Electromagnetic detection apparatus and method for monitoring slippage of externally prestressed tendons

Abstract

The application discloses an electromagnetic detection device and method for monitoring slippage of an external prestressed beam, and the device comprises an excitation coil, a coil type sensing element, a multi-channel acquisition instrument, and a signal control and demodulation instrument; the signal control and demodulation instrument drives the excitation coil to generate an electromagnetic field to magnetize the steel strand; the coil type sensing element is arranged correspondingly to the separate turns of the steel strand, and the magnetic characteristic signals of the steel strand in each turn layer are captured; the multi-channel acquisition instrument independently acquires the signals of each sensing element; the signal control and demodulation instrument adopts a full connection neural network trained by calibrated data to intelligently analyze the signals and accurately determine the specific turn layer and the number of slippage occurrence. The application realizes non-contact, nondestructive and real-time online monitoring of the anchoring state of the external prestressed beam, has the advantages of high precision, good adaptability and strong anti-interference capability, and effectively solves technical problems, such as the timely early warning, accurate positioning and quantitative evaluation of slippage accidents, which cannot be realized by traditional detection methods.

Description

Technical Field

[0001] This invention belongs to the field of bridge engineering structural health monitoring and safety assessment technology, specifically relating to an electromagnetic detection device and method for monitoring slippage of external prestressed tendons. Background Technology

[0002] External prestressing technology is one of the key technologies for large-scale concrete structures such as modern bridges and buildings. It applies loads to the structure through prestressed steel strands placed outside the concrete cross-section, offering advantages such as flexible strand placement, low friction loss, and ease of inspection and replacement. However, the anchoring system of the external prestressed strands, as the core component for transmitting and maintaining enormous prestress, is under constant high stress and directly exposed to the environment. This makes it highly susceptible to anchoring failure due to factors such as fatigue stress, stress corrosion, improper installation of wedges, or vibration, resulting in prestressed strand slippage. Such slippage accidents are sudden and can lead to a rapid loss of prestress, severely weakening the structure's load-bearing capacity and stiffness, and even causing catastrophic consequences such as structural collapse, posing a significant threat to public safety. Therefore, how to achieve accurate and rapid monitoring of the anchoring status of external prestressed strands has become a critical technical challenge that urgently needs to be solved in current engineering practice.

[0003] Currently, the industry lacks efficient, accurate, and easily implemented conventional methods for detecting and monitoring slippage or detachment of external prestressed strands. Traditionally, markings are made at the ends of the steel strands during tensioning, and slippage is later determined by manually measuring the change in distance between the marked points and the anchor end face. While this method is intuitive, it is inefficient and difficult to achieve large-scale, periodic, and rapid detection. Furthermore, due to the sudden nature of external prestressed strand detachment accidents, manual detection methods are insufficient for timely detection and effective handling.

[0004] Another approach is to indirectly infer whether slippage has occurred by monitoring changes in the tension of the prestressed strands. Conventional methods for this include ultrasonic guided wave methods (such as the technology in Chinese patent application CN117109797A) and magnetic flux sensor methods (such as the technology in Chinese patent application CN203490007U). However, these methods have significant technical limitations: since external prestressed strands are typically composed of multiple strands, monitoring the prestress of each strand individually would be too costly; and using whole-strand prestress monitoring cannot effectively distinguish whether the tension loss is caused by slippage of a single strand or overall prestress relaxation. Summary of the Invention

[0005] In view of the above, the present invention provides an electromagnetic detection device and method for monitoring slippage of external prestressed tendons, which can solve the problems of interference with the structure, low detection accuracy, poor real-time performance, and inability to quantitatively locate slippage in traditional detection methods.

[0006] An electromagnetic detection device for monitoring slippage of external prestressed tendons, comprising:

[0007] The excitation coil is arranged outside the external prestressed strand anchor head and is used to generate a magnetic field to magnetize the steel strand under current excitation.

[0008] Multiple coil-type sensing elements are installed on the outside of the external prestressed tendon anchor head to generate corresponding magnetic characteristic electrical signals through electromagnetic induction;

[0009] A multi-channel data acquisition instrument is used to independently acquire the magnetic characteristic electrical signals generated by each coil-type sensing element;

[0010] The signal control and demodulation instrument is used to provide current excitation to the excitation coil and to monitor the slippage state of the steel strands in the external prestressed bundle through diagnosis based on the acquired magnetic characteristic electrical signals.

