Online detection device and method for cracks of holding-up hammer of six-surface press based on multilayer PCB (printed circuit board) coils and multiple TMR (total magnetic resonance) sensors

Abstract

The invention provides a six-surface press holding-up hammer crack online detection device and method based on multilayer PCB coils and multiple TMR sensors, and the detection device comprises a power supply, a signal generation circuit, a power amplification circuit, a TMR sensor detection probe, a signal amplification processing module, a main control chip, and an alarm module. The power supply is connected with the signal generation circuit, the power amplification circuit and the TMR sensor detection probe and provides stable power supply for the signal generation circuit, the power amplification circuit and the TMR sensor detection probe. The signal generation circuit is connected with the power amplification circuit, the power amplification circuit is connected with the TMR sensor detection probe, the TMR sensor detection probe is connected with the signal amplification processing module, the signal amplification processing module is connected with the main control chip, and the main control chip is connected with the alarm module. According to the method, lift-off and edge effects are solved at the same time, and surface and internal cracks of the holding-up hammer of the six-surface press are detected and judged.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive detection of anvil cracks, and particularly relates to an on-line detection device and method for anvil cracks of a six-sided press based on a multi-layer PCB coil and multiple TMR (Tunnel Magneto Resistance) sensors. Background Art

[0002] In the domestic and foreign industrial production of synthetic diamond, the static pressure catalyst method is usually adopted to synthesize diamond. In this process, graphite is used as the raw material, and under the action of high temperature, high pressure and a catalyst metal, the crystal structure of graphite is changed, and then it is transformed into diamond. As a key component of the six-sided press for artificial synthesis of diamond, the tungsten carbide anvil will bear various stresses during the synthesis of diamond, including the alternating action of tensile, compressive, shear and thermal stresses, and it is in a subcritical stress state during operation. Tungsten carbide itself is a brittle metal, so external factors such as fatigue damage, process defects, improper anvil alignment and incorrect placement of the synthesis block may cause the anvil to crack. If the operator fails to detect the crack and breakage of the anvil in time and take pressure relief and stop heating operations, the pressure in the high-pressure cavity may leak, resulting in the flying of fragments and synthesis blocks, and even causing accidents. At the same time, multiple anvils may also be damaged, resulting in greater economic losses and potential hazards to personnel. Therefore, in order to ensure the reliability of artificial synthesis of diamond, it is necessary to regularly detect and maintain the six anvils of the press.

[0003] At present, manual inspection of anvils still dominates in industrial production. Manual inspection mainly involves first performing temperature and pressure reduction treatment on the synthesis cavity of the six-sided press, and then the inspector gently scratches each surface of the six anvils with a saw blade, and judges whether the anvil is damaged by the feel and sound when scratching the anvil. This detection method has disadvantages such as strong subjectivity and being greatly affected by the production environment. Although the manual detection method can reduce the occurrence of cylinder explosion accidents to a certain extent, it is still very difficult to thoroughly detect the cracks of the press anvil, especially the problem of detecting internal cracks of the anvil.

[0004] In order to solve the inaccurate problem of manual detection, the current research work on the detection of anvil crack breakage by scholars mainly includes: 1. Detection is carried out through image detection and recognition methods. However, the method of image recognition is greatly affected by the complex working conditions on the surface of the anvil. Many cracks are difficult to accurately detect. At the same time, the method of image recognition cannot detect the internal cracks and breakages on the surface of the anvil.

[0005] 2. Detection is carried out through acoustic methods. However, the noise is very loud when the press is working, and it is also very difficult to accurately detect various cracks by acoustic methods.

[0006] The utility model patent with the application number 202121837055.X discloses a six-sided press anvil crack detection system based on eddy current, including: a robotic arm with a fixture at its end; an eddy current detection probe installed in the fixture. The robotic arm is used to control the eddy current detection probe to contact the surface of the anvil, and the anvil is installed in a six-sided press. The eddy current detector is connected to the eddy current detection probe through a signal transmission line. The above utility model controls the eddy current detection probe through a robotic arm to detect the anvil in the six-sided press, with high detection sensitivity, improving the detection quality and efficiency, and can be widely applied to the crack detection technology field. However, due to the randomness of the damage crack direction of the anvil, the eddy current effect can be effectively detected when the crack is perpendicular to the movement direction of the probe, and there is no eddy current effect when the crack is horizontal to the movement direction of the probe. Only a single eddy current detection probe cannot accurately detect cracks in different random directions, resulting in the problem of missed crack detection. Summary of the Invention

[0007] Aiming at the technical problems in the prior art that the crack detection of the anvil of a six-sided press requires automatic detection of multiple presses, the efficiency of manual detection is low, and internal cracks cannot be detected, the present invention proposes an on-line crack detection device and method for the anvil of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors, which can detect the health status of the anvil of the six-sided press in real time during the process of artificial synthesis of diamond, so as to solve the problem that the current manual detection of the anvil status is subjective and missed detection leads to the risk of explosion of the synthesis cavity during the synthesis process.

