Dynamic monitoring device and method for scouring interface of offshore wind power pile foundation

Abstract

The invention discloses a dynamic monitoring device and method for a scouring interface of an offshore wind power pile foundation. The monitoring device comprises an internal heating temperature measurement optical cable, a heating unit, an optical fiber temperature measurement unit, a monitoring error correction unit, a heat conductivity coefficient calculation unit and a scouring area detection unit. During monitoring, the optical fiber temperature measurement unit transmits acquired heating temperature data to the monitoring error correction unit; the monitoring error correction unit adopts a mixed attention neural network algorithm to realize monitoring data error correction and then transmits the data to the heat conductivity coefficient calculation unit; the heat conductivity coefficient calculation unit calculates the heat conductivity coefficient of each medium in the length direction of the pile foundation by using the corrected temperature data, and transmits the heat conductivity coefficient to the scour area detection unit; the scouring area detection unit monitors the length and the state of a scouring area by calculating the heat conductivity coefficient gradient. According to the monitoring device and method, the defects of a traditional monitoring means are overcome, and the scouring state of the offshore wind power pile foundation is accurately and dynamically fed back.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind power pile foundation scour monitoring, and in particular to an offshore wind power pile foundation scour interface dynamic monitoring device and method. Background Art

[0002] Offshore wind power pile foundations are widely used in offshore engineering fields such as offshore platforms, docks, cross-sea bridges and offshore wind power, and are one of the important engineering structures in offshore engineering. However, since the seabed in offshore areas is mostly loose sedimentary soils such as silt, silt or fine sand, these special geological conditions make the seabed around the pile foundations prone to scouring and erosion under the action of waves and currents. Scouring will not only weaken the bearing capacity of the soil around the pile foundation, but may also cause the pile foundation to be exposed, tilted or even unstable, seriously threatening the long-term safety of offshore engineering facilities. Therefore, in order to ensure the safe operation of offshore engineering facilities, it is of great significance to conduct long-term and effective monitoring of the scouring conditions around offshore wind power pile foundations.

[0003] There are various types of underwater pile foundation scour monitoring technologies. Currently, the commonly used monitoring technologies include sonar, sliding magnetic ring, ground penetrating radar and fiber Bragg grating (FBG) sensors.

[0004] Sonar technology measures the shape and depth of seabed scour pits around pile foundations by emitting sound waves and receiving their reflected waves. When sound waves encounter a medium interface with different densities (such as the boundary between water and the seabed), reflections will occur. By measuring the return time and intensity of the sound waves, detailed information about the seabed can be obtained. However, the resolution of sonar is limited by the wavelength of the sound waves. In complex marine environments, multipath propagation may cause data distortion. In addition, the technology has poor real-time performance and is difficult to provide continuous dynamic monitoring of scours.

[0005] The sliding magnetic ring monitoring technology is based on the principle of electromagnetic induction. It senses the changes in the electromagnetic field in the water flow, monitors the speed and direction of the water flow in real time, and indirectly infers the movement of the soil. This method is suitable for monitoring the movement of soil during scouring. However, the sliding magnetic ring is very sensitive to soil type and water flow speed, which can easily lead to measurement errors. In addition, the installation and maintenance are relatively complicated, and the reliability and stability of the equipment underwater need to be ensured.

[0006] The working principle of ground penetrating radar is similar to that of sonar. It emits high-frequency electromagnetic waves to the seabed and uses the reflected echoes to obtain underground structure information. Its advantage is that it can provide high-resolution images and is suitable for various types of marine sediments. However, in seawater and saturated soil, the penetration ability of ground penetrating radar will be limited, resulting in severe signal attenuation. In addition, weather and seasonal changes will also affect the detection effect, resulting in reduced data reliability.

[0007] Fiber Bragg grating sensors use the propagation characteristics of light in optical fibers to obtain strain or temperature information along the optical fiber by measuring the wavelength change of the light reflected by the Bragg grating. This technology is highly sensitive and resistant to electromagnetic interference, but its monitoring is point-based and cannot achieve distributed monitoring.

[0008] In summary, although the existing technologies can monitor the scouring conditions of offshore wind turbine pile foundations to a certain extent, they are unable to achieve the effect of real-time, fully distributed monitoring. Therefore, how to achieve fully distributed real-time monitoring of offshore wind turbine pile foundations has become a technical problem that needs to be solved urgently. Summary of the invention

[0009] Purpose of the invention: In view of the technical problems existing in the prior art, the present invention proposes a dynamic monitoring device and method for the scour interface of an offshore wind power pile foundation. The monitoring error correction unit in the present invention performs error correction on the temperature data through a hybrid attention neural network algorithm to obtain a corrected temperature change rate; the thermal conductivity calculation unit calculates the thermal conductivity of each medium in the length direction of the pile foundation according to the corrected temperature change rate; the scour area detection unit obtains a thermal conductivity gradient curve according to the thermal conductivity, determines the scour area according to the thermal conductivity gradient curve, and judges the scour state of the pile foundation at the same time, thereby improving the accuracy of scour detection.

