in making the repair d)An instruction to cali out a specialist.

This choice gives complete control over how each fault condition, once diagnosed, is dealt with.

Two further fault parameters can be defined: a) Fault FREQUENCY; a value in the range 1 to 999 that indicates the relative rate of fault occurrence.

b) Fault COST WEIGHT; a value in the range 1 to 999 that grades the fault according to its impact on production.

Test Information East test is identified by a TEST NUMBER starting from 1 up to a practical maximum of 150 (in the portable form of computer). Each test is given a "TEST NAME" which can consist of up to nine lines, each 47 characters long, of alpha-numeric text and punctuation. A constraint on most TEST NAMES is that they must be worded as questions.

The tests are categorised by a TYPE reference letter. The various test types are described below.

Type Y: A test question Y for YES or N for NO, which can discriminate for or against all possible faults.

Type X: A test question, answer Y for YES or N for NO, which can discriminate for or against some, but no all possible faults.

Type D: A test question, as for type Y or type X, which allows D for DON'T KNOW as an answer.

Type D tests provide assistance for less experienced users.

Type M: A Menu test, or "multiple-choice" test question, where the test question is displayed with up to eight possible answers. A type M test is answered by keying the number (1 to 6 in the portable computer) of the selected answer. In appropriate circumstances a carefully devised type M test can be a very powerful fault discriminator.

Type G: A type G test is the means of access to a "sub-routine" consisting of a group of tests (type Y,X,M or D, as required) applicable to several identical areas of plant.

(Note: Test type G is so named from the computer term "GOSUB" meaning "go to a sub-routine" where a particular sequence of instructions, or in our case, tests, may be repeated several times in differing circumstances).

Type C: A comment or "dummy" test which does not require an answer; it is used to insert a comment between tests or it can be used to increase the space available for a TEST NAME Three further test parameters can be defined: a) test COST WEIGHT; a value in the range 1 to 999 that grades the test according to its "cost" in time, difficulty, need for special test equipment etc.

b) TEST/FAULT MATRIX; a statement of the result of the test given each of the possible faults.

c) type D test PROBABILITY; a value in the range 1 to 99 that represents a subjective assessment of the probability of a D answer to a type D test.

Decision Tree A decision tree is a method of formalising in a diagrammatic ortabularform the process of logical diagnosis. In this invention the decision tree is the prime controlling mechanism for deciding at any stage which fault has been identified or the next best test to perform or what comments are appropriate.

With such an arrangement we have found it is possible to ensure that in operation computer commands can be carried out by typing a single key of the computer keyboard in each instance.

The computer used in the fault detection method can, as hereinabove indicated, incorporate a fault finding programme which can take an operator through a decision tree. Thus on entry into the computer, a sequence of tests will be presented on the screen, the answer given via the keyboard determining the next action to be taken. Conveniently a range of responses can be provided by operation of a single key thus "Y" for YES, "N" for NO, etc., or a numeral in relation to a menu test.

The arrangement as hereinabove described enables a fault detecting arrangement to be set up by a skilled technician or engineer who has the ability and facility, for a given complex industrial operation, to insert into the computer all possible faults and tests for indicating those faults and their symptoms, and the arrangement then subsequently to be utilised by operators of considerably less skill who need only have the ability to answer questions posed from the computer subsequent to its programming by the skilled operator.

In addition the less skilled operator need only be required to utilise a single key for the question and answer sequence. In this way it will be seen that the method as herein above described can be utilised with a very user friendly computer programme.

Claims

1. A method of diagnosing faults in an industrial operation comprising the steps of compiling characteristics of faults possible in the operation together with characteristics of tests which will enable discrimination between the specified faults, compiling selection rules, taking into accountfaultand test characteristics, for the operational sequence of tests such that a logical order of application of selected tests is obtained and an operational decision tree formed, and operating the decision tree upon a fault occurring in the industrial operation so as to enable the fault to be properly diagnosed.

2. A method as claimed in claim 1 wherein the fault and test characteristics are compiled in a computer.

3. A method as claimed in claim 2 wherein the computer used is a micro-computer.

4. A method as claimed in claim 2 or 3 wherein the selection rules for the operational sequence of tests is used by the computer so that an operational decision tree is formed therein.

5. A method as claimed in claim 2,3 or 4 wherein the fault and test characteristics compiled include weighting of faults according to their importance, fault frequency probability, and test difficulty and cost.

6. A method as claimed in any one of claims 2 to 5 wherein the operational system for a decision tree and fault and test data is incorporated in the computer in permanent memory form.

7. A method as claimed in any one of claims 2 to 6 wherein data for the operational system for a decision tree and fault and test data is derived automatically from a matrix of fault and test characteristics.

8. A method as claimed in any one of Claims 2 to 7 wherein data forthe operational system for a decision tree and fault and test data is fed into the computer by an operator.

9. A method as claimed in any one of claims 2 to 8 wherein an interface between the industrial plant concerned and the computer is utilised such that operational signals can be used in the fault diagnosis operation of the computer.

10. A method as claimed in any one of claims 2 to 9 wherein a portable computer is utilised.

11. A method of diagnosing faults in an industrial operation substantially as described in the example hereinbefore given.