Preparation method of silver-based flexible sensor for identifying existing crack of bridge structure

Abstract

The invention discloses a preparation method of a silver-based flexible sensor for existing crack recognition of a bridge structure. The method comprises the following steps: step 1, designing a flexible sensor bionic framework based on human skin gradient elastic modulus characteristics; step 2, aiming at existing crack identification, referring to unique biomechanical characteristics and perceptual ability of human skin, and based on a bionic architecture, modifying a conductive silver nano-material on the surface of a thermoplastic polyurethane film by using a smart gravity-driven phase separation strategy caused by a concentration gradient, so as to obtain a conductive silver nano-material modified thermoplastic polyurethane film; the nano-composite film flexible sensor with the capability of being attached to and adapting to bridge surface change is formed; and step 3, integrating a power supply module, a data acquisition and preprocessing module and a data transmission module on the basis of the flexible sensor to obtain the flexible multifunctional integrated sensor. The sensor prepared by the method has high sensitivity, good flexibility and excellent durability, and can realize accurate identification and real-time monitoring of existing cracks of a bridge structure.

Description

Technical Field

[0001] The present invention belongs to the field of health monitoring of existing cracks in bridge structures, and relates to a preparation method of a flexible sensor, specifically to a preparation method of a silver-based flexible sensor for identifying existing cracks in bridge structures. Background Art

[0002] Accurately monitoring the development and evolution of existing cracks in bridge structures can effectively ensure the operational safety of bridge structures. As a device for detecting and transmitting information, it is crucial to develop its application in enhancing or replacing manual intervention in equipment. With the increase in the service life of bridges, the problem of cracks in bridge structures has gradually emerged, posing a serious threat to the safety, stability, and service life of bridges. Traditional bridge crack detection technologies, such as ultrasonic detection and impact elastic wave method, although can identify cracks to a certain extent, have problems such as complex operation, high cost, poor real-time performance, etc., and it is difficult for traditional sensors to achieve close fitting with the structure surface, resulting in low detection accuracy. Therefore, developing a new type, efficient, and low-cost bridge crack identification technology has important practical application value. Summary of the Invention

[0003] The present invention provides a preparation method of a silver-based flexible sensor for identifying existing cracks in bridge structures. The sensor prepared by this method has high sensitivity, good flexibility, and excellent durability, and can achieve precise identification and real-time monitoring of existing cracks in bridge structures.

[0004] The object of the present invention is achieved by the following technical solutions:

[0005] A preparation method of a silver-based flexible sensor for identifying existing cracks in bridge structures, comprising the following steps:

[0006] Step 1: Design a bionic architecture of a flexible sensor based on the characteristics of the gradient elastic modulus of human skin. The bionic architecture includes a base layer, a sensing layer, and a surface layer. The base layer and the surface layer are thermoplastic polyurethane (TPU) film layers, and the sensing layer includes one or more layers of conductive silver nanomaterial layers;

[0007] Step 2: For identifying existing cracks, drawing on the unique biomechanical properties and sensing capabilities of human skin, based on the bionic architecture designed in Step 1, use the dexterous gravity-driven phase separation strategy caused by the concentration gradient to modify the conductive silver nanomaterials on the surface of the thermoplastic polyurethane film to form a nano-composite film flexible sensor with the ability to adhere to and adapt to the changes on the bridge surface;

[0008] Step 3: Integrate a power module, a data acquisition and preprocessing module, and a data transmission module on the basis of the flexible sensor prepared in Step 2 to obtain a flexible multi-functional integrated sensor.

[0009] Compared with the prior art, the present invention has the following advantages:

[0010] 1. The present invention simulates the different responses of the human skin when subjected to external pressure or deformation, and the flexible strain sensor generates different electrical signal waveforms accordingly, realizing the recognition and monitoring of different strain signals. In view of the above characteristics, for the recognition of existing cracks, a bionic architecture of a flexible sensor based on the characteristics of the gradient elastic modulus of the human skin is designed.

[0011] 2. Based on the bionic architecture and oriented to the recognition of existing cracks, the present invention designs the asymmetry of the sensor array, and based on this, the flexible sensor realizes selective multi-functional responses to bending, compression, and stretching.

[0012] 3. The present invention uses the RNN+Attention attention mechanism to process the problem of crack trend recognition. RNN is good at processing sequence data and can be applied to analyze the time series data of crack development. On this basis, the attention mechanism is added, which can help the model focus on the most relevant parts of the input data and extract the characteristics of the flexible sensor response signal. On this basis, a fuzzy logic system and a simulated annealing algorithm are adopted to improve the recognition accuracy of the development and change of existing cracks.

[0013] 4. The present invention can effectively realize the interaction between the flexible sensor and the monitoring environment, and the flexible sensor for monitoring existing cracks has the characteristics of high precision, fast response, and can adapt to the surface deformation of the structure, and is easy to realize distributed monitoring. Compared with the existing methods, the present invention improves the accurate recognition rate of existing cracks in bridges. Brief Description of the Drawings

[0014] Figure 1 It is a flowchart of the preparation method of a silver-based flexible sensor for the recognition of existing cracks in bridge structures.

