An underwater wireless power transfer coil monitoring system and method

Abstract

The application discloses an underwater wireless energy transmission coil monitoring system and method, which comprises a plurality of grating sensors, optical fibers, wireless energy transmission coils, optical fiber demodulators, insulation wrapping layers and online monitoring devices; the optical fibers are wound on the wireless energy transmission coils; the grating sensors are equidistantly arranged on the surfaces of the wireless energy transmission coils and are connected with each other through the optical fibers; the two ends of the optical fibers are connected with the optical fiber demodulators; the grating sensors, the optical fibers and the wireless energy transmission coils are arranged in the insulation wrapping layers; the optical fiber demodulators are embedded in the insulation wrapping layers and are connected with the wireless energy transmission coils; and the optical fiber demodulators are connected with the online monitoring devices in communication, so that the technical problems that the existing underwater wireless energy transmission coil monitoring mode needs to consume a large amount of manpower and material resources, has a low degree of automation, cannot timely find the fault conditions of the wireless energy transmission coils and leads to an unsatisfactory monitoring effect are solved.

Description

Technical Field

[0001] This invention relates to the field of underwater coil fault monitoring technology, and in particular to an underwater wireless power transmission coil monitoring system and method. Background Technology

[0002] With economic and social development, electricity plays an increasingly important role in people's lives. Traditional power transmission often relies on massive power grids, which brings a series of problems such as frequent accidents and maintenance difficulties, as well as high construction costs and large investments of human and material resources. With the advancement of technology, more and more electrical devices, such as implantable and underwater devices, which operate in special environments, have an increasingly urgent need for wireless power transmission. Based on the huge commercial prospects, people have been committed to the research of wireless power transmission.

[0003] Underwater wireless power transfer coil systems have always been a significant challenge in the fields of marine engineering and underwater equipment. Due to the complex underwater environment, the long-term operating conditions of underwater wireless power transfer coil systems are also affected by various factors, such as high pressure, low temperature, and corrosion. These conditions may cause changes in temperature and stress inside the coil, affecting system performance and stability, and bringing unpredictability to the stable operation of underwater wireless power transfer systems.

[0004] Existing monitoring methods for underwater wireless power transfer coil systems mostly rely on underwater robots to scan and monitor the wireless power transfer coils when the system stops operating, or on manual inspection using monitoring instruments to determine if the wireless power transfer coils are faulty. These processes consume a lot of manpower and resources, have low automation, and cannot detect faults in the wireless power transfer coils in a timely manner, resulting in unsatisfactory monitoring results. Summary of the Invention

[0005] This invention provides an underwater wireless power transfer coil monitoring system and method, which solves the technical problems of existing underwater wireless power transfer coil monitoring methods that require a lot of manpower and material resources, have a low degree of automation, and cannot detect wireless power transfer coil failures in a timely manner, resulting in unsatisfactory monitoring results.

[0006] The first aspect of the present invention provides an underwater wireless power transfer coil monitoring system, the system comprising multiple grating sensors, optical fibers, wireless power transfer coils, an optical fiber demodulator, an insulating wrapping layer, and an online monitoring device;

[0007] The optical fiber is wound around the wireless power transmission coil.

[0008] Each of the grating sensors is equidistantly disposed on the surface of the wireless power transmission coil and interconnected with each other via the optical fiber.

[0009] Both ends of the optical fiber are connected to the optical fiber demodulator, and each of the grating sensors, the optical fiber, and the wireless power transmission coil are disposed within the insulating wrapping layer;

[0010] The fiber optic demodulator is embedded in the insulating wrapping layer and connected to the wireless power transmission coil.

[0011] The fiber optic demodulator is communicatively connected to the online monitoring device;

[0012] The grating sensor is used to acquire multiple dynamic time-series temperature stress signals of the wireless power transmission coil and transmit them to the online monitoring device through the fiber optic demodulator.

[0013] The online monitoring device is used to perform averaging calculation on multiple dynamic time-series temperature stress signals, output the average dynamic time-series temperature stress signal and compare it with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multi-modal one-dimensional model for fault prediction and the coil monitoring result is output.

[0014] Optionally, the online monitoring device includes a data acquisition module and a data processing module;

[0015] The data acquisition module is connected to the data processing module;

[0016] The data acquisition module is used to transmit multiple dynamic time-series temperature stress signals received from the fiber optic demodulator to the data processing module.