[0011] This electromagnetic detection device achieves non-destructive testing of the slippage state of steel strands in external prestressed bundles by coordinating the electromagnetic excitation field with the sensing signal.

[0012] Furthermore, the number of excitation coils is one, which is wound around the outer circumferential surface of a certain layer of steel strands outside the external prestressed strand anchor head through a coil frame, and ensures that the number of steel strands inside the coil is approximately close to the number of steel strands outside the coil, thereby providing a uniformly distributed excitation magnetic field for the steel strands outside the anchor head.

[0013] Furthermore, the coil-type sensing element adopts a ring layout, and its number is configured according to the coil arrangement of the steel strands inside the anchor head. Each coil-type sensing element is wound around the outer circumferential surface of the corresponding layer of steel strands through a coil frame, so as to realize independent monitoring of the slippage state of each layer of steel strands.

[0014] Furthermore, the signal control and demodulation instrument acquires the magnetic characteristic electrical signals of each coil-type sensing element in real time, and diagnoses whether the anchorage status of the external prestressed strand is normal through an algorithm model; when an abnormal state is detected, it can determine the location and number of steel strand slippages and provide diagnostic monitoring information.

[0015] Furthermore, the algorithm model employs a pre-trained fully connected neural network, whose inputs are the magnetic characteristic electrical signals of each coil-type sensing element and the ambient temperature, and whose outputs are the number and position of the slip strands.

[0016] Furthermore, in the pre-training process of the fully connected neural network: firstly, based on the finite element simulation model or laboratory calibration, the magnetic characteristic electrical signals, ambient temperature, and the number and location of slipped steel strands of each coil-type sensing element under all damage conditions are obtained as training data samples; then, the data samples are normalized and divided into training set, validation set, and test set according to the proportions; the neural network parameters are initialized, including the bias vector and weight matrix of each layer, learning rate, and optimizer; the training set samples are input one by one into the fully connected neural network, and the neural network outputs the corresponding prediction results, i.e., the number and location of slipped steel strands, through forward propagation; the loss function between the prediction result and the label is calculated; based on the loss function, the network parameters are iteratively updated using the gradient descent method by the optimizer until the loss function converges or reaches the maximum number of iterations, and the training is completed; after the training is completed, the neural network is validated using the validation set samples, and the neural network that performs best on the validation set is used as the final algorithm model.

[0017] Furthermore, the loss function is the sum of the slip quantity prediction error and the slip position prediction error.

[0018] A method for diagnosing the slippage state of external prestressed tendons based on the above-mentioned electromagnetic detection device includes the following steps:

[0019] (1) Use a signal control and demodulator to generate an excitation signal, drive the excitation coil to generate a magnetic field, and uniformly magnetize the steel strands on the outside of the prestressed tendon anchor head.

[0020] (2) Use a multi-channel acquisition instrument to acquire the magnetic characteristic electrical signals generated by each coil-type sensing element, and amplify and filter these electrical signals;

[0021] (3) Input the preprocessed electrical signal and ambient temperature together into the built-in algorithm model of the signal control and demodulator. The model outputs diagnostic results about the number and position of the slip strands.

[0022] This diagnostic method uses the output signal of a coil-type sensing element and the ambient temperature as inputs, and through a trained fully connected neural network model, it can accurately determine the number and location of slippage of the prestressed strands in the external prestressed bundle.

[0023] This invention aims to address the shortcomings of existing external prestressed tendon anchorage status detection technologies. Its technical solution, as an innovative non-contact, real-time monitoring method, has the technical advantage of not altering the original state and performance of the external prestressed tendons. It can achieve precise positioning and quantitative assessment of slippage, and has significant application value for the safety monitoring and maintenance of external prestressed tendons.