[0015] Figure 2 It is a bionic architecture diagram of a silver-based flexible sensor, with the green being the flexible base layer and the orange being the conductive sensing layer.

[0016] Figure 3 It is an asymmetric structure diagram of a silver-based flexible sensor array.

[0017] Figure 4 It is a data processing flowchart of the RNN+Attention attention mechanism.

[0018] Figure 5 It is an existing cracked concrete slab;

[0019] Figure 6 It is a test pressure frame;

[0020] Figure 7 It is a schematic diagram of sensor pasting;

[0021] Figure 8 Set for simply supported conditions;

[0022] Figure 9 For initial crack width measurement;

[0023] Figure 10 Schematic diagram of load test;

[0024] Figure 11 Resistance change of end - pasting method;

[0025] Figure 12 Resistance change of overall - pasting method;

[0026] Figure 13 Resistance change rate of end - pasting method;

[0027] Figure 14 Resistance change rate of overall - pasting method. Specific implementation manner

[0028] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings, but it is not limited thereto. Any modification or equivalent replacement of the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention shall be covered by the protection scope of the present invention.

[0029] The present invention provides a preparation method of a silver - based flexible sensor for identifying existing cracks in bridge structures, as Figure 1 shown. The method includes the following steps:

[0030] Step 1: Simulate that the human skin has different reactions when subjected to external pressure or deformation, and the flexible strain sensor then generates different electrical signal waveforms to realize the identification and monitoring of different strain signals. In view of the above characteristics, when applied to the identification of existing cracks, a flexible sensor bionic architecture based on the gradient elastic modulus characteristics of the human skin is designed. The specific steps are as follows:

[0031] Step 1 - 1: Based on the gradient elastic modulus characteristics, the skin can adapt to various deformations and accurately sense external stimuli. Simulate this skin characteristic and design a multi - level flexible sensor bionic architecture, which includes a base layer, a sensing layer, and a surface layer. Among them: the thickness, density, and elastic modulus between each layer are adjusted according to needs to achieve the required gradient distribution, so as to adapt to the complex changes of cracks on the bridge surface.

[0032] Step 1 - 2: Selecting appropriate materials is the key to realizing the gradient elastic modulus characteristics. Select conductive silver - nanomaterials to construct different sensing layers of the sensor.

[0033] Through the above steps, a flexible sensor bionic architecture for identifying and monitoring existing cracks in bridge structures is designed.

[0034] Step 2: For the identification of existing cracks, drawing on the unique biomechanical properties and sensing capabilities of human skin, based on the bionic architecture of the flexible sensor designed in Step 1, a novel AgNW@TPU nanocomposite film flexible strain sensor with the ability to adhere to and adapt to the changes in the bridge surface is formed by using a dexterous gravity-driven phase separation strategy induced by a concentration gradient to modify the thermoplastic polyurethane film with conductive silver nanomaterials (AgNWs). The specific steps are as follows:

[0035] Step 2-1: Film pretreatment: Clean and pretreat the TPU film to ensure its surface is flat and clean, which is conducive to the adhesion of AgNWs.

[0036] Step 2-2: Prepare a series of AgNWs solutions with different concentrations, and the concentration gradient of these solutions will be used for the subsequent gravity-driven phase separation process. Through precise concentration formulation and mixing, ensure that the concentration of the AgNWs solution forms a continuous gradient from high to low.

[0037] Step 2-3: Based on the bionic structure proposed in Step 1, with the identification of existing cracks as the guidance, design the asymmetry of the sensor array. By controlling the concentration gradient of AgNWs in the solution, use gravity drive to achieve the phase separation and orderly arrangement of AgNWs on the surface of the TPU film, and obtain an AgNW@TPU nanocomposite film flexible strain sensor with the ability to adhere to and adapt to the changes in the bridge surface. The specific steps are as follows:

[0038] Step 2-3-1: Impregnation process: Immerse the TPU film into the AgNWs solution.

[0039] Step 2-3-2: Phase separation control: With the identification of existing cracks as the guidance, by controlling parameters such as the impregnation speed, time, and temperature, the distribution and arrangement of AgNWs on the surface of the TPU film can be precisely controlled.

[0040] Step 2-3-3: Through appropriate drying and curing treatments, make AgNWs firmly embedded on the TPU film, generally controlling the temperature between 100 and 130 °C.