[0010] Technical solution: The dynamic monitoring device for the scour interface of an offshore wind power pile foundation of the present invention comprises an internal heating temperature measuring optical cable, a heating unit, an optical fiber temperature measuring unit, a monitoring error correction unit, a thermal conductivity calculation unit and an scour area detection unit; the internal heating temperature measuring optical cable is composed of an internal heating resistance wire and a temperature sensing optical fiber; the internal heating temperature measuring optical cable is fixed on the offshore wind power pile foundation; the internal heating resistance wire is connected to the heating unit, and the temperature sensing optical fiber is connected to the optical fiber temperature measuring unit; the optical fiber temperature measuring unit collects the temperature data of the internal heating temperature measuring optical cable during the heating process and transmits it to the monitoring error correction unit.

[0011] The method for dynamically monitoring the scouring interface of offshore wind power pile foundation of the present invention comprises the following steps:

[0012] (1) After the offshore wind power pile foundation is manufactured, the internal heating temperature measuring optical cable is fixed on the offshore wind power pile foundation; after the offshore wind power pile foundation is constructed, the internal heating resistance wire in the internal heating temperature measuring optical cable is connected to the heating unit, and waterproof protection is performed at the same time, and the temperature sensing optical fiber in the internal heating temperature measuring optical cable is connected to the optical fiber temperature measuring unit through an optical fiber jumper;

[0013] (2) The heating unit is powered on, and the optical fiber temperature measurement unit collects temperature data of the temperature measurement optical cable during the heating period;

[0014] (3) The optical fiber temperature measurement unit transmits the temperature data to the monitoring error correction unit, which uses a hybrid attention neural network algorithm to perform error correction on the temperature data to obtain a corrected temperature change rate; and transmits the corrected temperature change rate to the thermal conductivity calculation unit, which calculates the thermal conductivity in the length direction of the pile foundation according to the corrected temperature change rate; the process is as follows:

[0015] (3.1) The temperature change rate of medium i between heating time period t1 and t2 is measured and calculated by the optical fiber temperature measurement unit.

[0016]

[0017] Among them, medium i includes seawater, seabed sediments and the interface between seawater and seabed sediments, T is the temperature corresponding to two moments, T i (t1) is the temperature of the line source in medium i at time t1 measured by the optical fiber temperature measurement unit, T i (t2) is the temperature of the line source in medium i measured by the optical fiber temperature measurement unit at time t2;

[0018] Under laboratory conditions, high-precision temperature measuring equipment is used to measure the temperature change rate of medium i during the same heating period.

[0019]

[0020] The resulting As the target value of the supervised neural network; T i '(t1) is the temperature of the line source in medium i at time t1 measured in the laboratory, T i '(t2) is the temperature corresponding to the line source in medium i at time t2 measured in the laboratory;

[0021] Finally, the corrected temperature change rate K is obtained using the following error correction formula:

[0022]

[0023] in, Error correction for the temperature rate of change predicted by the hybrid attention neural network.

[0024] Using the hybrid attention neural network algorithm to obtain the temperature change rate error correction value The process is as follows: the temperature change rate in each medium The temperature change rate along the monitored optical fiber is used as the input temperature change rate feature matrix X of the hybrid attention neural network, and the temperature change rate feature map F of the input feature is extracted through a lightweight one-dimensional convolution layer c:

[0025] F c =σ(W*X+b)

[0026] Among them, W is the convolution kernel weight, b is the bias term, σ is the activation function; X is the input temperature change rate feature matrix of the neural network.

[0027] (3.2) Temperature change rate characteristic diagram F c Perform channel-by-channel maximum pooling and average pooling operations to obtain the maximum pooling result F max And the average pooling result F avg :

[0028]

[0029] Where c is the channel index, F c (X,c) represents the eigenvalue on channel c after the input temperature change rate feature matrix X is processed by convolution and activation function; C is the total number of channels.

[0030] Based on the global pooling channel attention mechanism, calculate the temperature change rate feature map F c The weight matrix A c :

[0031] A c =σ(FC(GAP(F c )))

[0032] Among them, GAP represents the c The average value of each channel of F is calculated to generate a feature vector of a specific length. C It means that the feature vector is weighted learned through the fully connected layer to generate the weight of each channel.

[0033] Dynamically adjust each channel to obtain the enhanced temperature change rate characteristic graph F e :

[0034] F e =A c ⊙F c

[0035] Among them, ⊙ represents the channel-by-channel point product.

[0036] (3.3) Using the spatial attention mechanism to capture the enhanced temperature change rate feature map F e For correlation between different positions, the process is as follows:

[0037] First, the spatial weight A is generated by multi-scale convolution s :

[0038] A s=σ(Conv2D(Concat(F max , F avg ))) Among them, Concat means to convert F max and F avg Splicing in the channel dimension generates a joint temperature change rate feature map; Conv2D means learning the spatial attention weight from the spliced ​​joint temperature change rate feature map through two-dimensional convolution to extract significant features and global distribution features.

