УДК 620.179.16
AN ALGORITHM FOR THE ESTIMATION OF THE QUALITY OF THE SPOT WELDS
Gediminas Genutis1, Elena JasiUniene1, Ruth Sanderson2 1Ultrasound Institute, Kaunas University of Technology, Studentu St. 50,
Kaunas, Lithuania 2TWI Ltd E-mail: elena.jasiuniene@ktu.lt
Abstract — Spot welds are widely used in industry, especially in automotive, so welds must be reliable. One of the approaches to estimate the quality of the spot welds is to use ultrasonic testing. Spot welding is used in automotive repair industry as well. However spot weld testing device, dedicated for the automotive repair industry, is not available. The objective of this work was to develop an algorithm for estimation of the quality of the spot weld, to be used in a simple device in automotive repair industry. The algorithm developed identifies the peaks in the acquired signal and separates the main peaks (ultrasonic signal reflected from the backwall of the weld) from intermediate peaks (which identify the defects in the weld or undersized weld). Then, according to the peaks distribution and peak amplitude variations in the signal the quality of the weld is determined and automatic pass or fail indication given. The algorithm was verified using the finite element model of the good weld, undersized weld and a weld with a pore.
Keywords: ultrasonic testing, spot weld, non-destructive evaluation, quality control, automotive.
АЛГОРИТМ ОЦЕНКИ КАЧЕСТВА ТОЧЕЧНОЙ СВАРКИ
Гедиминас Генутис1, Елена Ясюниене1, Рут Сандерсон2 Институт ультразвука, Каунасский технологический институт ул. Студенту, 50, Каунас, Литва 2TWI Ltd
Точечную сварку широко используют в промышленности, особенно в автомобильной, поэтому сварные соединения должны быть надежными. Один из способов оценки качества сварных соединений состоит в использовании у. з. контроля. Точечную сварку применяют также в авторемонтной отрасли, однако в настоящее время нет специализированного дефектоскопа для этой отрасли. Цель работы — разработать алгоритм для оценки качества точечной сварки, который можно использовать в простом дефектоскопе. Алгоритм идентифицирует пики анализируемого сигнала и разделяет основные пики у. з. сигналы, отраженные от донной поверхности) от пиков в промежуточной области (которые идентифицируются как дефекты в сварке или как область контакта уменьшенного размера). В зависимости от распределения пиков по времени и соотношения их амплитуд автоматически выдается сигнал годности или негодности объекта. Алгоритм был проверен, используя метод конечных элементов, на моделях годного сварного соединения, соединения с областью контакта уменьшенного размера и соединения с порой.
Ключевые слова: у. з. контроль, точечная сварка, неразрушающая оценка качества, автомобильная промышленность.
INTRODUCTION
Spot welds (Fig. 1) are widely used to join the sheets of metal of automotive components. Approximately 20 % to 25 % more spot welds than needed are made in order to guarantee the integrity of the final product [1]. In order to ensure the quality of the spot welds, they must be inspected. Ultrasonic non-destructive methods for evaluation of spot welds are used in automotive industry for many years [2]. However there is no available ultrasonic equipment, which could be used in automotive repair industry for the inspection of the spot welds, when car repairs are performed. For automotive repair industry the simple device, giving automatic pass or fail indication is needed. The pass/fail criteria are based on the weld size (diameter and thickness) and presence of defects in the weld. The weld is good, when there are no defects in the weld, it is not smaller than defined minimum nugget diameter and the thickness of the weld is more than 70 % of parent metal thickness [2].
Fig. 1. Spot weld.
In automotive industry for the evaluation of the quality of the spot welds ultrasonic pulse echo method is used. As thicknesses of the welded sheets are small (0.5—3 mm), high frequency transducers are used for the evaluation of the spot welds. Most frequently 15—20 MHz frequency transducers are used [3], [4]. In order to determine the diameter of the weld, the diameter of the probe must correspond to the diameter of the weld. The diameter of the probe must be approximately equal to the smallest allowable weld diameter [5]. So it has to be determined, if the tested weld is of the same diameter as the transducer, or smaller.
Classification of spot welds is based mainly on evaluation of the A-scan taking into account: reflections from the backwall of the welded structure; the appearance of intermediate echoes from the interface between the sheets in the case of bad welds or embedded defects and the thickness of the weld [2], [3], [4], [5], [6], [7]. However, in order to interpret the results correctly, the NDT specialists are needed.
The objective of this work was to develop an algorithm, which would give simple pass or fail indication depending on the quality of the weld. It has to automatically identify the main reflections (from the backwall of the weld) and intermediate reflections (in the case of the too small weld or defects in the weld) and to determine the parameters of the reflections (time and amplitude). When this information is available, the thickness and the approximate size of the weld can be determined, and information about the weld quality is obtained — if no intermediate reflections are found, then the weld can be identified as good.