[0041] Steps Two, Three, and Four: According to the gradient elastic modulus characteristics of the bionic architecture, combined with the possible directions and morphologies of cracks in practical applications, starting from a low-concentration AgNWs solution and gradually immersing it into a high-concentration AgNWs solution in sequence, repeat Steps Two, Three, and One to Steps Two, Three, and Three for multi-layer composite and surface treatment to construct an asymmetric sensing layer layout. Based on this, the AgNW@TPU nanocomposite film flexible sensor can achieve selective multi-functional responses to bending, compression, and tension, further improving the performance of the AgNW@TPU nanocomposite film. In this process, gravity will drive the AgNWs to move from the high-concentration area to the low-concentration area and form an orderly arrangement on the surface of the TPU film.

[0042] In this step, in order to ensure that the AgNW@TPU nanocomposite film flexible sensor can accurately identify existing cracks on structures such as bridges and achieve selective responses to various deformations such as structural strain and deflection changes in bending, compression, and tension, based on the principle of the bionic architecture, the characteristics of the human skin that can produce different responses when subjected to different stimuli are simulated. On this basis, the asymmetric structure of the sensor array is designed. In the design process of the sensor array, according to the gradient elastic modulus characteristics of the bionic architecture, combined with the possible directions and morphologies of cracks in practical applications, an asymmetric sensing layer layout is constructed. The differences in the size, shape, and arrangement of the sensing layer are constructed to adapt to deformations and pressure changes in different directions. The specific design requirements are as follows:

[0043] (1) Design the directional arrangement of the sensors: According to the possible directions of cracks, design a sensing layer with a directional arrangement to ensure that the sensor can produce a more obvious response under deformations or pressures in a specific direction.

[0044] (2) Design the density difference of the sensors: In different regions of the sensor, adjust the density of the sensing layer. In areas prone to cracks, increase the density of the sensing layer to improve the sensitivity and accuracy of detection.

[0045] (3) Design the multi-level structure of the sensors: Combining the gradient elastic modulus characteristics of the bionic architecture, design sensors with a multi-level structure. The sensing layers at different levels have different elastic moduli and sensitivities to meet the monitoring requirements under different deformation degrees.

[0046] Step Two, Four: Test the performance of the AgNW@TPU nanocomposite film flexible strain sensor obtained in Step Two, Three:

[0047] Adhesion test: Test the adhesion performance of the sensor on the surface of the bridge structure to ensure that it can adhere tightly and adapt to the complex changes on the bridge surface;

[0048] Sensitivity test: By applying different pressures and deformations, test the sensitivity and response speed of the sensor;

[0049] Durability test: Simulate the environmental conditions of the bridge structure in actual use to conduct a durability test on the sensor to ensure that it can work stably for a long time in a harsh environment.

[0050] Step 3: For the identification of existing cracks, integrate a power module, a data acquisition and preprocessing module, and a data transmission module on the basis of the flexible strain sensor prepared in Step 2 to obtain a flexible multifunctional integrated sensor. The specific steps are as follows:

[0051] Step 3-1: According to the requirements of crack identification, design the functions and structures of parts such as the power module, the data acquisition and preprocessing module, and the data transmission module. Among them: the power module is responsible for providing stable electrical energy for the entire sensor; the data acquisition and preprocessing module is responsible for collecting the response signals of the flexible sensor, and performing filtering, amplification, identification of existing cracks, and judgment of the crack development trend; the data transmission module is responsible for wirelessly transmitting the processed data to the host computer or the cloud server. Design flexible connectors based on flexible materials to connect each module and ensure the stability and reliability of the sensor in a complex environment.

[0052] In this step, the method for the data acquisition and preprocessing module to collect the response signals of the flexible sensor, and perform filtering, amplification, identification of existing cracks, and judgment of the crack development trend is as follows: Use the RNN+Attention attention mechanism to process the crack trend identification problem. RNN is good at processing sequence data and can be applied to analyze the time series data of crack development. On this basis, the attention mechanism is added, which can help the model focus on the most relevant parts of the input data and extract the characteristics of the flexible sensor response signals; on this basis, use the fuzzy logic system and the simulated annealing algorithm to improve the accuracy of identifying the development and change of existing cracks. The specific steps are as follows:

[0053] (1) Data collection: Collect the time series data of crack development. These data come from the response signals of the flexible sensor at different positions on the surface of the structure. The response signals include the sensor's compressive response, the sensor's tensile response, and the sensor's bending response. Among them:

[0054] Sensor compressive response: When the composite material is compressed, the space between the conductive silver nanomaterials decreases, and the conductive network becomes denser, resulting in a decrease in resistance. When subjected to pressure, the sensing layer will generate different electrical signal outputs according to the distribution and magnitude of the pressure, so as to achieve accurate identification of the pressure magnitude and position, as shown in formula (1):

[0055]

[0056] Wherein, p is the applied pressure (MPa); G is the compression modulus of the composite material (MPa); R is the resistance of the conductive particles (Ω); R0 is the initial resistance of the conductive particles (Ω); m is the mass of one electron (G); h is Planck's constant; d0 is the gap width (nm); is the percolation index.