[0039] Then, the optimized enhanced temperature change rate characteristic diagram is F final :

[0040] F final =A s ⊙F e

[0041] Finally, the optimized temperature change rate characteristic graph F is calculated by quadratic weighted integration. final Correction is performed to obtain the temperature change rate error correction value of medium i during the heating period

[0042]

[0043] Among them, W corr is the modified weight matrix, and Ω is the calculation area.

[0044] The loss function L is used to optimize the corrected weight matrix Wcorr and temperature change rate error correction value of the hybrid attention neural network The loss function is:

[0045]

[0046] Among them, w i is the weight for medium i, λ is the parameter of regularization strength, W corr is the modified weight matrix, λ||W corr || 2 is the regularization term. The loss function L updates the corrected weight matrix Wcorr at each iteration. When the loss function L reaches the minimum value, the final corrected temperature change rate K is output.

[0047] The thermal conductivity calculation unit calculates the medium thermal conductivity λ at the pile length l according to the corrected temperature change rate. l ;

[0048]

[0049] Among them, λ lis the thermal conductivity of the pile foundation at length l, in W / (m·K); q is the heating power of the linear heat source, in W / m; K is the temperature change rate at the pile foundation length l after correction by the hybrid attention neural network, in K.

[0050] (4) The scour area detection unit calculates the thermal conductivity gradient g in the length direction of the pile foundation based on the thermal conductivity of the medium. l , determine the scour area and detect the scour state based on the thermal conductivity gradient; thermal conductivity gradient g l for:

[0051]

[0052] Among them, λ l is the thermal conductivity at length l, λx is the thermal conductivity at length x adjacent to length l, X is the distance between the two measuring points, the value of X is L / S, where L is the total length of the monitoring section, and S is the sampling resolution.

[0053] The scour area detection unit detects the position difference of the thermal conductivity gradient mutation point between the thermal conductivity gradient curve and the reference curve when the pile foundation is installed, and obtains the scour area length d, which is as follows:

[0054] d=L2-L1, where L2 is the position of the mutation point in the thermal conductivity gradient curve, and L1 is the position of the mutation point in the reference curve.

[0055] Comparing the length of the scour area d with the original depth of the foundation D, the instability coefficient F is obtained:

[0056]

[0057] When the instability coefficient F is greater than the safety threshold f in the specification, the pile foundation scour state is unstable and the superstructure is at risk of collapse; when the instability coefficient F is less than the safety threshold f in the specification, the pile foundation scour state is stable and the superstructure is safe.

[0058] In step (3.3), the optimized temperature change rate characteristic graph F is obtained by quadratic weighted integration. final Correction is performed to obtain the temperature change rate error correction value of medium i during the heating period

[0059] In step (3.3), the loss function L is used to optimize the modified weight matrix Wcorr of the hybrid attention neural network and the temperature change rate error correction value The loss function is:

[0060]

[0061] Among them, w iis the weight for medium i, λ is the parameter of regularization strength, W corr is the modified weight matrix, λ||W corr || 2 is the regularization term; the loss function L updates the corrected weight matrix Wcorr at each iteration, and when the loss function L is minimized, the corrected temperature change rate K is output.

[0062] In step (1), the internally heated temperature measuring optical cable is fixed on the offshore wind power pile foundation by using waterproof glue.

[0063] In step (1), the optical cable needs to be protected with high-strength epoxy resin after being fixed.

[0064] In step (1), in order to ensure data quality, a section of internal heating temperature measurement optical cable is reserved at the bottom of the pile foundation.

[0065] In step (3), the heating power and heating time are set according to the designed burial depth of the pile foundation and the local average depth of seawater, and the resolution S and sampling frequency of the optical fiber temperature measurement unit when collecting data are determined according to the length of the pile foundation.

[0066] In step (4), in the process of monitoring error correction, in order to improve the calculation accuracy of the temperature change rate, the hybrid attention neural network is trained based on laboratory calibration data.

[0067] Working principle: The internal heating temperature measuring optical cable of the present invention is composed of an internal heating resistance wire and a temperature sensing optical fiber. When the offshore wind power pile foundation is produced, the internal heating temperature measuring optical cable is fixed on the pile foundation, and then the tail end of the internal heating resistance wire of the internal heating temperature measuring optical cable is welded together to form a passage. The head end of the internal heating resistance wire is connected to the heating unit; after the construction of the offshore wind power pile foundation is completed, the internal heating temperature measuring optical cable is connected to the optical fiber temperature demodulator; the heating unit heats for a certain period of time according to the set power, and the optical fiber temperature demodulator collects the temperature data within this time period and transmits it to the monitoring error correction unit; the monitoring error correction unit uses the hybrid attention neural network algorithm to perform error correction on the temperature data to obtain the corrected temperature change rate; the thermal conductivity calculation unit calculates the thermal conductivity of each medium in the length direction of the pile foundation according to the corrected temperature change rate; the scouring area detection unit obtains the thermal conductivity gradient curve according to the thermal conductivity, and determines the scouring area according to the thermal conductivity gradient curve, and judges the scouring state of the pile foundation at the same time.

