JPS6123966A - Automatic sensory inspection device - Google Patents

Abstract

PURPOSE:To perform sensory inspection automatically and accurately by dividing the frequency of a sample percussion sound through a band-pass filter, and referring to filter outputs of a learning storage memory classified by conforming and nonconforming articles, and sorting conforming sample from nonconforming samples. CONSTITUTION:The percussion sound that a sample 1 generates with a percussion device 2 is collected by a microphone 3, frequency-divided by a BPF4, and stored in an inspection data register 7. The contents of this register 7 are compared by a comparator 8 with outputs of the BPF4 which is stored previously in the learning storage memory 6 and classified by conforming and nonconforming articles to decide and display whether the sample 1 is normal or not on a result display device 9. Thus, the sensory inspection of a casting, etc., by percussion is performed easily and accurately by this automatic inspection.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an automatic sensory test device for automatically performing a sensory test based on tapping sounds on materials such as castings, and determining whether the material is good or bad.

Conventional Structures and Problems Castings are generally referred to as molloys because the crystalline state of the graphite contained in the castings is flaky graphite, which accounts for 17% or more. In order to make the casting sticky and ensure tensile strength, it is best to make it into spherical crystals. Magnesium is currently used as a catalyst to obtain spherical graphite. That is,
1,450°C ~], magnesium is placed in cast iron melted at 500°C and poured into a mold to obtain a casting.

However, since magnesium has a boiling point of approximately 1,100°C, sublimation occurs in molten cast iron, and the magnesium concentration gradually decreases during the process of pouring it into a mold.

When it decreases, graphite crystals become flaky and become a seven-loy casting. Castings that are defective and have little stickiness are used as parts in applications where human life is at stake, but when used in this way, it is necessary to conduct a 100% inspection using sensory testing. There is.

Sensory testing methods include a visual inspection of the state of graphite crystals using a microscope, and a hammering test based on the sound produced by striking a casting. The microscopy method requires considerable skill in applying light and in determining the results, and the testing efficiency is poor due to eye fatigue, so percussion tests are mainly used.

The method of hammering test is to strike the sample with a hammer or the like, and if the pitch is high, it is determined to be good, and if it is low, it is determined to be defective.

This method is correlated with microscopy and is commonly used. Good products make a clicking sound, while defective products make a clicking sound. Intermediate products make a clicking sound.Although there are samples of good and defective products, the sensory test standards are extremely vague.Still. However, since some parts are heavy, there is a problem in that physical stress is required for sensory tests.

Purpose of the Invention The present invention eliminates the problems of the above-mentioned conventional example,
The purpose is to be able to easily and accurately determine whether a casting is good or defective without requiring any skill, and to be able to determine whether the casting is good or defective in the same way even if the shape or weight of the inspected casting is changed. It is.

Structure of the Invention In order to achieve the above object, the present invention captures the tapping sounds of good and defective products with a microphone, and uses the outputs of n bandpass filters as standard bangs to learn and memorize the sounds of good and defective products. Next, the tapping sound of the sample to be inspected is captured by a microphone, and the output data of the same n bandpass filters is matched with the filter output data for good and defective products that has been learned and memorized in advance. This system automatically determines whether the product is good or bad based on physical quantities, whereas conventional selection is done based on intuition and experience.

DESCRIPTION OF THE EMBODIMENTS The configuration of an embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, 1 is a test sample.

2 is a percussion device, 3 is a microphone, 4 is an n-channel bandpass filter, 5 is a mode changeover switch 4 and 6 for switching between learning registration mode and inspection mode, and learning and storing filter output data for non-defective and defective products. memory,
Reference numeral 7 indicates a sample band in the inspection mode; a dust filter output data register; 8 a comparator; this comparator 8 compares the filter data of the non-defective sample and the sample. . 9 is a display device for displaying the comparison results.

In FIG. 1, a sample 1 is struck with a percussion device 2, and the sound received from a microphone 3 is passed through a bandpass filter 4. The bandpass filter 4 is composed of n channels from low frequency to high frequency, but n may be about 8. Further, it is sufficient that the frequency range can cover about 800 Hz to 3K) Iz.

FIGS. 2A and 2H show examples of bandpass filter output data for non-defective products and defective products. Good products make a high-pitched sound,
The output level of the bandpass filter on the high frequency side is high, and the output level of the bandpass filter on the low frequency side is high because a defective product produces a low sound. Since the output data of the bandpass filter 4 has characteristics, this is used to determine whether the product is good or defective.For this purpose, set the mode selector switch 5 to the learning mode in advance, sound a sample of the good product, and Output data for each channel of the bandpass filter is stored in the storage memory 6. When there are many non-defective samples, a learning function is provided so that the average value of all samples can be calculated and stored.

Next, the selector switch 5 is switched to the test mode, the tapping sound of the test sample is recorded in the register 7, and is matched with a previously stored standard slam by the comparator 8 for comparison. At this time, if there are more high-frequency components than in the non-defective product, the product is determined to be non-defective. If the number of low frequency components is higher than that of a good product, it is considered a defective product.

When it is desired to rank a good product, a defective product, or an intermediate product, the data for each rank is learned and stored in the learning memory 6 in advance, and the closest one is determined as a result by click matching, and the result is output to the result display 9. ○ The handling of the sample and the operation of the percussion device can of course be automated, and the output information from the result display 9 can be used to
It is also possible to separate samples into good and defective products.

As described above, the device of this embodiment can automatically classify pre-registered good products, defective products, or their ranks, making it possible to accurately and easily sort castings, which previously relied on skill and intuition. , and has the advantage of being automatic.

Since the hammering sound can still be learned and memorized by oneself, it has the advantage that even if the shape or weight of the casting is changed, good products and defective products can be sorted out by learning anew.

Effects of the Invention The present invention has a queen-like configuration and provides the following effects.

(a) It is more accurate than the conventional sorting based on intuition because the results of hammering sound analysis of good and defective products can be learned and memorized as a standard bang, and the material inspection can be matched and sorted using physical quantities.

(b) Since sample handling, tapping operations, etc. can be automated, the physical burden on humans is reduced.

(C) Since several standard patterns of hitting sounds can be registered, it is also possible to rank products from good to bad.

(d) Since the hammering sound can be learned and memorized, even if the type of casting is changed, it can be selected by learning a new one.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an automatic sensory testing device according to an embodiment of the present invention, and FIGS. 2A and 2B are diagrams showing band-pass filter output data of hitting sounds of non-defective products and defective products by the same device. 1. Sample, 2. Percussion device, 3. Microphone, 4. Bandpass filter, 5. Selector switch, 6. Study memory memo Ic. 7. Test data register, 8. Comparator, 9. Results display. Name of agent: Patent attorney Toshio Nakao, 1st person, 2nd person
figure

Claims (1)

[Claims] A microphone that picks up the sound of hitting the sample;
A band-pass filter that divides the collected hitting sound into frequencies, a storage means in which data of each band of this band-pass filter is stored in advance, and output data of the band-pass filter when a test sample is hit. a register to
An automatic sensory testing device comprising a comparator that compares the data stored in the register with the data stored in the storage means to determine the rank of good or bad of the sample to be tested.