[0024] This invention's electromagnetic detection device achieves high-precision, non-destructive testing of the anchorage status of externally prestressed tendons through the combination of an electromagnetic excitation field and an array of coil-type sensing elements. The device generates a controllable electromagnetic field through a signal control subsystem and acquires the detection signal in real time through a signal demodulation subsystem. Combined with a fully connected neural network for intelligent analysis, it can accurately determine the number and location of slippage. This invention's electromagnetic detection method has advantages such as strong real-time performance, wide adaptability, and high detection accuracy, effectively overcoming the problems of large interference, low accuracy, and poor real-time performance of traditional detection technologies. It can safely and reliably conduct long-term monitoring in complex environments, providing reliable data support for the safe operation of externally prestressed tendon bridges. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the electromagnetic detection device for monitoring slippage of external prestressed tendons according to the present invention.

[0026] Figure 2 This is a schematic diagram of the installation of the excitation coil and the array-type coil sensing element, where (a) to (d) correspond to the installation configurations in four typical anchorages.

[0027] Figure 3 A schematic diagram of the magnetic flux circuit generated by the excitation coil.

[0028] Figure 4 This is a schematic diagram of the fully connected neural network used in the diagnostic method of the present invention and its training process.

[0029] Figure 5 This is a schematic diagram of the diagnostic steps for slippage of external prestressed tendons according to the present invention.

[0030] Figure 6 This is a schematic diagram illustrating the monitoring results of the slippage state of the external prestressed bundle using a fully connected neural network, as described in this invention.

[0031] In the diagram: 1—Signal control and demodulator, 2—Signal connection line, 3—Display, 4—Excitation coil, 5—Coil-type sensing element, 6—Multi-channel acquisition instrument, 7—External prestressed tendon anchor plate, 8—External prestressed tendon anti-loosening plate, 9—Steel strand, 10—Inner layer steel strand, 11—Outer layer steel strand. Detailed Implementation

[0032] To describe the present invention in more detail, the technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] like Figure 1As shown, the electromagnetic detection device for monitoring slippage of external prestressed tendons according to the present invention includes a signal control and demodulator 1, a display 3, an excitation coil 4, a coil-type sensing element 5, and a multi-channel acquisition instrument 6. The signal control and demodulator 1 is connected to the excitation coil 4 to generate a controllable electromagnetic excitation field. The coil-type sensing element 5 captures the magnetic characteristic electrical signals. The multi-channel acquisition instrument 6 collects the signals from multiple coil-type sensing elements 5. Finally, the signal control and demodulator 1 performs intelligent diagnosis.

[0034] There is one excitation coil 4, which is uniformly wound on the coil frame and arranged outside the external prestressed tendon anchor head. This ensures that the number of steel strands inside the excitation coil 4 is as close as possible to the number of steel strands outside, and ensures that the steel strands outside the anchor head form a uniformly distributed excitation magnetic field.

[0035] The coil-type sensing element 5 adopts a ring layout design, and its number is strictly configured according to the coil arrangement characteristics of the steel strands 9 inside the anchor head. Each sensing element 5 is arranged on the outer circumferential surface of the corresponding layer of steel strands 9 through a coil frame to monitor the slippage state of each layer of steel strands 9. In this embodiment, the specific implementation structure of the coil-type sensing element 5 is as follows: each sensing element is composed of a ring coil wound with copper core polyurethane enameled round copper wire, which is fixed on the coil frame. The coil frame is customized according to the anchor specifications (such as a 27-hole anchor).

[0036] The coil frame is used to provide mechanical support. Its number is the sum of the required excitation coils 4 and the total number of coil-type sensing elements 5. The size of the coil frame is reasonably determined according to the specifications of the external prestressed tendon anchorage to adapt to different anchorage structures.

[0037] The signal control and demodulator 1 can generate pulse signals or AC signals and is connected to the excitation coil 4 for applying excitation signals.

[0038] The multi-channel acquisition instrument 6 is connected to each coil-type sensing element 5 through the signal connection line 2, so as to realize the independent acquisition of the magnetic characteristic electrical signal of each coil-type sensing element 5.

[0039] The signal control and demodulation instrument 1 can acquire the magnetic characteristic electrical signal of the coil-type sensing element 5 in real time, and determine whether the anchorage status of the external prestressed strand is normal through intelligent algorithm; when an abnormal state is detected, it can determine the location and number of steel strand slippage, and then transmit the monitoring and diagnostic information to the display 3 for display through the signal connection line 2.