[0068] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0069] (1) The monitoring method adopted by the present invention realizes an efficient heating process through self-heating technology. During the heating process, the heat loss is extremely low, which ensures the effectiveness of heating and reduces the influence of the external environment on the heating effect. At the same time, the present invention combines optical frequency domain reflectometry (OFDR) technology to improve the temperature measurement accuracy to 0.01°C, further improve the measurement accuracy of thermal conductivity, and provide reliable data support for determining the scour area, thereby improving the accuracy and reliability of scour detection.

[0070] (2) The present invention uses a hybrid attention neural network algorithm to correct the monitoring data error, further improving the accuracy of thermal conductivity calculation. The algorithm extracts key features from the monitoring data, eliminates noise and outliers, and ensures the reliability of the calculation results. The application of this technology provides strong support for the accurate detection of the scour area, further ensuring the stability and practicality of the monitoring system.

[0071] (3) The monitoring method of the present invention is implemented by optical fiber sensing technology, which has extremely strong anti-interference ability, especially in areas with complex electromagnetic environments, and will not be affected by electromagnetic interference, thus ensuring the stability of the monitoring data. In addition, through the application of optical fiber sensors, all-weather foundation scour detection is achieved, and the condition of offshore wind turbine pile foundations is continuously detected even under severe weather conditions. This all-weather real-time monitoring technology makes up for the shortcomings of traditional monitoring methods in terms of real-time and environmental adaptability, and enables more accurate dynamic feedback on the scour condition of offshore wind turbine pile foundations.

[0072] (4) The sensors used in the present invention have a high survival rate and can withstand the harsh conditions in the marine environment to ensure long-term stable operation. The design of the sensor enables it to perform distributed measurements along the length of the pile foundation, and can accurately capture the scour conditions at various locations of the pile foundation, and is no longer limited to monitoring a single measuring point. This distributed monitoring mode improves the comprehensiveness of the data and facilitates understanding of the scour distribution around the pile foundation. The sensor is easy to install and does not require complicated construction steps, and is suitable for monitoring needs of various types of offshore wind power pile foundations. In particular, after the construction of the offshore wind power pile foundation is completed, the monitoring method of the present invention can be put into use immediately for long-term monitoring of local scour areas of the pile foundation. This long-term monitoring capability not only provides continuous data support for the safe maintenance of offshore wind power pile foundations, but also indirectly reduces subsequent engineering costs, with significant economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a structural schematic diagram of a dynamic monitoring device for the scour interface of an offshore wind power pile foundation according to the present invention;

[0074] Figure 2 is the thermal conductivity curve in the length direction of the pile foundation in the present invention;

[0075] Figure 3 It is the thermal conductivity gradient curve in the present invention. DETAILED DESCRIPTION

[0076] like Figure 1 As shown, the offshore wind power pile foundation scour interface dynamic monitoring device of the present invention comprises an internal heating temperature measuring optical cable 1, a heating unit 2, an optical fiber temperature measuring unit 3, a monitoring error correction unit 4, a thermal conductivity calculation unit 5 and a scour area detection unit 6. The internal heating temperature measuring optical cable 1 is composed of two internal heating resistance wires 1-1 and a temperature sensing optical fiber 1-2.

[0077] The internal heating temperature measuring optical cable 1 is fixed on the offshore wind power pile foundation. In this embodiment, the internal heating temperature measuring optical cable 1 is pasted on the surface of the offshore wind power pile foundation. The internal heating resistance wire 1-1 in the internal heating temperature measuring optical cable 1 is connected to the heating unit 2, and the temperature sensing optical fiber 1-2 is connected to the optical fiber temperature measuring unit 3; the optical fiber temperature measuring unit 3 collects the temperature data of the internal heating temperature measuring optical cable 1 during the heating process and transmits it to the monitoring error correction unit 4; the monitoring error correction unit 4 performs error correction on the temperature data to obtain the temperature change rate, and then transmits it to the thermal conductivity calculation unit 5; the thermal conductivity calculation unit 5 calculates the medium thermal conductivity in the length direction of the pile foundation, and transmits it to the scouring area detection unit 6; the scouring area detection unit 6 obtains the thermal conductivity gradient curve according to the thermal conductivity, and then determines the scouring area and detects the scouring state.

[0078] The method for dynamically monitoring the scouring interface of offshore wind power pile foundation of the present invention comprises the following steps:

[0079] (1) The internal heating temperature measuring optical cable 1 is glued and fixed on the offshore wind power pile foundation; the two internal heating resistance wires 1-1 in the internal heating temperature measuring optical cable 1 at the tail of the pile foundation are welded together; the internal heating resistance wire 1-1 in the internal heating temperature measuring optical cable 1 at the top of the pile foundation is connected to the heating unit 2.

[0080] (2) After the pile foundation construction is completed, the temperature sensing optical fiber 1-2 is connected to the optical fiber temperature measurement unit 3 through an optical fiber jumper; the heating unit 2 is powered on, and the optical fiber temperature measurement unit 3 collects temperature data in the length direction of the pile foundation during the heating time.