[0008] To achieve the above object, the technical solution of the present invention is realized as follows: An on-line crack detection device for the anvil of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors, characterized by including a power supply, a signal generation circuit, a power amplification circuit, a TMR sensor detection probe, a signal amplification and processing module, a main control chip, and an alarm module. The power supply is respectively connected to the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe to provide stable power supply to the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe. The signal generation circuit is connected to the power amplification circuit, the power amplification circuit is connected to the TMR sensor detection probe, the TMR sensor detection probe is connected to the signal amplification and processing module, the signal amplification and processing module is connected to the main control chip, and the main control chip is connected to the alarm module.

[0009] Preferably, the signal amplification and processing module is connected to the main control chip through an AC-to-DC effective value conversion chip, and the main control chip is connected to the upper computer.

[0010] Preferably, the main control chip is connected to the upper computer. The alarm module includes an indicator light and a buzzer connected in parallel. When alarming, the buzzer alarms and the indicator light flashes.

[0011] Preferably, the TMR sensor detection probe includes a probe circuit board. A multi-layer PCB coil is provided on the probe circuit board. The multi-layer PCB coil is connected to a power amplification circuit. A plurality of TMR sensors are provided on the multi-layer PCB coil. The TMR sensors are used to detect tunnel magnetoresistance effect signals. The TMR sensors are connected to an effective value conversion chip through a signal amplification and processing module, and the AC signal is converted into a DC signal and then enters the main control chip for digital signal processing.

[0012] Preferably, there are 3 TMR sensors, and the 3 TMR sensors are arranged in a 120-degree array on the multi-layer PCB coil.

[0013] Preferably, the signal generation circuit uses a signal generation chip to generate a sine excitation signal, and the signal frequency of the signal generation circuit is 10 kHz; the power amplification circuit includes a pre-amplifier and an amplifier chip. The pre-amplifier is connected to the signal generation circuit, the pre-amplifier is connected to the amplifier chip, and the amplifier chip is connected to the multi-layer PCB coil of the TMR sensor detection probe. The excitation signal after power amplification is passed into the multi-layer PCB coil to generate an excitation magnetic field.

[0014] Preferably, the power supply includes a transformer. One end of the transformer is connected to the mains power supply, and the other end of the transformer is connected to a rectifier bridge; the signal amplification and processing module is a differential instrumentation amplifier chip, which realizes the amplification of the output signal of the TMR sensor; the main control chip is provided with an analog-to-digital conversion module for collecting and processing the amplified signal of the TMR sensor and the signal after AC-to-DC conversion; the main control chip uses the BP neural network algorithm to fuse the data of the 3-way TMR sensors to realize the detection of the top hammer crack damage.

[0015] Preferably, the multi-layer PCB coil is a circular coil. The circular coil is a 6-8 layer PCB coil. Three TMR sensors are arranged in a 120-degree array at the center of the multi-layer PCB coil. The TMR sensors are connected to the signal amplification and processing module. An input interface for the coil excitation signal connected to the multi-layer PCB coil is provided on the probe circuit board, and the input interface for the coil excitation signal is connected to the power amplification circuit; a power signal input interface is provided on the probe circuit board, and the power signal input interface is connected to the output end of the power supply; a signal amplifier chip output signal interface connected to the signal amplification and processing module is provided on the probe circuit board, and the signal amplifier chip output signal interface is connected to the effective value conversion chip.

