CN1341908A - Automatic analysis and identification device of internal defect in cast and its analysis and identification method - Google Patents

Abstract

The invention relates to an automatic analysis and identification device for casting internal defects and an analysis and identification method thereof. The device is composed of a display screen, a casing and a panel, a connecting cable, a power switch, various function keys, and a circuit board. The corresponding points are connected; the circuit is composed of A/D conversion circuit, digital signal processor, synchronous random access memory, programmable read-only memory, address decoding and chip selection controller, serial interface circuit, liquid crystal display output circuit through the power supply Lines and respective signal lines are connected. The invention can analyze and identify the internal defects of castings in real time, is accurate, reliable and has high efficiency, and overcomes and solves the shortcomings and problems existing in the prior art.

Description

Automatic analysis and identification device for casting internal defects and its analysis and identification method

(1) Technical field:

The invention is an automatic analysis and identification device for internal defects of castings and an analysis and identification method thereof, belonging to digital image analysis and identification technology and equipment.

(two) background technology:

Metal casting is an industrial process of processing metal into the desired geometric shape. The main process is to pour the molten metal into the mold, and obtain the shape required by the design after cooling. In this production process, due to the influence of many factors such as process design, material formula, production equipment, environmental conditions, etc., castings inevitably form various casting defects. As the requirements for the quality of casting products are getting higher and higher, higher requirements are also put forward for the detection, analysis and identification methods of casting product quality. Due to the diversity of casting products, the detection of internal defects of castings generally uses X-ray machines or layer scanning (CT) instruments to obtain defect images inside castings, and then the inspectors follow the non-destructive inspection standards formulated by the International Standard Testing Committee ASTM. The reference image of the international standard for casting inspection is compared with the actual inspection image. The inspection process often requires repeated comparison, identification and judgment of defect images of up to 7 categories and 8 levels that appear in the same type of product. In this long-term and boring inspection and identification work environment, it is easy to cause eye fatigue, so It will inevitably directly affect the accuracy and efficiency of detection. That is to say, there are shortcomings and problems in the prior art, such as the detection and identification personnel are very strenuous, labor-intensive, easily fatigued, low in detection and identification efficiency, poor in accuracy, and large in randomness.

The purpose of the present invention is to overcome and solve the problem that there is no special device for automatic detection, analysis and recognition of internal defects of castings or existing detection, analysis and recognition technology, and the existence of detection and recognition personnel is very laborious, laborious, easy to fatigue, low in efficiency, large in randomness, and accurate In order to solve the shortcomings and problems of poor performance, research, design, and invent a device that can automatically analyze and identify the internal defect images of castings that are acquired online in real time, and its analysis and identification methods have high precision, good accuracy, high reliability, and high efficiency. , Low labor intensity and low cost.

(3) Contents of the invention:

The present invention is achieved through the following technical solutions: the schematic diagram of the external structure of the casting internal defect automatic analysis and identification device is shown in Figure 1; its circuit block diagram is shown in Figure 2; the circuit schematic diagram of one of its implementations is shown in Figure 3 shown. It consists of a liquid crystal display 1, a casing and a panel 2, a connecting cable 3, a power switch 4, function keys 5, 6, 7, and circuit boards. In the shell 2, the liquid crystal display 1 is installed on the panel 2 of the casing and is electrically connected to the corresponding point of the liquid crystal display output circuit on the circuit board through the display signal line; the power switch 4 and the function keys 5, 6, 7 are installed on the machine respectively. The lower part of the panel 2 of the shell is electrically connected to the power supply terminal on the circuit board and the corresponding points of each corresponding functional circuit through the power line and each functional signal line respectively; one end of the connecting cable 3 passes through the hole reserved on the side of the casing 2 and the circuit The corresponding points of the corresponding circuit on the board are connected, and the other end is connected with the corresponding communication signal output end of the X-ray machine or CT scanner; the circuit consists of an A/D conversion circuit, a digital signal processor (DSP), a synchronous random access memory (SRAM), programmable read-only memory (EPROM), address decoding and chip select controller, serial interface circuit, and liquid crystal display output circuit are jointly electrically connected, and the interconnection relationship is: A/D conversion circuit respectively through analog Signal lines and digital signal lines are electrically connected to X-ray machines or CT scanners and digital signal processors (DSP); The signal line and the serial interface signal line are electrically connected with the synchronous random access memory (SRAM), the programmable read-only memory (EPROM), the address decoding and chip selection controller, and the serial interface circuit respectively; the address decoding and the chip The selection controller is electrically connected to the synchronous random access memory (SRAM) and the programmable read-only memory respectively through decoding and control signal lines; the serial interface circuit is electrically connected to the liquid crystal display output circuit through the display signal line. As shown in Figure 3, wherein: the A/D conversion circuit is composed of integrated circuit IC1; the digital signal processor is composed of chip IC2; the serial interface can adopt the serial interface in the digital signal processor chip IC2, and the liquid crystal display output circuit It is composed of J1; the synchronous random access memory is composed of IC9, IC10 and IC11; the programmable read-only memory is composed of IC7; the address decoding and chip selection controller is composed of IC3.