[0081] (3) The temperature data collected by the optical fiber temperature measurement unit 3 is transmitted to the monitoring error correction unit 4, and the monitoring error correction unit 4 uses a hybrid attention neural network algorithm to perform error correction on the temperature data to obtain a corrected temperature change rate; the thermal conductivity calculation unit 5 calculates the thermal conductivity of the medium at each position along the length direction of the pile foundation by using the corrected temperature change rate;

[0082] (4) The scour area detection unit 6 calculates the thermal conductivity gradient in the length direction of the pile foundation according to the medium thermal conductivity, and determines the length of the scour area and the instability coefficient according to the result, thereby determining the scour state.

[0083] Among them, in step (1), the internal heating temperature measuring optical cable 1 is fixed on the offshore wind power pile foundation by using waterproof quick-drying glue; secondly, the optical cable is protected by using high-strength epoxy resin.

[0084] In step (1), in order to ensure data quality, a section of internal heating temperature measurement optical cable is reserved at the bottom of the pile foundation. The internal heating resistance wire 1-1 is welded and waterproofed.

[0085] In step (2), the heating time and heating power of the heating unit 2 are set according to the designed buried depth of the pile foundation and the local average depth of seawater.

[0086] In step (2), the sampling resolution S and acquisition frequency of the optical fiber temperature measurement unit 3 are determined according to the total length of the pile foundation.

[0087] In step (3), in the process of monitoring error correction, in order to improve the calculation accuracy of the temperature change rate, the hybrid attention neural network is trained based on laboratory calibration data:

[0088] (3.1) First, the temperature change rate of medium i between the heating time period t1 and t2 is measured and calculated by the optical fiber temperature measurement unit.

[0089]

[0090] Wherein, medium i includes seawater, seabed sediments and the interface between seawater and seabed sediments, and T is the temperature corresponding to two moments. i (t1) is the temperature of the line source in medium i measured by the optical fiber temperature measurement unit 3 at time t1, T i (t2) is the temperature of the line source in medium i measured by the optical fiber temperature measurement unit at time t2;

[0091] Under laboratory conditions, use high-precision temperature measurement equipment to measure the temperature change rate of medium i during the same heating period to obtain accurate values The formula is as follows:

[0092]

[0093] The resulting As the target value of the supervised neural network. i '(t1) is the temperature of the line source in medium i at time t1 measured in the laboratory, T i '(t2) is the temperature corresponding to the line source in medium i at time t2 measured in the laboratory;

[0094] Finally, the corrected temperature change rate K is obtained using the following error correction formula:

[0095]

[0096] in, Error correction value predicted by hybrid attention neural network.

[0097] Error correction value is obtained using hybrid attention neural network algorithm The process is as follows: the temperature change rate in each medium The temperature change rate along the monitored optical fiber is used as the input temperature change rate feature matrix X of the hybrid attention neural network, and the temperature change rate feature map F of the input feature is extracted through a lightweight one-dimensional convolution layer c :

[0098] F c =σ(W*X+b)

[0099] Among them, W is the convolution kernel weight, b is the bias term, σ is the activation function; X is the input temperature change rate feature matrix of the neural network.

[0100] (3.2) Then the temperature change rate characteristic diagram F c Perform channel-by-channel maximum pooling and average pooling operations to obtain the maximum pooling result F max And the average pooling result F avg :

[0101]

[0102] Where c is the channel index, F c (X,c) represents the eigenvalue on channel c after the input temperature change rate feature matrix X is processed by convolution and activation function, and C is the total number of channels.

[0103] Calculate the temperature change rate characteristic graph F c The weight matrix A c :

[0104] A c =σ(FC(GAP(F c )))

[0105] Among them, GAP represents the c The average value of each channel of F is calculated to generate a feature vector of a specific length; C It means that the feature vector is weighted learned through the fully connected layer to generate the weight of each channel.

[0106] Dynamically adjust each channel to obtain the enhanced temperature change rate characteristic graph Fe :

[0107] F e =A c ⊙F c Where ⊙ represents the channel-wise dot product.

[0108] (3.3) Using the spatial attention mechanism to capture the enhanced temperature change rate feature map F e For correlation between different positions, the process is as follows:

[0109] First, the spatial weight A is generated by multi-scale convolution s :

[0110] A s =σ(Conv2D(Concat(F max , F avg )))

[0111] Concat means to convert F max and F avg The joint temperature change rate feature map is generated by splicing in the channel dimension. Conv2D means learning the spatial attention weight from the spliced ​​joint temperature change rate feature map through two-dimensional convolution to extract significant features and global distribution features.

[0112] Then, the optimized enhanced temperature change rate characteristic diagram is F final :

[0113] F final =A s ⊙F e

[0114] Finally, the optimized temperature change rate characteristic graph F is calculated by quadratic weighted integration. final Correction is performed to obtain the temperature change rate error correction value of medium i during the heating period

[0115]

[0116] Among them, W is the modified weight matrix and Ω is the calculation area.