[0057] Sensor tensile response: The uniformly distributed conductive silver nanomaterials are interconnected in the polymer matrix to form a reliable conductive framework. During the tensile process, due to the sliding of the two-dimensional nano-structures, these connections gradually decrease, resulting in an increase in resistance. When the sensor array is subjected to tensile deformation, due to the differential design of the sensing layer in terms of size and arrangement, the sensor can identify the direction and degree of the tensile force and output the corresponding electrical signal, as shown in Equation (2):

[0058]

[0059] Wherein, γ is the percolation coefficient of the composite material.

[0060] Sensor bending response: When the sensor array is subjected to bending deformation, due to the asymmetry and directional arrangement of the sensing layer, the sensor can identify the direction and degree of the bending and generate corresponding changes in the electrical signal.

[0061] The relationship between the bending angle and the motion value can be expressed by Equation (3):

[0062]

[0063] Wherein, L0 represents the initial length of the sensor; ΔL represents the change value of the sensor length; α represents the bending angle of the sensor; b represents the tangent length of the sensor.

[0064] When the sensor is in an ideal state, b can be taken as L0 / 2, and substituting it into Equation (3), it is converted into Equation (4):

[0065]

[0066] When the sensor bends outward or inward at any angle, L0 and ΔL can be directly measured, and α can be calculated by Equation (4). Then, comparing the calculation results with the calculation results of Equation (5) to verify the sensor effect:

[0067]

[0068] Wherein, ΔR represents the change value of the resistance of the signal element; Δl represents the tensile amount of the substrate; K represents the sensitivity coefficient of the sensor.

[0069] (2) Data preprocessing: Clean the collected data to remove noise and outliers, ensuring data accuracy and reliability. Standardize the data so that data from different sources and scales can be compared and analyzed within the same framework.

[0070] (3) Build an RNN-based crack trend recognition model: The cyclic structure of RNN enables it to capture the temporal dependencies in sequential data and analyze the development and changes of cracks over time. Train the RNN model using the preprocessed data, and optimize the model's performance by adjusting model parameters and training strategies to accurately identify the development trend of cracks.

[0071] Initialize the hidden state, that is, set the initial hidden state to a vector of all zeros.

[0072] h0 = 0 (6)

[0073] Hidden state update formula:

[0074] h t = tanh(W hh h t-1 + W hx x t + b h ) (7)

[0075] In the formula, h t represents the hidden state at the current moment; h t-1 represents the hidden state at the previous moment; x t represents the input at the current moment; W hh represents the weight matrix from the hidden state to the hidden state; W hx represents the weight matrix from the input to the hidden state; b h represents the hidden state offset.

[0076] Output calculation formula:

[0077] y t = σ(W yh h t + b y ) (8)

[0078] In the formula, y t represents the output at the current moment; W yh represents the hidden state at the previous moment; b y represents the input at the current moment.

[0079] On this basis, establish a loss function model:

[0080]

[0081] Wherein, L represents a certain loss function, such as cross-entropy loss; represents the correct output.

[0082] (4) On the basis of the RNN model, an Attention mechanism is introduced. By calculating the attention weights of each position in the input sequence, the model is helped to focus on the part most relevant to the crack development trend. During the training process, the attention weights are optimized through the backpropagation algorithm, enabling the model to more accurately capture the key information in the input data.

[0083] On this basis, the calculation formula for the hidden state of RNN+Attention is proposed:

[0084] h t = f(h t-1 , x t ) (10)

[0085] Wherein, h t represents the hidden state at the current moment; h t-1 represents the hidden state at the previous moment; x t represents the input at the current moment; f represents the transfer function of the RNN cell.

[0086] The calculation formula for the attention score is established:

[0087] e t,i = g(h t , s i ) (11)

[0088] Wherein, e t,i represents the attention score at the current moment; h t represents the hidden state at the current moment; s i represents the i-th hidden state in the encoded sequence; g represents the scoring function of the attention mechanism.

[0089] The attention weight formula is introduced:

[0090]

[0091] Wherein, α t,i represents the attention weight of the i-th encoded hidden state at the current moment.

[0092] On this basis, the calculation formula for the context vector is established:

[0093]

[0094] Wherein, c t represents the context vector at the current moment, which is the weighted sum of the hidden states in the encoded sequence.

[0095] Next, construct the calculation formula for the decoder hidden state:

[0096] s t = f(s t-1 , y t-1 , c t ) (14)

[0097] In the formula, s t represents the decoder hidden state at the current moment; s t-1 represents the decoder hidden state at the previous moment; y t-1 represents the output at the previous moment; c t represents the context vector at the current moment; f represents the transition function of the decoder.

[0098] Use the RNN and Attention mechanisms to extract features from the flexible sensor response signals, which may include the width, length, development speed, etc. of the cracks. Fuse the extracted features to form a comprehensive feature vector for characterizing the development trend of the cracks.