[0017] The data processing module is used to perform averaging calculation on multiple dynamic time-series temperature stress signals, output the average dynamic time-series temperature stress signal and compare it with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multi-modal one-dimensional model for fault prediction and the coil monitoring result is output.

[0018] The dynamic time-series temperature-stress signal includes a dynamic time-series temperature signal and a dynamic time-series stress signal.

[0019] Optionally, the dynamic time-series temperature stress average signal includes a dynamic time-series temperature average signal and a dynamic time-series stress average signal; the preset temperature stress signal threshold includes a preset temperature signal threshold and a preset stress signal threshold; the preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module; the data processing module is specifically used for:

[0020] The mean values ​​of each dynamic time-series temperature signal and each dynamic time-series stress signal are calculated to generate a dynamic time-series temperature mean signal and a dynamic time-series stress mean signal, which are then compared with preset temperature signal thresholds and preset stress signal thresholds.

[0021] If the dynamic time-series average temperature signal is less than a preset temperature signal threshold and the dynamic time-series average stress signal is less than a preset stress signal threshold, then the signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series average temperature signal and the dynamic time-series average stress signal respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data.

[0022] A convolutional network is used to perform data mapping on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data;

[0023] The output module fuses and predicts temperature mapping data and stress mapping data to output coil monitoring results.

[0024] Optionally, the one-dimensional temperature sequence data includes Fourier temperature amplitude, Fourier temperature phase, first-order differential temperature signal, second-order differential temperature signal, and Hilbert transform temperature signal; the one-dimensional stress sequence data includes Fourier stress amplitude, Fourier stress phase, first-order differential stress signal, second-order differential stress signal, and Hilbert transform stress signal; the data processing procedure of the signal processing module is specifically as follows:

[0025] The input dynamic time-series average temperature signal is subjected to Fourier transform, first-order derivative, second-order derivative and Hilbert transform respectively, and the corresponding Fourier transform temperature signal, first-order derivative temperature signal, second-order derivative temperature signal and Hilbert transform temperature signal are output.

[0026] The input dynamic time-series mean stress signal is subjected to Fourier transform, first-order differential, second-order differential and Hilbert transform respectively, and the corresponding Fourier transform stress signal, first-order differential stress signal, second-order differential stress signal and Hilbert transform stress signal are output.

[0027] Based on the Fourier transform temperature signal, determine the Fourier temperature amplitude and Fourier temperature phase.

[0028] Based on the Fourier transform stress signal, the Fourier stress amplitude and Fourier stress phase are determined.

[0029] Optionally, the convolutional network includes four cascaded convolutional layers, an adaptive max-pooling layer, and a fully connected layer; the data processing procedure of the convolutional network is specifically as follows:

[0030] Four cascaded convolutional layers are used to perform convolution operations on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively to generate one-dimensional temperature convolutional data and one-dimensional stress convolutional data.

[0031] The one-dimensional temperature convolutional data and the one-dimensional stress convolutional data are pooled by an adaptive maximum pooling layer, and the one-dimensional temperature pooled data and the one-dimensional stress pooled data are output.

[0032] The one-dimensional temperature pooling data and the one-dimensional stress pooling data are input into the fully connected layer for feature mapping to generate temperature mapping data and stress mapping data.

[0033] Optionally, the output module includes a data fusion layer and a classifier; the data processing procedure of the output module is specifically as follows:

[0034] The temperature mapping data and the stress mapping data are fused using a multi-parameter fusion algorithm through the data fusion layer to generate temperature-stress fused data.

[0035] A classifier is used to classify and predict the temperature stress fusion data, and the coil monitoring results are output.

[0036] Optionally, the data processing module is further configured to:

[0037] When the coil monitoring results show an abnormality, a coil warning command is generated.

[0038] Optionally, the online monitoring device further includes an alarm module;

[0039] The alarm module is connected to the data processing module;

[0040] The alarm module is used to issue an alarm in response to the coil warning command sent by the data processing module.

[0041] Optionally, the grating sensor includes a temperature grating sensor and a stress grating sensor; the optical fiber includes a temperature optical fiber and a stress optical fiber;

[0042] The stress fiber is wound around the wireless power transmission coil along the length of the wireless power transmission coil, and the temperature fiber is spirally wound around the wireless power transmission coil.