[0117] The loss function L is used to optimize the corrected weight matrix Wcorr and temperature change rate error correction value of the hybrid attention neural network The loss function is:

[0118]

[0119] Among them, w i is the weight for medium i, λ is the parameter of regularization strength, W corr is the modified weight matrix, λ||Wcorr || 2 is the regularization term; the loss function L updates the corrected weight matrix Wcorr at each iteration, and when the loss function L is minimized, the corrected temperature change rate K is output.

[0120] In step (3), the thermal conductivity of the medium at the length l of the pile foundation is l Calculated by the following formula:

[0121]

[0122] Among them, λ l is the thermal conductivity of the pile foundation at the length l, in W / (m·K), q is the heating power of the linear heat source, in W / m, and K is the temperature change rate at the length l of the pile foundation after correction by the hybrid attention neural network.

[0123] In step (4), the process of the scouring area detection unit 6 detecting the scouring area according to the medium thermal conductivity gradient is as follows:

[0124] The present invention adopts a linear heat source heating method to measure the thermal conductivity in the length direction of the offshore wind power pile foundation. Because different types of media have different thermal conductivities due to different heat absorption capacities, the thermal conductivity in the entire length direction of the pile foundation is derived to obtain the thermal conductivity change rate, i.e., the thermal conductivity gradient. The thermal conductivity gradient will suddenly change at the place where the thermal conductivity changes (i.e., the junction of the seabed and the seawater). Therefore, the thermal conductivity gradient curve when the pile foundation is just installed is used as a reference, and the scouring area detection unit 6 compares the thermal conductivity gradient curve obtained during the monitoring process with the reference curve, and obtains the range of the scouring area based on the position difference of the mutation point between the two.

[0125] In step (4), the flushing area detection unit 6 calculates the thermal conductivity gradient g using the thermal conductivity obtained in step (3). l , the formula is as follows:

[0126]

[0127] Among them, λ l is the thermal conductivity at length l, λx is the thermal conductivity at length x adjacent to length l, X is the distance between the two measuring points, the value of X is L / S, where L is the total length of the monitoring section, and S is the sampling resolution.

[0128] In step (4), the scour area detection unit 6 detects the position difference of the thermal conductivity gradient mutation point between the thermal conductivity gradient curve and the reference curve to obtain the scour area length d, which is as follows:

[0129] d=L2-L1

[0130] Among them, L2 is the position of the mutation point in the thermal conductivity gradient curve, and L1 is the position of the mutation point in the reference curve.

[0131] Comparing the length of the scour area d with the original depth of the foundation D, the instability coefficient F is obtained:

[0132]

[0133] Combined with the safety threshold f in the specification, when the instability coefficient F is greater than f, the pile foundation scour state is unstable and the superstructure is at risk of collapse; conversely, the pile foundation scour state is stable and the superstructure is safe.

[0134] Example

[0135] The scouring condition of a certain offshore wind power pile foundation is monitored by using the dynamic monitoring device and method for the scouring interface of the offshore wind power pile foundation of the present invention. The offshore wind turbine has a capacity of 6MW, adopts a single pile foundation, the pile is made of steel, the pile foundation diameter is 5m, the total length of the pile foundation is 50m, the foundation burial depth is 25m, the average water depth of the detection site is 20m, and the marine sediment type is fine sand with a particle size range of 0.063mm-0.1mm.

[0136] The method for dynamically monitoring the scour interface of an offshore wind power pile foundation of the present invention comprises the following steps:

[0137] (1) After the pile foundation is completed, two 50m long internal heating and temperature measurement optical cables 1 are symmetrically pasted on both sides of the pile foundation using waterproof quick-drying glue;

[0138] (2) The tail end of the internal heating temperature measuring optical cable 1 is cut off, and the two internal heating resistance wires 1-1 are welded together by electric welding, and then tightly wrapped with waterproof insulating tape;

[0139] (3) After the pile foundation construction is completed, the internal heating resistance wire 1-1 in the internal heating temperature measuring optical cable at the top of the pile foundation is connected to the heating unit 2, and the temperature sensing optical fiber 1-2 in the internal heating temperature measuring optical cable 1 is connected to the optical fiber temperature measuring unit 3 through an optical fiber jumper;

[0140] (4) The heating power q of the heating unit 2 is adjusted to 80 W, the heating time is set to 1 minute, and then the power is turned on. The optical fiber temperature measurement unit 3 uses a resolution of 1 cm and a collection frequency of 10 Hz to collect temperature data during the heating time, and the data is used as a reference;

[0141] (5) After the pile foundation is put into use, the local scouring status of the pile foundation is monitored regularly every day. The heating power of the heating unit 2 is adjusted to 80W, the heating time is set to 1 minute, and then the power is turned on. The optical fiber temperature measurement unit 3 uses a resolution of 1 cm to collect temperature data during the heating time.