[0040] This embodiment takes an external prestressed tendon with 27 anchorages as an example, which includes an external prestressed tendon anchor plate 7, an external prestressed tendon anti-loosening plate 8, and 27 steel strands 9.

[0041] Both the excitation coil 4 and the coil-type sensing element 5 are installed on the outside of the external prestressed tendon anti-loosening plate 8. The excitation coil 4 is installed outside the second layer of steel strands, so that there are 12 steel strands inside the excitation coil 4 and 15 steel strands outside the excitation coil 4. The coil-type sensing element 5 is installed on the outer periphery of each layer of steel strands 9, for a total of three coil-type sensing elements 5 installed on three layers of steel strands 9.

[0042] The excitation coil 4 and the coil-type sensing element 5 are designed in shape and size according to the model of the external prestressed tendon anchor to be measured. Figure 2 Examples of installation configurations of the excitation coil 4 and the coil-type sensing element 5 in several typical anchor models are shown.

[0043] The magnetic flux loop generated in the steel strand by the excitation coil 4 under the action of the excitation signal is as follows: Figure 3 As shown, the inner steel strand 10 generates an upward magnetic flux, which enters the outer steel strand 11 through the air or anti-corrosion grease. The magnetic flux of the outer steel strand 11 is directed downward, and finally returns to the inner steel strand 10 through the air or anti-corrosion grease, thus forming a closed magnetic flux loop.

[0044] Based on the above-described detection device structure, this embodiment also provides a fully connected neural network-based algorithm for diagnosing and monitoring the slippage state of external prestressed tendons. This algorithm is integrated into a signal control and demodulation unit, specifically:

[0045] First, based on the finite element simulation model or actual measurement calibration, the magnetic characteristic electrical signals, temperature, slippage quantity, and position distribution of the coil-type sensing elements under all damage conditions are obtained as the dataset for training the fully connected neural network. The fully connected neural network is then trained using this dataset. The inputs to the fully connected neural network are the output electrical signals and temperature of each coil-type sensing element, and the outputs are the slippage quantity and position distribution.

[0046] Input the dataset into... Figure 4 In the fully connected neural network shown, the dataset is normalized so that the data values ​​in the training, validation, and test sets are all within [0,1]. The dataset is divided into training, validation, and test sets in a 3:1:1 ratio. The training performance evaluation index of the fully connected neural network is the sum of the slip position prediction error and the slip number prediction error. Through continuous training and testing, the slip position and slip number output by the neural network meet the accuracy requirements.

[0047] The process of using a trained fully connected neural network for the diagnosis and monitoring of slippage of external prestressed tendons is as follows: Figure 5 As shown:

[0048] Step 1: Magnetize the outer steel strand at the anchoring end by generating an excitation magnetic field through an excitation coil.

[0049] The second step is to acquire the magnetic characteristic electrical signal through a coil-type sensing element and amplify and filter it in the time domain.

[0050] The third step is to input the magnetic characteristic electrical signals collected by the coil-type sensing element and the ambient temperature into the trained fully connected neural network to perform intelligent diagnosis of the anchorage status of the external prestressed tendons.

[0051] Figure 6 The results of monitoring slippage of external prestressed strands in this embodiment show that the monitoring accuracy for early slippage of external prestressed strands, i.e., when the total number of slipped steel strands is less than 6, reaches 100%. This demonstrates that the electromagnetic detection device and method for external prestressed strand slippage of the present invention have high reliability and can meet the requirements of actual external prestressed strand slippage monitoring.

[0052] The above description of the embodiments is provided to enable those skilled in the art to understand and apply the present invention. Those skilled in the art can readily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without creative effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made to the present invention by those skilled in the art based on the disclosure thereof should be within the scope of protection of the present invention.