[0016] Preferably, the detection method is as follows: The signal generation circuit generates a sinusoidal excitation signal with a signal frequency of 10 kHz. The power amplifier circuit amplifies the power of the sinusoidal excitation signal. The sinusoidal excitation signal after power amplification is connected to the multi-layer PCB coil of the TMR sensor detection probe to generate an excitation magnetic field. When the TMR sensor detection probe moves on the surface of the anvil, the TMR sensor detects the tunneling magnetoresistance effect signal generated on the surface of the anvil. The signal collected by the TMR sensor is amplified by the signal amplification processing module, and then after the conversion from AC to RMS value, it becomes a digital signal through the analog-to-digital conversion of the main control chip. The three converted digital signals are transmitted in real time to the upper computer for serial port waveform display with QT as the development platform through serial port communication. The anvil crack and breakage recognition model judges and processes the digital signals during detection. When cracks are detected on the surface and inside of the anvil, an audible and visual alarm is given through the alarm module to remind to replace the anvil.

[0017] Preferably, the BP neural network algorithm is trained with the sensor output data without cracks, the sensor output data with cracks, the probe lift-off sensor output data, and the anvil edge effect sensor output data collected by the TMR sensor to obtain the anvil crack and breakage recognition model. The anvil crack detection model includes an input layer, a hidden layer, and an output layer. The outputs of the three TMR sensors are respectively used as three input signals, and the output of the output layer is crack or no crack. The hyperbolic tangent function "tansig" and the linear function purelin are respectively selected as the activation functions of the hidden layer and the output layer of the BP neural network, and the BP neural network is trained using the newff function and the gradient-based backpropagation algorithm.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: A detection probe based on a multi-layer PCB coil, multiple TMR sensors, and related circuits and detection algorithms is designed to solve the problem of the inability to detect the cracks in the anvil of a six-sided press online. The multi-layer PCB coil is used as the excitation coil to replace the traditional wound coil, greatly reducing the thickness and volume of the coil, facilitating the insertion into the internal gap of the anvil of the six-sided press through a manipulator or manual driving during detection, and moving on the surface of each anvil for detection. Three TMR sensors are used to obtain the tunnel magnetoresistance effect signals during detection to replace a single eddy current detection probe, solving the problem of the influence of cracks in different directions on the magnetic field, and facilitating subsequent data processing and the judgment of cracks and edge effects. The output signals of the three TMR magnetic sensors are amplified by a differential instrumentation amplifier and then converted into DC signals through an AC-to-RMS conversion chip for acquisition and processing by the main control chip. When the probe passes through the surface of the anvil of the six-sided press by a manipulator or manual driving, the main control chip collects the amplified and transformed output signals of the three TMR sensors, processes the collected data in real time using the BP neural network algorithm, and simultaneously solves the lift-off and edge effect problems, realizing the detection and judgment of the surface and internal cracks of the anvil of the six-sided press. When cracks are detected on the surface and inside the anvil, an audible and visual alarm is given through the alarm module to remind the replacement of the anvil. Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0020] Figure 1 It is a schematic diagram of the circuit board of the detection probe of the present invention.

[0021] In the figure, 1 is the probe circuit board, 2 is the multi-layer PCB coil, and 3 is the TMR sensor.

[0022] Figure 2 It is a principle block diagram of the present invention.

[0023] Figure 3 It is a working flow chart of the detection device of the present invention.

[0024] Figure 4 It is the algorithm structure diagram of the present invention Detailed Embodiments

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0026] Embodiment 1 As Figure 2 shown, the present invention provides an on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors, including a power supply, a signal generation circuit, a power amplification circuit, a TMR sensor detection probe, a signal amplification and processing module, a main control chip, and an alarm module. The power supply is respectively connected to the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe to provide stable power supply for the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe. The signal generation circuit is connected to the power amplification circuit, the power amplification circuit is connected to the TMR sensor detection probe, the TMR sensor detection probe is connected to the signal amplification and processing module, the signal amplification and processing module is connected to the main control chip through an effective value conversion chip, and the main control chip is connected to the alarm module. The signal generation circuit is responsible for generating a sine excitation signal, which will be transmitted to the multi-layer PCB coil of the TMR sensor detection probe to excite a magnetic field. To further improve the system performance, the power amplification circuit is used to amplify the power of the sine excitation signal to ensure that the system has sufficient sensitivity and stability. The amplified excitation signal is connected to the multi-layer PCB coil of the TMR sensor detection probe. The alarm module includes an indicator light and a buzzer connected in parallel. When alarming, the buzzer alarms and the indicator light flashes. The main control chip is connected to the upper computer for displaying the processed signal, and the signal can be manually monitored.