[0043] The temperature grating sensor and the stress grating sensor are disposed adjacent to each other on the surface of the wireless power transmission coil;

[0044] Each of the temperature grating sensors is interconnected via the temperature optical fiber, and each of the stress grating sensors is interconnected via the stress optical fiber.

[0045] Each of the temperature grating sensors, each of the stress grating sensors, the temperature optical fiber, the stress optical fiber, and the wireless power transmission coil are disposed within the insulating wrapping layer;

[0046] Both ends of the temperature fiber and both ends of the stress fiber are connected to the fiber demodulator.

[0047] The temperature grating sensor is used to acquire the dynamic time-series temperature signal of the wireless power transmission coil and transmit it to the data acquisition module through the fiber optic demodulator.

[0048] The stress grating sensor is used to acquire the dynamic time-series stress signal of the wireless power transmission coil and transmit it to the data acquisition module through the fiber optic demodulator.

[0049] The second aspect of this invention provides a method for monitoring underwater wireless power transfer coils.

[0050] When multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, and compared with a preset temperature stress signal threshold.

[0051] If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, then the average dynamic time-series temperature stress signal is input to the preset multimodal one-dimensional model, wherein the preset multimodal one-dimensional model includes a signal processing module, a convolutional network and an output module, and the average dynamic time-series temperature stress signal includes the average dynamic time-series temperature signal and the average dynamic time-series stress signal.

[0052] The signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series average temperature signal and the dynamic time-series average stress signal, respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data.

[0053] A convolutional network is used to perform data mapping on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data;

[0054] The output module fuses and predicts the temperature mapping data and the stress mapping data to output the coil monitoring results.

[0055] As can be seen from the above technical solutions, the present invention has the following advantages:

[0056] The first aspect of the technical solution of the present invention provides an underwater wireless power transfer coil monitoring system. This system includes multiple grating sensors, optical fibers, a wireless power transfer coil, an optical fiber demodulator, an insulating sheath, and an online monitoring device. The optical fibers are wound around the wireless power transfer coil. Each grating sensor is equidistantly disposed on the surface of the wireless power transfer coil and interconnected via optical fibers. Both ends of the optical fibers are connected to the optical fiber demodulator. Each grating sensor, optical fiber, and wireless power transfer coil is disposed within the insulating sheath. The optical fiber demodulator is embedded in the insulating sheath and connected to the wireless power transfer coil. The optical fiber demodulator is communicatively connected to the online monitoring device. First, the grating sensors acquire multiple dynamic time-series temperature stress signals from the wireless power transfer coil and transmit them to the online monitoring device via the optical fiber demodulator. Then, the online monitoring device averages the multiple dynamic time-series temperature stress signals. The calculation process outputs a dynamic time-series average temperature stress signal and compares it with a preset temperature stress signal threshold. Finally, if the dynamic time-series average temperature stress signal is less than the preset temperature stress signal threshold, the dynamic time-series average temperature stress signal is input into a preset multi-modal one-dimensional model for fault prediction, and the coil monitoring result is output. This scheme, by comparing the dynamic time-series average temperature stress signal with a preset temperature stress signal threshold and combining the dynamic time-series average temperature stress signal input into a preset multi-modal one-dimensional model for fault prediction, can replace the traditional manual judgment method, promptly detect faults in the wireless power transmission coil, and achieve better monitoring results. At the same time, by performing threshold judgment and model fault judgment based on the dynamic time-series average temperature stress signal, the accuracy of the coil monitoring results is further improved.

