·Table of Contents
·Methods and Instrumentation
Two approaches to ultrasonic control problem solvingVitaly Davidenko
The ultrasonic control and testing development (ultrasonic NDT or UNDT) within the last decade happened in an atmosphere of progressing intellectualization of ultrasonic flaw detectors at the expense of broad introduction in this branch of instrument making of means of computer science and computer facilities: microprocessors, digital displays, memory devices, progressive software (SAFT, TOFD) etc. On a hum noise of rough process the certain philosophy of ultrasonic NDT intellectualization and methods it of reaching is precisely traced. The main idea consists of releasing the UNDT-operator of the most complex and responsible process of settting-up the ultrasonic flaw detector when object is under the control and to put this function to the device.
Realization of this idea implemented by qualified experts, which beforehand made set-up certain type of flaw detector with pre-known converters on the testing of specific object under the adopted speciation. The outcomes of tunings as appropriate positions of controls with various parameters were entered into flaw detector's memory under numbers, assigned to them, of verification modes. After that NDT-operator was enough only under number of mode call from memory necessary set-up of the flaw detector, which were installed automatically.
As a good idea of flaw detectors intelligence rise collides with essential difficulties both quantitative, and qualitative character. At once there is a problem: how many different modes in NDT device are necessary for storing? The answer is: as it is more - it is better. For the last decade in the answer to the first problem the precise geometrical progression is looked through: quantity of modes in memory of flaw detectors grew under the law 10n (n = 1, 2, 3). What will be further? Whether it is necessary to accumulate core budgets of modes up to sizes 104... 105?
The second problem concerns quality of tunings. What parameters is necessary to store, what their range and digitization? In a cast of initial verification modes could not still actuate parameters of converters, objects under testing (OT) and defects because of so-called "damnations of multiregularity ", requiring increase of storage devices volumes up to the improbable sizes.
The especial difficulty is the account of converter's parameters, which it is necessary to know a minimum of four: prism delay, sensitivity, operating frequency and angle of a prism. Thus it is necessary to take into account such factors, as the same type converter's parameters tolerance, actually converter's individual originality, and also instability of parameters of any converter owing to natural wear during operation. For withdrawal from this straits searched on any other ways:
The technique of intellectualization of ultrasonic flaw detectors by reminder of normal verification modes gradual has developed into a cybernetic technique " of a black box ", distinguished by the statistical approach to study of dependences between inputs and outputs of a "black box ", which ultrasonic flaw detector's electro-acoustic channel in this case is. Due to determining not only testing data-ins (modes), but also output echo-signal's parameters (amplitude and delay) under certain conditions scanning ensured with the well-experienced operators, has appeared a capability of realization of idea of a "black box " by training.
Thus the not statistical approach will be realized, when all testing results are fixed, and strictly selective, when the maximum values of amplitude and appropriate delay of an echo-signal are fixed only. The full set of input and output data fixed at a certain converter's position on an OT, represents one fragment of "black box "description. Such fragments should be very much to make a full description of ultrasonic flaw detector's electro-acoustic channel. The separate description fragments are grouped in indications of equal decay coefficients of ultrasound, and also on an indication of equality of the sizes of artificial reflectors, derivating so-called AVG-curves. On these curves the identification of defects discovered in actual OT is made and their equivalent size is determined.
The present technique of flaw detectors intellectualization allows to decide all UNDT problems without using of electro-acoustic channel's mathematical description (or model) and to avoid of significant errors, intrinsic to it. However thus it is necessary to pay by a stiff high bid of flaw detector's memory training and boundedness of converter's selection. Besides the further improvement of ultrasonic flaw detectors with the given technique calls owing to notorious "damnations of multiregularity". If not to fall outside the limits the requirements to manual UNDT, alternative of training of intellectual ultrasonic flaw detectors on a cybernetic method of a "black box" can be only method of a "grey box". The idea of this method consists that the electro-acoustic channel functionality is partly opened, with further converting of flaw's parameters (sizes and coordinates) into measuring parameters of echo-signal (amplitude and delay) with some unknown parameters of an actual system "flaw detector- converter-object under test" (DCO) wich will be necessary to determine by training (or setting-up) on a method of a "grey box". As the method of flaw coordinates-to-echo delay conversion and back simple and fully linear, main goal of conversions is reflecting area to echo-signal amplitude conversion with allowance of delay and damping in OT.