[0099] (5) Construct a fuzzy logic system. A fuzzy logic system is a logic system based on fuzzy set theory for processing imprecise information, allowing variables to vary between 0 and 1, and is used to handle the uncertainties in the development of cracks. Define fuzzy sets, fuzzy rules, and fuzzy inference mechanisms, transform the problem of identifying the development trend of cracks into a fuzzy logic problem, and use the simulated annealing algorithm to optimize the parameters of the fuzzy logic system. Apply the optimized fuzzy logic system to the identification of crack development changes and output the identification results. The specific steps of defining fuzzy sets, fuzzy rules, and fuzzy inference mechanisms include: defining fuzzy variables and their membership functions, establishing fuzzy rules, selecting the Mamdani inference mechanism, and converting the fuzzy output into a clear result using the centroid method and the maximum membership degree method to identify the development trend of cracks. Use the simulated annealing algorithm to avoid getting stuck in local optimal solutions during the search process and improve the accuracy of crack development change identification.

[0100] (6) Use an independent test dataset to evaluate the fuzzy logic system constructed in step (4), calculate indicators such as the accuracy and recall rate of the model, and evaluate the performance of the model. Improve the model according to the evaluation results, including adjusting the model structure, optimizing parameter settings, introducing new features, etc., to improve the performance and accuracy of the model.

[0101] Step 32: Assemble the power module, data acquisition and preprocessing module, data transmission module, and flexible strain sensor through flexible connectors to form a complete flexible multifunctional integrated sensor. During the assembly process, it is necessary to ensure that the connection between the modules is firm and reliable. Perform functional tests on the assembled sensor, including the output voltage of the power module, the response speed of the data acquisition and preprocessing module, the transmission distance and stability of the data transmission module, etc. The test results meet the design requirements.

[0102] A method for identifying existing cracks in a bridge structure using the flexible strain sensor prepared by the above method comprises the following steps:

[0103] Step 1: Install the flexible multifunctional integrated sensor on the surface of the structure to be monitored. The flexible strain sensor simulates the reaction of human skin when it is subjected to external pressure or deformation. The change of the flexible sensor response signal is used to identify the existence and development trend of cracks. The specific method is as follows:

[0104] Step 1: Accurately capture and measure these micro-to-macroscopic crack deformations through changes in the electrical signal waveform. On this basis, the rate, direction, and frequency parameters of the deformation are determined through changes in the electrical signal waveform, so as to achieve the purpose of real-time and continuous monitoring of the changes in existing cracks in the bridge structure, and use flexible strain sensors to detect the existence and expansion of cracks.

[0105] Step 1 and 2: When a crack appears, different elastic modulus layers in the flexible strain sensor will be subjected to different degrees of stress concentration, causing changes in electrical parameters such as resistance inside the sensor. By monitoring the changes in electrical parameters, the existence and expansion of the crack can be accurately identified.

[0106] Step 2: The sensor transmits the collected data wirelessly to the host computer or cloud server for further noise reduction and data mining.

[0107] Step 3: Based on the treatment results, determine the severity and development trend of the cracks to provide a scientific basis for the maintenance and repair of the bridge structure.

[0108] Example:

[0109] In order to obtain the changing law of the electrical parameters of the sensor during the crack development under the action of load, this embodiment uses a beam with a single crack as a carrier, pastes the trial-produced sensor at the crack development point, connects a digital milliohm meter, measures the resistance change in real time, draws a load-resistance change diagram, and verifies the accuracy of the model and the stability and practicality of the sensor.

[0110] 1. Experimental design

[0111] The materials and equipment used in the test are as follows: epoxy resin glue, digital milliohm meter, digital hydraulic press, concrete slab with cracks.

[0112] Table 1 Test Equipment for Crack Identification

[0113]

[0114] Test conditions: The resistance change is measured by a digital milliohm meter. To make the resistance measurement more stable, silver tape is used to bond and lead out the wires at both ends of the sensor. The two ends of the slab are simply supported and constrained, and a concentrated force is applied vertically. The magnitude and change rate of the vertical concentrated force are controlled by a hydraulic jack. A reaction frame is set at the upper end, and the lower end is loaded by the jack. The resistance change is displayed in real time by the digital milliohm meter.

[0115] Test procedure:

[0116] (1) Material preparation: The sensors used in this test are self-made flexible sensors. There are 2 plain concrete slabs, divided into a slab with cracks and a slab without cracks. The structural dimensions are: 32 cm × 25 cm × 4 cm, a pressure test bench, a digital milliohm meter (with accurate readings to the milliohm level), and epoxy resin glue.