[0142] (6) The temperature data collected by the optical fiber temperature measurement unit 3 is transmitted to the monitoring error correction unit 4, and the monitoring error correction unit 4 uses a hybrid attention neural network algorithm to perform error correction on the monitoring data to obtain the temperature change rate K at each position after correction;

[0143] (4) The monitoring error correction unit 4 transmits the corrected temperature data to the thermal conductivity calculation unit 5. The thermal conductivity calculation unit 5 calculates the medium thermal conductivity at the pile foundation length l according to the following formula (the result is as follows Figure 2 ):

[0144]

[0145] Among them, λ l is the thermal conductivity of the pile foundation at length l, in W / (m·K), q is the heating power of the linear heat source, in W / m; K is the temperature change rate at the pile foundation length l after correction by the hybrid attention neural network, in K;

[0146] (8) The scour area detection unit 6 calculates the thermal conductivity gradient according to the following formula (the result is as follows Figure 3 ):

[0147]

[0148] Among them, λ l is the thermal conductivity at length l, λx is the thermal conductivity at length x adjacent to length l, X is the distance between the two measuring points, and its value is L / S, where L is the total length of the monitoring section and S is the sampling resolution.

[0149] The results show that one year after the pile foundation was put into use, the foundation scour depth d was 7.5m and the instability coefficient F was 0.3, which is close to the instability coefficient of 0.4 specified in the specification, and anti-scour measures need to be taken.

Claims

1. A dynamic monitoring device for the scouring interface of offshore wind power pile foundation, characterized in that: The invention comprises an internal heating temperature measuring optical cable (1), a heating unit (2), an optical fiber temperature measuring unit (3), a monitoring error correction unit (4), a thermal conductivity calculation unit (5) and a flushing area detection unit (6); the internal heating temperature measuring optical cable (1) comprises an internal heating resistance wire (1-1) and a temperature sensing optical fiber (1-2); the internal heating temperature measuring optical cable (1) is fixed on an offshore wind power pile foundation; the internal heating resistance wire (1-1) is connected to the heating unit (2), and the temperature sensing optical fiber (1-2) is connected to the optical fiber temperature measuring unit (3); the optical fiber temperature measuring unit (3) collects temperature data of the internal heating temperature measuring optical cable (1) during the heating process and transmits the data to the monitoring error correction unit (4).

2. A method for dynamic monitoring of the scouring interface of offshore wind power pile foundation, characterized in that: The method is implemented by the offshore wind power pile foundation scour interface dynamic monitoring device according to claim 1, and comprises the following steps: (1) fixing the internal heating temperature measurement optical cable (1) on the offshore wind power pile foundation; connecting the internal heating resistance wire (1-1) to the heating unit (2), and connecting the temperature sensing optical fiber (1-2) to the optical fiber temperature measurement unit (3); (2) The heating unit (2) is powered on, and the optical fiber temperature measuring unit (3) collects temperature data of the temperature measuring optical cable during the heating time; (3) The optical fiber temperature measurement unit (3) transmits the temperature data to the monitoring error correction unit (4); the monitoring error correction unit (4) uses a hybrid attention neural network algorithm to perform error correction on the temperature data, and transmits the corrected temperature change rate to the thermal conductivity calculation unit (5); the thermal conductivity calculation unit (5) calculates the thermal conductivity in the length direction of the pile foundation according to the corrected temperature change rate; the process is: (3.1) The temperature change rate K of medium i between heating time period t1 and t2 is measured and calculated by the optical fiber temperature measurement unit (3). Ti : Among them, medium i includes seawater, seabed sediments and the interface between seawater and seabed sediments, T i (t1) is the temperature of the line source in medium i at time t1 measured by the optical fiber temperature measurement unit, T i (t2) is the temperature corresponding to the line source in medium i at time t2 measured by the optical fiber temperature measurement unit; The temperature change rate of medium i in the same heating period is measured in the laboratory. The resulting As the target value of the supervised neural network; T i '(t1) is the temperature of the line source in medium i at time t1 measured in the laboratory, T i '(t2) is the temperature corresponding to the line source in medium i at time t2 measured in the laboratory; The corrected temperature change rate K is obtained using the following error correction formula: in, The error correction value of the temperature change rate predicted by the hybrid attention neural network; Calculate the temperature change rate error correction value The process is as follows: the temperature change rate in each medium The temperature change rate along the monitored optical fiber is used as the input temperature change rate feature matrix X of the hybrid attention neural network, and the temperature change rate feature map F of the input feature is extracted. c : F c =σ(W*X+b) Among them, W is the convolution kernel weight, b is the bias term, σ is the activation function; X is the input temperature change rate feature matrix of the neural network; (3.2) Temperature change rate characteristic diagram F c Perform channel-by-channel maximum pooling and average pooling operations to obtain the maximum pooling result F max And the average pooling result F avg : Where c is the channel index, F c (X, c) represents the eigenvalue on channel c after the input temperature change rate feature matrix X is processed by convolution and activation function; C is the total number of channels; Calculate the temperature change rate characteristic graph F c The weight matrix A c : A c =σ(FC(GAP(F c ))) Among them, GAP represents the c The average value of each channel of F is calculated to generate a feature vector of a specific length; C Indicates that the feature vector is weighted learned through the fully connected layer to generate the weight of each channel; Adjust each channel to obtain the enhanced temperature change rate characteristic diagram F e : F e =A c ⊙F c Among them, ⊙ represents channel-by-channel point product; (3.3) Using the spatial attention mechanism to capture the enhanced temperature change rate feature map F e For correlation between different positions, the process is as follows: Generate spatial weights A through multi-scale convolution s : A s =σ(Conv2D(Concat(F max ,F avg ))) Concat means to convert F max and F avg Splicing in the channel dimension to generate a joint temperature change rate feature map; Conv2D means learning spatial attention weights from the spliced ​​joint temperature change rate feature map through two-dimensional convolution to extract significant features and global distribution features; The optimized enhanced temperature change rate characteristic diagram is F final : F final =A s ⊙F e The optimized temperature change rate characteristic diagram F final Correction is performed to obtain the temperature change rate error correction value of medium i during the heating period Among them, W corr is the modified weight matrix, Ω is the calculation area; The thermal conductivity calculation unit (5) calculates the medium thermal conductivity λ at the pile foundation length l according to the corrected temperature change rate. l ; Where q is the heating power of the linear heat source, in W / m; K is the temperature change rate at the length l of the pile foundation corrected by the hybrid attention neural network, in K; (4) Scour area detection unit (6) Calculate the thermal conductivity gradient g in the length direction of the pile foundation based on the medium thermal conductivity l for: Among them, λ l is the thermal conductivity at length l, λx is the thermal conductivity at length x adjacent to length l, X is the distance between the two measuring points, X is L / S, L is the total length of the monitoring section, and S is the sampling resolution; The scour area detection unit (6) detects the position difference of the thermal conductivity gradient mutation point between the thermal conductivity gradient curve and the reference curve when the pile foundation is installed, and obtains the scour area length d: d=L2-L1 Wherein, L2 is the position of the mutation point in the thermal conductivity gradient curve, and L1 is the position of the mutation point in the reference curve; Comparing the length of the scour area d with the original depth of the foundation D, the instability coefficient F is obtained: When the instability coefficient F is greater than the safety threshold f, the pile foundation scour state is unstable; when the instability coefficient F is less than the safety threshold f, the pile foundation scour state is stable.