AN ALGORITHM
An algorithm for automatic identification and separation of the main and intermediate reflections and determination of their parameters was developed.
Reflections from the backwall of the weld and intermediate reflections can be identified using the following steps, when the A-scan signaly(k) (Fig. 2a) is available, where k stands for a sample number:
1. The envelope (Fig. 2b) of the analysed signal y(k) (Fig. 2a) is calculated using the discrete Hilbert transform:
y* (t ) =
I N-1 / N
II y (k [1 -(-1)k-1 ] coth)
(k -t )-v ' N
(1)
where coth is the hyperbolic cotangent, N is the length of the analysed signal in samples, t is the sample number of the Hilbert transform, and k is the sample number of the analysed signal.
2. The obtained signal is normalized according to the maximum value of the signal:
y (t)
Уьп (f) = ■
max
{Л (t)}'
(2)
3. The expected thickness of the weld is converted into samples using the ultrasound velocity vu, the sampling period At and the expected weld thickness d:
2d At ■ v..
(3)
4. In order to find multiple reflections in the weld the given signal y(t) has
to be divided into equal segments, which correspond to the approximate thickness of the weld as shown in Fig 3. For that the signal must be expanded to the length
1.0
0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8
-1.0
П-Г-1-1-1-1-Г-1-г
1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 Samples
b
1.0
0.9 0.8 0.7 0.6 0.5 0.4 0.3 2 0.1 0
0
1 1 1
.....» ......i-
..... ...... ... ...... r......
...... ......i- ----- .....
L j ik^y V\ J k v/\J
1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 Samples
Fig. 2. Original signal (a) and its envelope (b).
ns =
a
1.0
0.9 0.8 0.7 0.6
s
a 0.5
s
0.4 0.3 0.2 0.1
0
1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 Samples
Fig. 3. Clipped version of signal segmentation.
1 I 1
1 2 : 3 r
4- - -i-*■< -►
n s n. . 1 n s ns
, r
:
..... 1 :
'1 ьА Livj у
which can be divided by ns without remainder. This can be achieved using the following equation:
N =
n n.
(4)
where [ ] denotes the operator which rounds towards the higher integer, n is the length using:
length of the signal yhn(t). Then, zeros are appended to the end of the signal yhn(t)
v {t)_{y*n(t), ift<n; (5)
yhn (t )"[0, if n < t < n . (5)
5. In order to find the main backwall reflections the indices of the elements which have the highest amplitude are found in the matrix M in each row:
lj = arg k{max{M(kJ)}}, k = 1...ns, j = 1...r. (6)
6. The positions of the main reflections (from a backwall of the weld) qj are shown in Fig. 4 and can be estimated using the following equation:
q = lj + (j - 1)xns. (7)
7. The signaly' (t) is divided into equal segments hereby creating the matrix M with ns columnsnand r rows as shown in Fig. 3, where
N
r = -. (8)
ns
8. As can be seen from Fig. 4 intermediate reflections y usually occur between two main reflections. The amplitudes and the arrival timesj of them should be determined also. So, the most appropriate way to search for intermediate reflections
Samples
Fig. 4. Clipped version of the signal with marked main and intermediate reflections.
is in between two main reflections. First of all, the start and the stop positions for search of intermediate reflections must be determined:
A = min + 7, 1 - y}; (9a)
Bi = max{^ + ^ j - y}; (9b)
8. = arg{max{y'Jk)}}, k = A ...Bp (9c)
where A and B are the start and end points for the search of intermediate reflections in jeach sejgment of the signal and 8 is the sample number in the segment with the highest amplitude value. 1
The intermediate reflections are identified only if their amplitude is higher than defined threshold. The exact position of intermediate reflection y is then calculated using following equation: 1
y i = ф i +
Пл. 3
(10)
where [ J denotes the operator which rounds toward the nearest lower integer.
9. According to the time delay between successive reflections the thickness of the spot weld can beestimated. This is accomplished using the following equation:
H =
2 • v,
(11)
where x is the time delay between successive reflections.
Using the developed algorithm the backwall and intermediate reflections are identified and their parameters determined in the acquired data. The obtained parameters of these reflections allow classifying spot welds into good and bad.
T
ANALYSIS OF THE REF
Для дальнейшего прочтения статьи необходимо приобрести полный текст. Статьи высылаются в формате PDF на указанную при оплате почту. Время доставки составляет менее 10 минут. Стоимость одной статьи — 150 рублей.