Claims

1. An electromagnetic detection device for monitoring the slippage of an externally prestressed tendon, characterized in that Comprise: An excitation coil arranged outside the external prestressed tendon anchor head for generating a magnetic field to magnetize the steel strand under current excitation; A plurality of coil-type sensing elements installed outside the external prestressed tendon anchor head to generate corresponding magnetic characteristic electrical signals through electromagnetic induction; A multi-channel acquisition instrument for independently collecting the magnetic characteristic electrical signals generated by each coil-type sensing element; A signal control and demodulation instrument for providing current excitation for the excitation coil and monitoring the steel strand slip state in the external prestressed tendon through diagnosis based on the collected magnetic characteristic electrical signals; The number of the excitation coil is one, which is wound on the outer peripheral surface of a certain layer of steel strand outside the external prestressed tendon anchor head through the coil framework, and the number of steel strands inside the coil is approximately close to the number of steel strands outside the coil, thereby providing a uniformly distributed excitation magnetic field for the steel strands outside the anchor head; The coil-type sensing elements adopt a ring layout, and their number is configured according to the distribution of the steel strands inside the anchor head, and each coil-type sensing element is wound on the outer peripheral surface of the corresponding layer of steel strand through the coil framework; The signal control and demodulation instrument acquires the magnetic characteristic electrical signals of each coil-type sensing element in real time, diagnoses whether the anchoring state of the external prestressed tendon is normal through an algorithm model, and when an abnormal state is detected, it can determine the position and number of the slipped steel strands and provide diagnostic monitoring information.

2. The electromagnetic detection apparatus of claim 1, wherein: The algorithm model uses a pre-trained fully connected neural network, whose input is the magnetic characteristic electrical signals of each coil-type sensing element and the environmental temperature, and the output is the number and position of the slipped steel strands.

3. The electromagnetic detection apparatus of claim 2, wherein: In the pre-training process of the fully connected neural network, first, the magnetic characteristic electrical signals of each coil-type sensing element, the environmental temperature, the number and position of the slipped steel strands under all damage conditions are obtained according to the finite element simulation model or laboratory calibration as data samples for training; then the data samples are normalized and divided into training set, validation set and test set in proportion; the neural network parameters are initialized, including the bias vector and weight matrix of each layer, learning rate and optimizer; the training set samples are input into the fully connected neural network one by one, and the corresponding prediction results, i.e. the number and position of the slipped steel strands, are obtained by forward propagation of the neural network, the loss function between the prediction results and the labels is calculated, and the network parameters are updated iteratively by gradient descent method according to the loss function using the optimizer until the loss function converges or the maximum number of iterations is reached, and the training is completed; after the training is completed, the neural network is verified using the validation set sample, and the neural network with the best performance on the validation set is used as the final algorithm model.

4. The electromagnetic detection apparatus of claim 3, wherein: The loss function is the sum of the slip number prediction error and the slip position prediction error.

5. A method for diagnosing the slip state of an external prestressed tendon based on the electromagnetic detection device according to any one of claims 1-4, comprising the following steps: (1) generating an excitation signal using the signal control and demodulation instrument to drive the excitation coil to generate a magnetic field and uniformly magnetize the steel strands outside the external prestressed tendon anchor head; (2) The magnetic characteristic electric signals generated by each coil type sensing element are collected by a multi-channel acquisition instrument, and the electric signals are amplified and filtered; (3) The preprocessed electric signals and the environmental temperature are input into the algorithm model built in the signal control and demodulation instrument, and the model outputs the diagnosis results about the number and position of the slipped steel strands.

6. The method of claim 5, wherein: The algorithm model adopts a pre-trained fully connected neural network, the input of which is the magnetic characteristic electric signals of each coil type sensing element and the environmental temperature, and the output is the number and position of the slipped steel strands; in the pre-training process of the fully connected neural network, firstly, the magnetic characteristic electric signals of each coil type sensing element, the environmental temperature, the number and position of the slipped steel strands under all damage conditions are obtained according to a finite element simulation model or laboratory calibration, as data samples for training; then the data samples are normalized and divided into a training set, a validation set and a test set in proportion; the neural network parameters are initialized, including the bias vector and weight matrix of each layer, the learning rate and the optimizer; the training set samples are input into the fully connected neural network one by one, the corresponding prediction results, i.e. the number and position of the slipped steel strands, are obtained by forward propagation of the neural network, the loss function between the prediction results and the labels is calculated, the network parameters are updated by gradient descent method through the optimizer according to the loss function, until the loss function converges or the maximum number of iterations is reached, and the training is completed; after the training is completed, the neural network is verified by using the validation set samples, and the neural network with the best performance on the validation set is taken as the final algorithm model.

Images