The method for automatic detection, analysis and recognition with the casting internal defect automatic analysis and identification device of the present invention is to classify and feature extract the reference images of the casting detection international standard; the casting internal defect images collected on-site are automatically analyzed and identified by the casting internal defects After the device performs feature extraction, pattern matching, and defect analysis and recognition, the category and level of the detected defect are displayed on the monitoring screen; the method steps are as follows: (1) firstly, the reference image of the casting inspection international standard is analyzed, and after adopting such as Defect analysis algorithm shown in accompanying drawing 5 carries out processing and feature extraction to the reference image of casting defect standard, forms a cover and is used for the model of casting defect analysis identification and is stored in the memory in the analysis identification device of the present invention; Defect analysis algorithm comprises The inventor aimed at the types of defects such as elongated pinholes, circular pinholes and spongy shrinkage porosity in castings, which are difficult to analyze and identify with traditional existing methods, using wavelet theory to creatively map the images of these types of defects Carry out directional filtering in the airspace respectively, and then select a suitable moving window, according to the area occupied by the defect shrinkage cavity in different windows, the wavelet classification analysis algorithm is used to classify and distinguish different defects; (2) Then the internal defects of castings collected on site The analog signal output by the X-ray machine or CT layer scanner is input to the A/D converter of the device of the present invention for analog/digital conversion, and the automatic analysis and identification device for internal defects of the casting adopts the defect analysis algorithm shown in Figure 5 to carry out online The feature extraction of the image that obtains in real time; (3) carry out pattern matching then with the feature obtained and the feature of the standard reference image stored in the above-mentioned memory in the device, then carry out defect type recognition and defect grade recognition, finally in The category and the level of the detected defect are displayed on the monitoring screen; the more specific process is as follows: after the device of the present invention is powered on, the digital signal processor DSP chip is reset, and the self-guiding program that is solidified inside will be stored in the programmable read-only memory EPROM. The program and data are moved to the high-speed synchronous random access memory SRAM, and then the digital signal processor DSP starts the image analysis and recognition algorithm, and runs the defect analysis algorithm once for each frame, which takes about 22 milliseconds, and then the analysis is performed by the serial communication port. And recognition result is output to liquid crystal display screen; Because the operation speed of digital signal processor DSP is enough to finish casting image defect analysis and recognition algorithm (comprising preprocessing) in one frame, device of the present invention can work by full-duplex mode.

Compared with the prior art, the present invention has the following advantages and beneficial effects: (1) the present invention provides a real-time on-line casting internal defect image analysis and recognition device and analysis and recognition method for casting production; (2) the present invention can completely overcome and Solve the shortcomings and problems of inspectors' eye fatigue, poor accuracy, high randomness, low efficiency, unreliability, etc. in the existing casting quality inspection technology; (3) the present invention can improve the accuracy and accuracy of casting quality inspection, analysis, and identification Reliability, improve the efficiency of casting inspection, reduce the labor intensity of the inspection, analysis and identification process, and provide advanced technical means for computer integrated manufacturing, so that the production quality data obtained online can be automatically processed, stored and queried; (4) this The device and method of the invention can be used to analyze and identify any other casting internal defect images and graphics obtained by electronic methods; (5) the device of the invention has low cost, fast image processing speed, strong confidentiality and high precision.

