Apparatus for nondestructive on-line inspection of electric resistance welding states and a method thereof

• US 5,764,859 A
• Filed: 10/02/1995
• Issued: 06/09/1998
• Est. Priority Date: 10/01/1994
• Status: Expired due to Fees

• Current Assignee: Orion Metal Company, Ltd.
• Original Assignee: Orion Metal Company, Ltd.
• Inventors: Kim, Sang Hee, Kwon, Jin Oh
• Primary Examiner(s): Hafiz, Tariq R.
• Assistant Examiner(s): Shah, Sanjiv
• Application Number:
  US08/537,507
• Time in Patent Office: 981 Days
• Field of Search: 395/22, 395/97, 219/109, 219/110, 219/117.1, 219/497
• US Class Current: 706/20
• CPC Class Codes: B23K 11/252 using digital means
• CPC Class Codes: B23K 11/253 the measured parameter being a displacement or a position
• CPC Class Codes: B23K 11/258 the measured parameter being a voltage
• CPC Class Codes: B23K 31/006 relating to using of neural networks
• CPC Class Codes: G06N 3/08 Learning methods

Abstract

Apparatus for inspecting an electric resistance welding state including a first electrode connected to a power source, a second electrode connected to another terminal of the power source, and a welding object interposed between the first and the second electrodes. A voltage waveform measuring system includes a first analog-to-digital converter for detecting voltage applied, during a welding process, to both ends of the welding object.

An electrode movement measuring system includes a sensor for detecting a change of a gap between the first and the second electrodes during the welding process, and a second analog-to-digital converter for receiving an output of the sensor.

A computer system which includes a neural network inspection system for receiving outputs from the voltage waveform measuring system and from the electrode movement measuring system is provided.

Claims

1. An on-line nondestructive inspecting apparatus for electric resistance welding, comprising:

an electric resistance welding machine having a first electrode connected to a higher voltage electrical potential terminal of a power source, a second electrode connected to a lower voltage electrical potential terminal of said power source, said first and second electrodes being arranged to have at least two welding objects interposed therebetween;

a voltage waveform measuring system including a noise filtering means and a first analog-to-digital converter;

said noise filter means being arranged for transferring the voltage applied to the welding objects and a low frequency band pass filter to said first analog-to-digital converter;

said voltage waveform measuring system being arranged for providing a voltage waveform measurement output;

an electrode movement measuring system including a sensing means, a second analog-to-digital converter and a digital filter;

said sensing means being arranged for detecting variation of a gap between said first and second electrodes during said electric resistance welding and providing an output;

said second analog-to-digital converter being arranged for transferring said output of said sensing means to said digital filter;

said electrode movement measuring system being arranged for providing an electrode movement measurement output; and

a neural network inspection system arranged for receiving said outputs of said voltage waveform measuring system and said electrode movement measuring system, comprising a multilayered neural network having an input layer, at least one hidden layer, and an output layer;

said at least one hidden layer being interposed between said input and output layers;

said neural network being arranged for said neural network inspection system to learn offline; and

a computer system serving said neural network inspection system, said computer system comprising first software for converting signals digitalized by said first analog-to-digital converter to absolute value signals, and for detecting variations of peak value of said absolute value signals;

said neural network being arranged for using learning data acquired from off line destructive testing of spot welds made by said electric resistance welding machine, correlated with measurements of decrease in welding voltage as contact resistance of the welding objects welded in a learning stage decreases and the gap between the first and second electrodes increases, as welding progresses, and storing said learning data in said computer system; and

said neural network inspection system being arranged for using said learning data, and said variations in peak value and said electrode movement measurement output, for providing a real time determination of integrity of welding being performed by use of said electric resistance welding machine.

2. The apparatus as claimed in claim 1, wherein:

said hidden layer has twice as many nodes as said input layer.

3. The apparatus as claimed in claim 1, wherein:

said sensing means comprises a non-contact-type optical sensor.

4. The apparatus as claimed in claim 1, wherein:

said neural network is further arranged for using learning data as to welding conditions including welding time, kinds of welding machine, and thickness of welding objects, previously stored in said computer system, to learn.

5. The apparatus as claimed in claim 1, wherein said computer system further comprises:

second software for eliminating noise and vibration from the signals digitalized and filtered through said second analog-to-digital converter and said filtering means, by use of a moving average method, for obtaining data with a sampling time corresponding to the wavelength of welding frequency.

6. A method for inspecting and performing an electric resistance welding state, comprising an off-line learning process and an inspecting process,

said off-line learning process including:

(a) a first step comprising the sub-steps of (i) applying a voltage signal to both ends of each of at least two welding objects through first and second welding electrodes, and thereby combining the welding objects;

(ii) converting said voltage signal into digital data;

(iii) detecting variations of peak values of said digital data; and

(iv) storing said detected variations in a computer system;

(b) a second step comprising monitoring the welding state of said welding objects with a destructive inspecting method; and

(c) a third step comprising the sub-steps of (i) repeating said first and second steps;

(ii) applying actual output data and predetermined output data to a neural network contained in said computer system, said actual output data representing variations of said peak values, and said desired output data corresponding to respective welding states; and

(iii) determining weight values of the neural network by using a back propagation algorithm; and

said inspecting process including;

(d) applying said variations detected in said first step to said neural network having the weight values determined in sub-step (iii) of said third step; and

(e) performing an on-line inspection of the welding states of the welded objects;

said inspecting process being repeatedly performed by using weight values determined in said off-line learning process with conditions of said off-line learning process and said inspection process being identically set;

said computer system comprising first software for converting signals digitized by first analog-to-digital convertor to absolute value signals and for detecting variations of peak value of said absolute value signals;

said neural network being arranged for using learning data acquired from off line destructive of spot welds made by electric welding machine correlated with measurements of decrease in welding voltage as contact resistance of the welding objects welded in a learning stage decreases and the gap between the first and second electrode increases, as welding progresses and storing said learning data in said computer system; and

said neural network inspection system being arranged for using said learning data and said variations in peak value and electrode movement measurement output for providing a real time determination of integrity of welding being performed by use of said electric welding machine.