[0117] (2) Sensor pasting: To obtain the resistance change law under known crack development conditions, first use silver glue to paste both ends of the sensor, lead out the wires, measure and record the initial resistance of the sensor used, and record the initial crack width value. Connect the milliohm meter to one of the sensors first. Then, to calculate the resistance change results of different pasting methods simultaneously, the end-pasting method and the overall-pasting method are respectively adopted at different positions on the same specimen; the end-pasting method is to paste across the crack, and the colloid only fixes both ends of the sensor; the overall-pasting method requires ensuring that the whole sensor is closely attached to the structural surface without air bubbles and the surface is flat. Moreover, a small section of sensor needs to be set in the non-crack area close to the sensor to calculate the crack width. Finally, to ensure the stability of the sensor pasting, use epoxy resin glue to paste the sensor on the surface of the cracked slab at least 30 minutes in advance, pay attention to controlling no air bubbles in the sensor, and keep the surface flat ( Figure 7 where 1, 2, and 3 are the end-pasted sensor, non-crack area sensor, and overall-pasted sensor respectively).

[0118] (3) Boundary condition setting: To facilitate observing the deformation of the sensor and the development of cracks, place the jack under the specimen, and use concrete blocks to act as simply supported conditions above the specimen, and the pressure press acts as the reaction frame.

[0119] (4) Loading process: For the concrete slab with cracks, after the sensors are pasted in advance, first, a concentrated load is applied at the bottom of the beam using a hydraulic jack, and the growth rate is controlled to increase in steps. The change law of the resistance at both ends of the sensor is read in real time through a digital milliohm meter. Secondly, after the concentrated load gradually increases with time until it stabilizes, the final value of another sensor is read and recorded. Finally, the corresponding crack width value is measured. Due to the insufficient accuracy of crack measurement, the accuracy of the algorithm is considered to be preliminarily verified, and the crack is developed to a larger width so that the error of the calculation result is smaller.

[0120] 2. Experimental results

[0121] As the load changes continuously, the crack width also develops continuously, and the resistance increases continuously. When the load is applied to 2240 N, the displacement of the concrete slab expands sharply, approaching failure, and the test is stopped. The changes in load and resistance are shown in Tables 2 and 3.

[0122] Table 2 Resistance-load change table for end-pasting method

[0123]

[0124] Table 3 Resistance-load change table for overall-pasting method

[0125]

[0126] After obtaining the load and resistance change data, the corresponding load-resistance change diagram is made using origin as shown in Figure 11 and Figure 12 shown.

[0127] It can be found through the experiment that in the initial stage of load action, due to the small change in crack width, the resistance shows a linear change law. When the load exceeds 2240 N, the resistance shows a non-linear change, which is consistent with the electrical characteristics of the sensor. By comparing the two methods, it can be found that the experimental effect of the end-pasting method is more stable.

[0128] After data processing, the change situation of the resistance change rate when the load changes can be obtained, and the fitting using origin is shown in Figure 13 and Figure 14 shown.

[0129] It can be obtained from the figure that in the initial stage of load action, the resistance change rates of the two pasting methods basically show a linear change. When the load reaches 2000 N, considering that the cracks do not show a linear development, the obtained resistance change rate also suddenly increases.

[0130] 3. Algorithm verification

[0131] Scheme design: Before the test, calibrate the sensitivity of the used sensors to obtain the sensor sensitivity K. After the test results are stable, first, measure the initial resistance R0 of the sensor when it is pasted, and measure the initial crack width l0. After the loading is completed, use tools to measure the width l2 of the crack after it has developed. The crack development width is the difference between the two measured values. To reduce the test error, continuous loading can be applied during the loading process to make the crack develop to a larger width, approaching the failure of the test block.

[0132] The test effect of the end-pasting method is better. Therefore, based on the linear sensor used in the end-pasting method, K = 0.3625 mm -1 .

[0133] Use the end-pasting method for calculation verification. From the formula, it can be seen that:

[0134]

[0135] The theoretical crack development width is obtained as 0.68 mm. For the overall pasting method, analyze using the measured resistance change rate:

[0136]

[0137] The resistance of the small sensor in the non-crack area satisfies:

[0138]

[0139] Substitute the initial crack width l0 (estimated to be 1 mm) into formula (5) for calculation, and the crack development width can be obtained as 0.44 mm.

[0140] Through qualitative calculation, comparing the two calculation methods, it can be found that the calculation method corresponding to the end-pasting method is relatively simple. The crack development widths at different locations are different, and the crack development width near the concentrated load is larger, which is consistent with the actual working conditions. Since the overall pasting method considers the strain effect in the non-crack area, additional sensors need to be installed to measure the strain in the non-crack area. At the same time, the resistance change basically shows a linear growth trend with the increase of the load, which is consistent with the model, verifying the accuracy of the model.

Claims

1. A preparation method of a silver-based flexible sensor for identifying existing cracks in bridge structures, characterized in that The method includes the following steps: Step 1: Design a bionic architecture of a flexible sensor based on the characteristics of the gradient elastic modulus of human skin. The bionic architecture includes a base layer, a sensing layer, and a surface layer. The base layer and the surface layer are TPU film layers, and the sensing layer includes one or more layers of conductive silver nanomaterial layers; Step 2: For the identification of existing cracks, drawing on the unique biomechanical properties and sensing capabilities of human skin, based on the bionic architecture designed in Step 1, use the dexterous gravity-driven phase separation strategy caused by the concentration gradient to modify the conductive silver nanomaterials on the surface of the thermoplastic polyurethane film to form a nanocomposite film flexible sensor with the ability to adhere to and adapt to the changes on the bridge surface; Step 3: Integrate a power module, a data acquisition and preprocessing module, and a data transmission module on the basis of the flexible sensor prepared in Step 2 to obtain a flexible multifunctional integrated sensor.

2. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 1, characterized in that The specific steps of Step 1 are as follows: Step 1-1: Based on the characteristics of the gradient elastic modulus, the skin can adapt to various deformations and accurately sense external stimuli. Simulate this skin characteristic and design a bionic architecture of a flexible sensor, which includes a base layer, a sensing layer, and a surface layer; Step 1-2: Select conductive silver nanomaterials to construct different sensing layers of the sensor.

3. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 1, characterized in that The specific steps of Step 2 are as follows: Step 2-1: Film pretreatment: Clean and pretreat the TPU film to ensure that its surface is flat and clean, which is beneficial to the attachment of AgNWs; Step 2-2: Prepare a series of AgNWs solutions with different concentrations; Step 2-3: Based on the bionic structure proposed in Step 1, with the identification of existing cracks as the guide, design the asymmetry of the sensor array. By controlling the concentration gradient of AgNWs in the solution, use gravity drive to achieve the phase separation and orderly arrangement of AgNWs on the surface of the TPU film, and obtain an AgNW@TPU nanocomposite film flexible strain sensor with the ability to adhere to and adapt to the changes on the bridge surface.

4. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 3, characterized in that The specific steps of Step 2-3 are as follows: Step 2-3-1: Impregnation process: Immerse the TPU film into the AgNWs solution; Step 2-3-2: Phase separation control: With the identification of existing cracks as the guide, precisely control the distribution and arrangement of AgNWs on the surface of the TPU film by controlling the impregnation speed, time, and temperature parameters; Step 2-3-3: Through drying and curing treatment, make AgNWs firmly embedded in the TPU film; Step 2-3-4: According to the gradient elastic modulus characteristics of the bionic architecture, combined with the possible directions and forms of cracks in practical applications, starting from the low-concentration AgNWs solution and gradually immersing it into the high-concentration AgNWs solution, repeat Steps 2-3-1 to 2-3-3 for multi-layer composite and surface treatment to construct an asymmetric sensing layer layout. Based on this, the AgNW@TPU nanocomposite film flexible sensor can make selective multifunctional responses to bending, compression, and tension to further improve the performance of the AgNW@TPU nanocomposite film.

5. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 4, characterized in that In Step 2-3-4, the specific design requirements for constructing an asymmetric sensing layer layout are as follows: (1) Design the directional arrangement of sensors: According to the possible directions of cracks, design a sensing layer with a directional arrangement to ensure that the sensor can produce a more obvious response under deformation or pressure in a specific direction; (2) Design the density difference of sensors: In different regions of the sensor, adjust the density of the sensing layer. In the areas prone to cracks, increase the density of the sensing layer to improve the sensitivity and accuracy of detection; (3) Design the multi-layer structure of sensors: Combining the gradient elastic modulus characteristics of the bionic architecture, design a sensor with a multi-layer structure. The sensing layers of different layers have different elastic moduli and sensitivities to meet the monitoring requirements under different deformation degrees.

6. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 1, wherein The specific steps of Step 3 are as follows: Step 3-1: According to the requirements of crack identification, design the functions and structures of the power supply module, data acquisition and preprocessing module, and data transmission module. Among them: The power supply module is responsible for providing stable electrical energy for the entire sensor; The data acquisition and preprocessing module is responsible for collecting the response signals of the flexible sensor, and performing filtering, amplification, existing crack identification, and crack development trend judgment; The data transmission module is responsible for wirelessly transmitting the processed data to the upper computer or cloud server; Design flexible connectors based on flexible materials to connect each module and ensure the stability and reliability of the sensor in a complex environment; Step 3-2: Assemble the power supply module, data acquisition and preprocessing module, data transmission module, and flexible strain sensor through flexible connectors to form a complete flexible multi-functional integrated sensor.