(4) Description of drawings:

Fig. 1 is a schematic diagram of the external structure of the automatic analysis and identification device for internal defects of castings; Fig. 2 is a block diagram of its circuit; Fig. 3 is a schematic circuit diagram of one of its implementations; Fig. 4 is a block diagram of the program flow for online analysis and identification of internal defects of castings; Fig. 5 is a flow chart of the defect analysis algorithm of the automatic analysis and identification device for internal defects of castings. In the figure: 1 is the liquid crystal display screen, 2 is the casing and panel, 3 is the connecting cable, 4 is the power switch, 5, 6, and 7 are function keys, which can respectively realize the functions of pause, track and image selection; Fig. 3 Points with the same designation shall be electrically connected.

(5) Implementation method:

The casting internal defect automatic analysis and identification device and method thereof of the present invention can be realized by means of software, hardware or a combination of software and hardware, such as writing device processing software with Fortran and C language on a general-purpose computer to realize the method of the present invention, or using a general-purpose Programmable devices such as single-chip microcomputers can be used to replace part of the software functions with hardware, and can also be implemented by using a general-purpose or special-purpose digital signal processor DSP and adding a special-purpose accelerated processor on a general-purpose computer. In the preferred embodiment described now, the method and device for automatic analysis and identification of internal defects of the casting are realized by combining software and hardware, and its software flow block diagrams are as shown in Figure 4 and Figure 5; (1) according to Figure 1 As shown, the design, processing, and manufacture of the casing and panel 2 can be made of aluminum alloy materials and processed by general machining methods. After processing, arrange and install the LCD screen 1 and power switch 4 purchased in the market as shown in Figure 1 and each function key 5, 6, 7; (2) as shown in Fig. 2 and Fig. 3, use a computer to draw the circuit board, and screen the components to install the circuit board, for example: the TLC5510 of TI company can be selected for the A/D conversion circuit Chip; digital signal processor DSP can choose TMS320C31 floating-point DSP chip produced by Texas Instruments; synchronous random access memory SRAM can choose SL8X350 memory chip; programmable read-only memory EPROM can choose TMS27C512 chip; serial interface circuit can use The serial interface in the digital signal processor chip; the address decoding and chip selection controller can choose the SN74HC138 integrated part; (3) After installing the circuit board and simply powering on and debugging, write the software program as shown in Figure 4 and Figure 5 And store it in SRAM and EPROM; (4) connect the liquid crystal display screen 1, power switch 4 and each function key 5, 6, 7 installed on the casing panel respectively through the display signal line, the power line and each function signal line. Corresponding points of respective corresponding circuits on the circuit board in the casing 2 are electrically connected; through the signal cable 3, the corresponding analog signal input circuit on the circuit board is electrically connected with the corresponding points of the X-ray machine or CT machine, so that it can be better implemented this invention.