[0027] The power supply adopts a 50W dual 12V EI-type transformer. One end of the transformer is connected to the AC power supply of the city power, and the other end of the transformer is connected to a rectifier bridge. Since the power supply power required by the subsequent power amplification circuit is relatively large, a KBL608 rectifier bridge is used to provide sufficient power. The rectifier bridge can output relative positive and negative voltages to provide power for the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe.

[0028] The signal generation circuit uses an ICL8038 signal generation chip to generate a sine excitation signal, and the signal frequency of the signal generation circuit is 30 kHz. The power amplification circuit uses an NE5532 as the pre-amplifier, and a subsequent LM1875T amplifier chip is used for power amplification. The amplified excitation signal is input into the TMR sensor detection probe to generate an excitation magnetic field.

[0029] AsFigure 1 As shown in the figure, the TMR sensor detection probe is the core of the entire on-line detection device. The TMR sensor detection probe is a non-destructive probe composed of a multi-layer PCB coil 2 and multiple TMR sensors, including a probe circuit board 1. A multi-layer PCB coil 2 is provided on the probe circuit board 1. The multi-layer PCB coil 2 constitutes an excitation coil. The multi-layer PCB coil 2 is connected to a power amplifier circuit. The excitation signal generated by the power amplifier circuit is connected to the multi-layer PCB coil 2 to generate an alternating magnetic field. Multiple TMR sensors 3 are provided on the multi-layer PCB coil 2. There are 3 TMR sensors 3, and the 3 TMR sensors 3 are evenly distributed on the multi-layer PCB coil 2 at 120 degrees. The TMR sensor 3 is used to detect the tunneling magnetoresistance effect signal. When there is a crack, the signals of the three TMR sensors 3 change. The TMR sensor 3 is connected to an effective value conversion chip through a signal amplification processing module, converts the AC signal into a DC signal, and then enters the main control chip to be processed into a digital signal. The TMR2001 sensor, as a sensor element for detecting the tunneling magnetoresistance effect, is affected by various factors during the detection process of the detection probe, such as the movement direction of the probe, the orientation and length difference of the crack, etc. It is difficult to accurately determine the damage condition relying only on a single output signal. The present invention proposes to use the induction chips of multiple TMR sensors 3 to solve the problem that it is difficult for a single TMR induction chip to accurately identify due to various factors. Considering the needs of cost and detection accuracy, the detection probe is designed with three TMR sensors welded in a 120-degree array on the multi-layer PCB coil. The signal amplification processing module is a differential instrumentation amplifier chip, which realizes the amplification of the output signal of the TMR sensor.

[0030] The multi-layer PCB coil 2 is a circular coil. Figure 1 The circular coil in the upper part is a 6-8 layer PCB coil. The multi-layer PCB coil 2 is connected to a power amplifier circuit and serves as a detection excitation coil. The central part of the PCB coil is three TMR sensors 3, namely S1, S2, and S3, which are distributed at 120 degrees and are used to detect the magnitude of the tunneling magnetoresistance effect signal. When the TMR sensor detection probe moves on the surface of the anvil, when there is no crack on the surface or inside, the output signals of the three TMR sensors 3 change very little. When there is a crack on the surface or inside or the detection probe moves to the edge of the anvil, the outputs of the three TMR sensors 3 will change to varying degrees. U1, U2, and U3 are the signal amplifier chips of the three TMR sensors 3, which amplify the signals collected by the TMR sensors. P1 is the input interface of the coil excitation signal, and the input interface of the coil excitation signal is connected to the power amplifier circuit. P2 is the power signal input interface, which is connected to the output end of the power supply. P3 is the output signal interface of the signals of the three TMR sensors 3 after passing through the signal amplifier chip, and P3 is connected to the AC-to-DC effective value conversion chip.

[0031] The main control chip is the 32-bit STM32F407ZGT6 chip of STMicroelectronics. An analog-to-digital conversion module is provided inside the main control chip to collect and process the amplified and AC-to-DC converted signals output by the sensor. The main control chip uses the BP neural network algorithm to fuse and process the data of the three TMR sensors 3 to achieve the detection of the anvil crack damage. For the integrity and portability of the system, a complete circuit system is designed, including: a power supply, a signal generation circuit with adjustable frequency, a power amplification circuit, a sensor detection probe, and a main control chip, enabling the system to detect the changes in the output signals of the three TMR sensors 3 caused by various-direction damage cracks on the surface and inside of the anvil, and giving an alarm reminder through the alarm module.