3. The method for dynamic monitoring of the scouring interface of offshore wind power pile foundation according to claim 1 is characterized by: In step (3.1), the temperature change rate feature map F of the input feature is extracted through a lightweight one-dimensional convolutional layer c : F c =σ(W*X+b) Among them, W is the convolution kernel weight, b is the bias term, σ is the activation function; X is the input temperature change rate feature matrix of the neural network.

4. The method for dynamic monitoring of the scour interface of offshore wind power pile foundation according to claim 1 is characterized by: In step (3.2), based on the global pooling channel attention mechanism, the temperature change rate feature map F is calculated c The weight matrix A c : A c =σ(FC(GAP(F c ))) Among them, GAP represents the c The average value of each channel of F is calculated to generate a feature vector of a specific length; C It means that the feature vector is weighted learned through the fully connected layer to generate the weight of each channel.

5. The method for dynamic monitoring of the scouring interface of offshore wind power pile foundation according to claim 1 is characterized by: In step (3.3), the optimized temperature change rate characteristic graph F is obtained by quadratic weighted integration. final Correction is performed to obtain the temperature change rate error correction value of medium i during the heating period 6. The method for dynamic monitoring of the scouring interface of offshore wind power pile foundation according to claim 2 is characterized by: In step (3.3), the loss function L is used to optimize the modified weight matrix Wcorr of the hybrid attention neural network and the temperature change rate error correction value ΔK Ti ; The loss function is: Among them, w i is the weight for medium i, λ is the parameter of regularization strength, W corr is the modified weight matrix, λ||W corr || 2 is the regularization term; the loss function L updates the corrected weight matrix Wcorr at each iteration, and when the loss function L is minimized, the corrected temperature change rate K is output.

7. The method for dynamic monitoring of the scour interface of offshore wind power pile foundation according to claim 2, characterized in that: In step (1), the internal heating temperature measuring optical cable (1) is fixed on the offshore wind power pile foundation by using waterproof glue.

8. The method for dynamic monitoring of the scour interface of offshore wind power pile foundation according to claim 2 is characterized by: In step (1), the internal heating temperature measuring optical cable (1) is fixed and then protected by epoxy resin.

9. The method for dynamic monitoring of the scour interface of offshore wind power pile foundation according to claim 1, characterized in that: In step (1), an internal heating temperature measuring optical cable is reserved at the bottom of the pile foundation.

10. The method for dynamic monitoring of the scour interface of offshore wind power pile foundation according to claim 1, characterized in that: In step (2), the heating time and heating power of the heating unit (2) are determined according to the designed buried depth of the pile foundation and the average depth of seawater.

Images