The definition of unknown parameters of electro-acoustic channel's conversion algorithm (of a "grey box") - ultrasound decay coefficient in OT and sensitivity of DCO system, should be made without any external helps using of internal properties of a DCO system, which should be well known to operator. This setting-up named adaptive.
There are only two of correct DCO system's setting-up internal indications:
Conformity between numerical readout of reflector's size on a certain scale of the equivalent sizes shown by corrected echo-signal's amplitude, and actual size of the artificial reflector in actual OT sample. Using these indications for adaptive setting-up of a DCO system can be carried out by the following concepts basis.
It is possible to consider this expression also as solution of a inverse ultrasonic testing problem, since it installs connection of flaw's equivalent size with all basic testing parameters: by two measured echo-signal parameters (D and T) and two adaptive settings (G and K). Now solution of this problem implements by:
Structuring of input data under the formula of adduced amplitude Q = D + G× T - K, taking into account as well ultrasonic oscillations frequency by adaptive setting-up, dramatically reduces apriority memory volumes of intellectual flaw detectors and makes a procedure of their setting-up easy for any operator. Besides there can be one (and may be a biggest advantages of a "grey box" method) it is a possibility to use with such flaw detector any types of converters, irrespective to their sensitivity, frequency and other parameters. Due to this the requirements to converters are dramatically softened, as well as the necessity of their personalising and cataloguing goes out and accordingly price decreases.
Thus the difference between converters will indicate that they will provide determination of equivalent sizes of defects of specific size on different maximum depth, which thus and is a generalizing measure of their individual sensitivity. The marginal depth of identification of allowed flaws should be a unique criterion of suitability any of converter for testing of particular OT under the established standards. Today classification of converters by their operating frequencies though will save the place due to the strongest influence to limiting of sounding depth, but will lose the domination when converters will be select.
The more in-depth comparison of ultrasonic flaw detectors intellectualization methods based on cybernetic "black box" and "grey box" principles, is shown in table 1. It is clear, that there are differences on ten major parameters which are indicative of fundamental difference of two approaches to creation of intellectual ultrasonic flaw detectors.
|No||Training parameters||Methods of training|
|"black box"||"grey box"|
|1.||Time schedule||Before manufacturing of flaw-detectors series||Immediately before testing of each OT (object of testing)|
|2.||Place||Authorized laboratory of firm-flaw-detectors manufacturer||UNDT operator's working place|
|3.||Trainer||Scientific employee of firm-manufacturer||UNDT operator|
|4.||Manuals||Series of certificated samples from materials with known acoustic properties and a set of artificial different size and type reflectors on different depths||Technological sample from OT material with one or two specific type and size reflectors not requiring of certification|
|5.||Pre-knowledge||Basic technological information, list of parameters, being a subject to measurement and the record-keeping||Table of conformity between adduced amplitude and equivalent flaw size|
|6.||Acquired knowledge||Data tables with some inputs parameters and one output, appropriate to flaw's equivalent size||Adaptive parameters: ultrasound decay coefficient and sensitivity of DCO system|
|7.||Outcomes training registration||Internal device's non-volatile memory (ROM)||Operating memory device - RAM without indication on flaw-detector's screen|
|8.||Learning tool||Converters with individual numbers||Any approaching converters|
|9.||Knowledge check during the learning||None||Two cases:|
1) whether the full amplitudes from equal reflectors are equal one to another?
2) what is the reflector's size in a sample?
|10.||Identification of new object with educational samples (after training activity)||Acoustic properties of new OT are determined: a decay coefficient and speed of a sound; OT identification with an appropriate sample is proceeded.||It is not required, since the technological sample completely corresponds OT, and the learning process is a part of basic activity during testing|
|11.||Learning cost||Very expensive||Very cheap|
|Table 2: The comparative characteristics of cybernetic methods of intellectual ultrasonic flaw detectors training|
|© AIPnD , created by NDT.net|||Home| |Top||