7. The method for preparing a silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 6, characterized in that In Step 3-1, the method for the data acquisition and preprocessing module to collect the response signals of the flexible sensor and perform filtering, amplification, existing crack identification, and crack development trend judgment is as follows: (1) Data collection: Collect time series data of crack development, which are from the response signals of the flexible sensor at different positions on the surface of the structure; (2) Data preprocessing: Clean the collected data to remove noise and outliers to ensure the accuracy and reliability of the data; Perform standardization processing on the data so that data from different sources and different scales can be compared and analyzed in the same framework; (3) Build a crack trend identification model based on RNN: The cyclic structure of RNN enables it to capture the time-dependent relationships in the sequence data and analyze the development and change of cracks over time; Use the preprocessed data to train the RNN model, and optimize the performance of the model by adjusting the model parameters and training strategies so that it can accurately identify the development trend of cracks; Initialize the hidden state, that is, set the initial hidden state as a vector of all zeros: h0＝0 Hidden state update formula: h t = tanh(W hh h t-1 + W hx x t + b h ) where h t represents the hidden state at the current moment; h t-1 represents the hidden state at the previous moment; x t represents the input at the current moment; W hh represents the weight matrix from the hidden state to the hidden state; W hx represents the weight matrix from the input to the hidden state; b h represents the hidden state offset; Output calculation formula: y t = σ(W yh h t + b y ) where y t represents the output at the current moment; W yh represents the hidden state at the previous moment; b y represents the input at the current moment; On this basis, establish a loss function model: Wherein, L represents a certain loss function, such as cross-entropy loss; represents the correct output; (4) On the basis of the RNN model, introduce the Attention attention mechanism. By calculating the attention weights of each position in the input sequence, help the model focus on the part most relevant to the crack development trend; During the training process, optimize the attention weights through the backpropagation algorithm so that the model can more accurately capture the key information in the input data; Propose the RNN+Attention hidden state calculation formula: h t = f(h t-1 , x t ) where h t represents the hidden state at the current moment; h t-1 represents the hidden state at the previous moment; x t represents the input at the current moment; f represents the transfer function of the RNN cell; Establish a calculation formula for attention scores: e t,i = g(h t , s i ) where, e t,i represents the attention score at the current moment; h t represents the hidden state at the current moment; s i represents the i-th hidden state in the encoding sequence; g represents the scoring function of the attention mechanism. Introduce an attention weight formula: where α t,i represents the attention weight of the i-th encoded hidden state at the current moment. Establish a calculation formula for context vectors: where c t represents the context vector at the current moment, which is the weighted sum of the respective hidden states in the encoding sequence; Then construct a calculation formula for the decoder hidden state: s t = f(s t-1 , y t-1 , c t ) where s t represents the decoder hidden state at the current moment; s t-1 represents the decoder hidden state at the previous moment; y t-1 represents the output at the previous moment; c t represents the context vector at the current moment; f represents the transition function of the decoder; Use the RNN and Attention mechanisms to extract features from the flexible sensor response signals, fuse the extracted features to form a comprehensive feature vector for characterizing the development trend of cracks; (5) Construct a fuzzy logic system, define fuzzy sets, fuzzy rules, and a fuzzy inference mechanism, transform the problem of identifying the development trend of cracks into a fuzzy logic problem, use the simulated annealing algorithm to optimize the parameters of the fuzzy logic system, and apply the optimized fuzzy logic system to the identification of crack development changes to output the identification results; (6) Use an independent test data set to evaluate the constructed fuzzy logic system, and improve the model according to the evaluation results to improve the performance and accuracy of the model.

8. The preparation method of the silver-based flexible sensor for identifying existing cracks in bridge structures according to claim 7, characterized in that The specific steps of defining fuzzy sets, fuzzy rules, and a fuzzy inference mechanism include: defining fuzzy variables and their membership functions, establishing fuzzy rules, selecting the Mamdani inference mechanism, and converting the fuzzy output into a clear result using the centroid method and the maximum membership degree method to identify the development trend of cracks; using the simulated annealing algorithm to avoid falling into local optimal solutions during the search process and improve the accuracy of crack development change identification.

9. A method for identifying existing cracks in a bridge structure using a flexible strain sensor prepared by the method according to any one of claims 1-8, characterized in that The method includes the following steps: Step 1: Install a flexible multi-functional integrated sensor on the surface of the structure to be monitored. The flexible strain sensor simulates the reaction of the human skin when subjected to external pressure or deformation, and identifies the presence and development trend of cracks through changes in the flexible sensor response signals; Step 2: The sensor wirelessly transmits the collected data to the host computer or cloud server for further noise reduction and data mining processing; Step 3: According to the processing results, judge the severity and development trend of the cracks to provide a scientific basis for the maintenance and repair of the bridge structure.

10. The method for identifying existing cracks in a bridge structure using a flexible strain sensor according to claim 9, characterized in that The specific method of Step 1 is as follows: Step 1-1: Accurately capture and measure micro to macroscopic crack deformations through changes in the electrical signal waveform. On this basis, judge the rate, direction, and frequency parameters of the deformation through changes in the electrical signal waveform to achieve the purpose of real-time and continuous monitoring of existing crack changes in the bridge structure, and use the flexible strain sensor to detect the presence and expansion of cracks; Step 1-2: When cracks appear, different elastic modulus layers in the flexible strain sensor will be subjected to different degrees of stress concentration, resulting in changes in the electrical parameters inside the sensor. By monitoring the changes in the electrical parameters, accurately identify the presence and expansion of cracks.

Images