In the implementation process, more specific implementation and the performance of the device of the present invention are as follows: 1. the best embodiment of the device of the present invention is to adopt the DSP chip as the core, and is equipped with external circuits such as image acquisition, storage and liquid crystal display. The main features of the image processor are low cost and fast image processing speed, which can fully meet the real-time on-line detection analysis and identification requirements of the foundry production line; in addition, through the hardware of the software, the software has strong confidentiality and reliable operation The core part of this device adopts the floating-point DSP chip TMS320C31 produced by Texas Instruments, whose instruction cycle is 33ns and operating frequency is 40MHz, which can be used to realize various DSP algorithms; in addition, the chip also provides rich hardware resources: The data bus and address bus are 32-bit and 24-bit respectively. The separate program bus, address bus and direct memory access (DMA) enable addressing, reading and writing data and DMA operations to be performed in parallel, and can also access up to 16M of 32 Bit-byte memory space; 64-byte high-speed cache and 2K-byte fast RAM greatly reduce the number of off-chip accesses and improve the running speed of the program; the way to load the boot program is to load the program and data from the low-speed EPROM or The serial port is loaded into fast RAM; the on-chip serial port enables TMS320C31 to directly exchange data with serial external devices, supporting 8, 16, 24 and 32-bit data exchange; two timers can be used to realize various functions; ②A The /D conversion circuit adopts TLC5510 of TI Company as the A/D conversion chip, its conversion precision is 8 bits, CMOS level, sampling frequency is 8MHz, parallel output mode. It uses a half-flash fast structure, works with a single 5-volt power supply, and consumes only 100mW of power. It also includes an internal sampling and holding circuit, a high-impedance parallel port and an internal reference resistor; ③The analog signal of the X-ray machine or CT machine is input from the analog input port ANALOG IN of the A/D converter TLC5510 (IC1), as shown in Figure 3 AIN is shown in the middle, and capacitors C1 and R2 act as decoupling grounding; the input signal is converted into CMOS level by level shifter 74HC25 (IC3), and enters buffer 74603 (IC4). Since the output of IC1 is TTL level, so In the embodiment, IC3 is used for level conversion; output D0-D7 of IC1 is connected to A0-A7 of IC3, output Y0-Y7 of IC3 is CMOS level, connected to A0-A7 of buffer IC4, and output B0 of IC4 -B7 is connected with the data lines D0-D7 of the DSP chip TMS320C31 (IC2), and a selector 74157 (IC5), which is equivalent to a data acquisition switch, controls the output enable terminal OE of the register IC4 through the pin 1Y, when When OE is low level, data output is allowed; when OE is high level, D0-D7 of IC1 is in a high-impedance state. When collecting data, the DSP internal program sets the input pin XF0 to "0", makes the OE end of IC1 connected to IC2's XF0 trigger on the falling edge and thus starts the analog-to-digital conversion, and at the same time makes the output pin XF1 Set "1", through the selection pin SLE connected to it, the selector 74157 selects the 1A terminal connected to the serial communication terminal STRB of the DSP, the buffer starts to work, and the digital data is read from the buffer into the data line of the DSP ;When the data of one frame is collected, set XF1 to "0", the selector output Y is B (because B is high level), the buffer is closed, and stop sending data; through the write/read port W/R of IC2 Connection, where the clock signal CLK of the A/D converter is obtained by DSP frequency division. Resistors R1 and R4 plus adjustable resistance R3 divide the voltage to provide reference voltage to the reference voltage terminal REF of IC1; ④The real-time running program and data of IC2 are stored in the synchronous random access memory SRAM, which is four pieces of 256K×8 bits The storage chip SL8×350 (IC8, IC9, IC10 and IC11) constitutes a 256K×32-bit storage space, among which the address lines A0-A14 are connected to the address lines of IC2, and the data lines D0-D7 are respectively connected to the IC2’s D0-D7, D8-D15, D16-D23, D24-D31 are connected to the data bus; in addition, IC2 is also connected to a programmable read-only memory EPROM of TMS27C512 (IC7) for storing programs and initializing data, forming a 32K× 8 storage space; 16 address lines A0-A15 of EPROM are directly connected to A0-A15 of IC2, 8 data lines are connected to D0-D7 of IC2, chip select terminal CE and its own output enable terminal OE After being connected, it is connected to the output terminals Y1 and Y2 of the decoder SN74HC138 (IC6); the pins A, B and C of IC6 are respectively connected to the address lines A18, A22 and A23 of IC2; in addition, IC2 also outputs to a 64 The ×12 liquid crystal dot matrix display (J1) is used as a status display, and the product of the Japanese seiko Epson company model sed1330 can be selected; A13 of IC2 is used as the output of the liquid crystal dot matrix to enable connection to its CE terminal. Its D0-D19 time-sharing is used as the line signal of liquid crystal dot matrix, and D20-D31 is used as the line scan signal of liquid crystal dot matrix; 5. Fig. 4 is the flow chart of the method and device processing method that embodies casting internal defect automatic analysis identification of the present invention. Since the inspection, analysis and identification of casting quality are mainly based on the international standards for non-destructive testing provided by the ASTM committee, the automatic analysis and identification of casting internal defects still follow this standard, but it cannot directly use the standard diagrams provided by the committee like relying on human eye analysis. However, it must be processed so that the automatic analysis device can complete pattern matching and recognition through relatively simple digital operations; The casting defect image collected online is also subjected to feature analysis and feature extraction, and then the obtained features are matched with the features corresponding to the standard defect image, and the type and level of the casting defect are analyzed according to the algorithm introduced below , finally the analysis results are displayed on the display screen of the casting internal defect automatic analysis and identification device of the present invention; The mode of operation in the chip is to improve the reliability and the speed of real-time processing. The analysis and recognition algorithm of the casting defect image includes some methods commonly used in the field of image processing technology, such as: image segmentation involved in image preprocessing , edge detection, image enhancement and image binary processing, and the calculation of image geometric features involved in the analysis and recognition algorithm, such as inertia calculation, area calculation, etc. Research such as the wavelet classification analysis algorithm for gray texture images. The algorithm proposed by the inventor takes into account seven types of casting defects. A, B, C, D, E, and F as shown in Figure 3 represent air holes, elongated pinholes, round pinholes, spongy shrinkage, low Seven types of defects such as density foreign matter and shrinkage cavity. The algorithm shown in Figure 3 recognizes F and G as a class of defects during the automatic analysis and identification process, and then uses the same non-destructive detection method but a physical method to detect F and G Defects, namely low-density foreign matter and shrinkage cavity defects are identified.