[0032] As Figure 3 shown, when detecting the press anvil, driven by the manipulator or manually, the TMR sensor detection probe closely adheres to the surface of the anvil and moves uniformly through each surface of the anvil. When the TMR sensor detection probe moves on the surface of the anvil, the main control chip real-time collects the amplified and changed DC output signals of the three TMR sensors 3. Each of the three channels collects 50 - 100 data each time and uses the BP neural network algorithm to fuse and process the data. The judgment algorithm is one of the cores of the system. By collecting the output data of the TMR sensor 3 without cracks, the output data of the TMR sensor 3 with cracks, the output data when the probe is lifted from the sensor, and the output data of the anvil edge effect sensor, and training through the BP neural network algorithm, an anvil crack damage recognition model is obtained. The anvil crack damage recognition model judges and processes the data collected during detection, so as to effectively detect the crack damage according to the real-time collected data, prevent false alarms caused by lifting and moving to the edge, and effectively solve the on-line detection of cracks on the surface and inside of the anvil. When cracks are detected on the surface and inside of the anvil, an audible and visual alarm is given through the alarm module to remind to replace the anvil.

[0033] As Figure 4 shown is the structure diagram of the BP neural network algorithm, including an input layer, a hidden layer, and an output layer. The outputs of the three TMR sensors serve as the three input signals of the BP neural network, and the output y indicates the presence and absence of cracks. The anvil crack detection model is closer to a complex non-linear function. The hyperbolic tangent function "tansig" and the linear function purelin are respectively selected as the activation functions of the hidden layer and the output layer of the BP neural network. The newff function and the gradient-based backpropagation algorithm (train function) are used to train the BP neural network. Through the training of the TMR sensor data, finally, it is determined that the hidden layer network is 2 layers, and the neurons in the hidden layer are 7 and 8 respectively, achieving an identification effect of more than 99.5%.

[0034] Embodiment 2 An online crack detection method for the anvil of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors is as follows: A signal generation circuit generates a sinusoidal excitation signal, and the signal frequency of the signal generation circuit is 30 kHz. The power amplification circuit uses NE5532 as the preamplifier, and the amplifier chip LM1875T is used for power amplification subsequently. The amplified excitation signal is connected to the multi-layer PCB coil 2 of the TMR sensor detection probe to generate an excitation magnetic field. When the TMR sensor detection probe moves on the surface of the anvil, 3 TMR sensors 3 detect the tunnel magnetoresistance effect signals generated on the surface of the anvil. The signals collected by the 3 TMR sensors are amplified by the signal amplification processing module, and then after the alternating current to effective value conversion, they become digital signals through the analog-to-digital conversion inside the CPU of the main control chip. The digital signals converted by the 3-channel TMR sensors are transmitted in real time to the serial port waveform display upper computer with QT as the development platform through serial port communication. Through the output data of the sensor without cracks, the output data of the sensor with cracks, the output data of the probe lift-off sensor, and the output data of the anvil edge effect sensor, the BP neural network algorithm is trained to obtain the anvil crack and damage recognition model. Then, the model is implemented through the CPU software. When cracks are detected on the surface and inside of the anvil during detection, an audible and visual alarm is given through the alarm module to remind to replace the anvil.

[0035] Other structures and principles are the same as those of Embodiment 1.

[0036] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An on-line crack detection device for the anvil of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors, characterized in that, It includes a power supply, a signal generation circuit, a power amplification circuit, a TMR sensor detection probe, a signal amplification and processing module, a main control chip, and an alarm module. The power supply is respectively connected to the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe, and provides stable power supply to the signal generation circuit, the power amplification circuit, and the TMR sensor detection probe; the signal generation circuit is connected to the power amplification circuit, the power amplification circuit is connected to the TMR sensor detection probe, the TMR sensor detection probe is connected to the signal amplification and processing module, the signal amplification and processing module is connected to the main control chip, and the main control chip is connected to the alarm module.

2. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 1, wherein, The signal amplification and processing module is connected to the main control chip through an AC-to-DC effective value conversion chip, and the main control chip is connected to the upper computer.

3. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 2, characterized in that, The main control chip is connected to the upper computer. The alarm module includes an indicator light and a buzzer connected in parallel. When alarming, the buzzer gives an alarm and the indicator light flashes.

4. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to any one of claims 1-3, characterized in that, The TMR sensor detection probe includes a probe circuit board (1). A multi-layer PCB coil (2) is provided on the probe circuit board (1). The multi-layer PCB coil (2) is connected to the power amplification circuit. A plurality of TMR sensors (3) are provided on the multi-layer PCB coil (2). The TMR sensors (3) are used to detect tunnel magnetoresistance effect signals. The TMR sensors (3) are connected to the effective value conversion chip through the signal amplification and processing module, and the AC signal is converted into a DC signal and enters the main control chip for digital signal processing.

5. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 4, characterized in that, There are 3 TMR sensors (3), and the 3 TMR sensors (3) are arranged in a 120-degree array on the multi-layer PCB coil (2).

6. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 5, characterized in that, The signal generation circuit uses a signal generation chip to generate a sine excitation signal, and the signal frequency of the signal generation circuit is 10 kHz; the power amplification circuit includes a pre-amplifier and an amplifier chip. The pre-amplifier is connected to the signal generation circuit, the pre-amplifier is connected to the amplifier chip, and the amplifier chip is connected to the multi-layer PCB coil (2) of the TMR sensor detection probe. The amplified excitation signal is input into the multi-layer PCB coil (2) to generate an excitation magnetic field.

7. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 5 or 6, characterized in that, The power supply includes a transformer. One end of the transformer is connected to the mains, and the other end of the transformer is connected to a rectifier bridge; the signal amplification and processing module is a differential instrumentation amplifier chip, which realizes the amplification of the output signal of the TMR sensor (3); an analog-to-digital conversion module is provided in the main control chip to collect and process the amplified signal of the TMR sensor (3) and the signal after AC-to-DC conversion; the main control chip uses the BP neural network algorithm to fuse the data of the 3 TMR sensors (3) to realize the detection of the top hammer crack damage.

8. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 7, characterized in that, The multi-layer PCB coil (2) is a circular coil, which is a 6-8 layer PCB coil. Three TMR sensors (3) are arranged in a 120-degree array at the central part of the multi-layer PCB coil (2). The TMR sensors (3) are connected to a signal amplification and processing module. An input interface for the coil excitation signal connected to the multi-layer PCB coil (2) is provided on the probe circuit board (1). The input interface for the coil excitation signal is connected to a power amplification circuit; a power signal input interface is provided on the probe circuit board (1), and the power signal input interface is connected to the output terminal of a power supply. A signal amplifier chip output signal interface connected to the signal amplification and processing module is provided on the probe circuit board (1), and the signal amplifier chip output signal interface is connected to an effective value conversion chip.

9. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to any one of claims 5, 6, and 8, characterized in that, The detection method is as follows: A signal generation circuit generates a sinusoidal excitation signal with a signal frequency of 10 kHz. The power amplification circuit amplifies the power of the sinusoidal excitation signal. The sinusoidal excitation signal after power amplification is connected to the multi-layer PCB coil (2) of the TMR sensor detection probe to generate an excitation magnetic field. When the TMR sensor detection probe moves on the surface of the anvil, the TMR sensor (3) detects the tunnel magnetoresistance effect signal generated on the surface of the anvil. The signal collected by the TMR sensor (3) is amplified by the signal amplification and processing module, and then after the AC-to-effective value conversion, it becomes a digital signal through the analog-to-digital conversion of the main control chip. The three converted digital signals are transmitted in real time to the upper computer for serial port waveform display with QT as the development platform through serial port communication; the anvil crack and breakage recognition model judges and processes the digital signals during detection. When cracks are detected on the surface and inside of the nail hammer, an audible and visual alarm is given through the alarm module to remind to replace the anvil.

10. The on-line crack detection device for the top hammer of a six-sided press based on a multi-layer PCB coil and multiple TMR sensors according to claim 9, characterized in that, The BP neural network algorithm is trained with the output data of the sensor without cracks, the output data of the sensor with cracks, the output data of the probe lift-off sensor, and the output data of the anvil edge effect sensor collected by the TMR sensor (3) to obtain the anvil crack and breakage recognition model; The anvil crack detection model includes an input layer, a hidden layer, and an output layer. The outputs of the three TMR sensors are respectively used as three input signals, and the output of the output layer is crack or no crack; the hyperbolic tangent function "tansig" and the linear function purelin are respectively selected as the activation functions of the hidden layer and the output layer of the BP neural network, and the BP neural network is trained using the newff function and the gradient-based backpropagation algorithm.

Images