Claims (2)

1, a kind of internal defect in cast is analyzed recognition device automatically, it is characterized in that: it is connected and composed by LCDs (1), casing and panel (2), stube cable (3), power switch (4), function key (5), (6), (7), the common installation of circuit board, its mutual alignment and annexation are: circuit board is fixedly installed in the casing (2), and the panel (2) that LCDs (1) is loaded on casing is gone up and is electrically connected mutually by liquid crystal display output circuit respective point on display signal line and the circuit board; Power switch (4) and function key (5), (6), (7) be installed on respectively panel (2) bottom of casing and respectively by power lead, each function signal line respectively with circuit board on power end, each corresponding function circuit respective point be electrically connected mutually; One end of stube cable (3) passes the hole of reserving casing (2) side and is connected with related circuit respective point on the circuit board, and its other end is connected with X-ray machine or CT plane scan instrument respective communication signal output part; Its circuit selects controller, serial interface circuit, liquid crystal display output circuit to be electrically connected jointly by A/D change-over circuit, digital signal processor, synchronous RAM, programmable read only memory, address decoding and sheet to constitute, and its interconnected relationship is: the A/D change-over circuit is electrically connected with X-ray machine or CT plane scan instrument, digital signal processor respectively mutually by analog signal line, digital signal line respectively; Digital signal processor selects controller, serial interface circuit to be electrically connected mutually with synchronous RAM, programmable read only memory, address decoding and sheet respectively by synchronous access signal wire, storage signal line able to programme, decoding and control signal wire, serial line interface signal wire respectively; Address decoding and sheet select controller to be electrically connected mutually with synchronous RAM, programmable read only memory respectively by decoding and control signal wire respectively; Serial interface circuit is electrically connected with the liquid crystal display output circuit mutually by display signal line.

2, a kind ofly analyze the method that recognition device carries out internal defect in cast analysis identification automatically with the described internal defect in cast of claim 1, it is that the reference picture that casting detects international standard is classified and feature extraction; The internal defect in cast image of collection in worksite shows the classification and the grade of the defective of surveying on the monitoring screen after internal defect in cast is analyzed recognition device automatically and carried out feature extraction, pattern match and defect analysis identification; Its method step is as follows: (1) at first will cast the reference picture that detects international standard and analyze, through adopting the defect analysis algorithm that the reference picture of casting defect standard is handled and feature extraction, form the model that a cover is used for analyzing identification and deposit in the storer of analysis recognition device of the present invention; The defect analysis algorithm comprises that the inventor is at this several defect type images travel direction filtering in the spatial domain respectively of the microscler pin hole in the foundry goods, circular pin hole and spongy shrinkage porosite, select suitable moving window again, come the small echo classification analysis algorithm that different defectives are classified and distinguished at the shared area of defective shrinkage cavity in the different windows; (2) A/D converter that then simulating signal of the X-ray machine of the internal defect in cast of collection in worksite or the output of CT plane scan instrument is input to apparatus of the present invention carries out mould/number conversion, analyzes recognition device automatically through internal defect in cast and adopts the defect analysis algorithm to carry out the feature extraction of the image that online in real time obtains; (3) then resulting feature and the feature that is stored in the canonical reference image in the above-mentioned storer in the device are carried out pattern match, then carry out the grade identification of defect type identification and defective, on the monitoring screen, show the classification and the grade of the defective of surveying at last; More specifically process is as follows: after apparatus of the present invention power up, the digital signal processor DSP chip reset, the program and the data that will be stored among the programmable read only memory EPROM by the self initializing program of inside solidification move to high-speed synchronous random access memory SRAM, digital signal processor DSP begins image analysis and recognizer then, will be analyzed with recognition result by serial communication port and output to LCDs; The arithmetic speed of digital signal processor DSP is enough to finish foundry goods image defect analysis and recognizer in a frame, device is pressed full duplex mode work.

Images