[0057] The second aspect of the above-mentioned technical solution of the present invention provides an underwater wireless power transfer coil monitoring method. When multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, which is then compared with a preset temperature stress signal threshold. If the dynamic time-series temperature stress average signal is less than the preset temperature stress signal threshold, the dynamic time-series temperature stress average signal is input to a preset multimodal one-dimensional model. The preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module. The dynamic time-series temperature stress average signal includes a dynamic time-series temperature average signal and a dynamic time-series stress average signal. The signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series temperature average signal and the dynamic time-series stress average signal, respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data. The convolutional network is used to process the one-dimensional temperature sequence data... The system maps temperature and stress data separately to one-dimensional stress sequence data, generating temperature and stress mapping data. The output module then fuses and predicts these data to output coil monitoring results. This approach, by comparing the dynamic time-series average temperature and stress signal with a preset temperature and stress signal threshold, and inputting the dynamic time-series average temperature and stress signal into a preset multimodal one-dimensional model for fault prediction, outputs coil monitoring results. Compared to existing methods of scanning and monitoring wireless power transfer coils using underwater robots or manually determining faults using monitoring instruments, this method replaces traditional manual methods, enabling timely detection of wireless power transfer coil faults and achieving better monitoring results. Furthermore, the sequential comparison of threshold judgments and model fault judgments based on the dynamic time-series average temperature and stress signal further improves the accuracy of coil monitoring results. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a schematic diagram of the structure between multiple grating sensors, optical fibers, insulating wrapping layers, and wireless power transfer coils in an underwater wireless power transfer coil monitoring system provided in Embodiment 1 of the present invention.

[0060] Figure 2 This is a schematic diagram of the structure between the wireless power transfer coil, the fiber optic demodulator, and the online monitoring device in an underwater wireless power transfer coil monitoring system provided in Embodiment 1 of the present invention.

[0061] Figure 3 This is a schematic diagram of the structure of a pre-set multimodal one-dimensional model provided in Embodiment 1 of the present invention;

[0062] Figure 4 This is a schematic diagram illustrating the operation of an underwater wireless power transfer coil monitoring system according to Embodiment 1 of the present invention.

[0063] Figure 5 This is a flowchart illustrating the steps of an underwater wireless power transfer coil monitoring method provided in Embodiment 2 of the present invention.

[0064] The meanings of the labels in the attached figures are as follows:

[0065] 1. Wireless power transmission coil; 2. Temperature fiber optic cable; 3. Temperature grating sensor; 4. Stress fiber optic cable; 5. Stress grating sensor; 6. Insulating wrapping layer; 7. Fiber optic demodulator; 8. Online monitoring device. Detailed Implementation

[0066] This invention provides an underwater wireless power transfer coil monitoring system and method to solve the technical problems of existing underwater wireless power transfer coil monitoring methods, which require a large amount of manpower and material resources, have a low degree of automation, and cannot detect wireless power transfer coil malfunctions in a timely manner, resulting in unsatisfactory monitoring results.

[0067] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0068] Please see Figures 1-2 , Figure 1 This is a schematic diagram of the structure between multiple grating sensors, optical fibers, insulating wrapping layers, and wireless power transfer coils in an underwater wireless power transfer coil monitoring system provided in Embodiment 1 of the present invention.

[0069] This invention provides an underwater wireless power transfer coil monitoring system, which includes multiple grating sensors, optical fibers, a wireless power transfer coil 1, an optical fiber demodulator 7, an insulating wrapping layer 6, and an online monitoring device 8. The optical fibers are wound around the wireless power transfer coil 1. Each grating sensor is equidistantly arranged on the surface of the wireless power transfer coil 1 and interconnected with each other through the optical fibers. Both ends of the optical fibers are connected to the optical fiber demodulator 7. Each grating sensor, optical fiber, and wireless power transfer coil 1 are disposed within the insulating wrapping layer 6. The optical fiber demodulator 7 is embedded in the insulating wrapping layer 6 and connected to the wireless power transfer coil 1. The optical fiber demodulator 7 is communicatively connected to the online monitoring device 8.

[0070] It should be noted that by wrapping the optical fiber around the wireless power transfer coil 1, and equidistantly arranging each grating sensor on the surface of the wireless power transfer coil 1 and interconnecting them with optical fibers, a high degree of coupling can be achieved between the wireless power transfer coil 1, the grating sensors, and the optical fiber. At the same time, by placing each grating sensor, optical fiber, and wireless power transfer coil 1 within the insulating wrapping layer 6, it is possible to better adapt to the extreme conditions in the underwater environment while ensuring transmission efficiency and insulation performance. When the underwater wireless power transfer coil system is transmitting power, the wireless power transfer coil can supply power to the grating sensors and the online monitoring device 8, ensuring the normal operation of the online monitoring functions of the grating sensors and the online monitoring device 8.

[0071] In this embodiment, based on the structure of the wireless power transmission coil 1 itself, fiber optic gratings (grating sensors, optical fibers) are alternately wound around the wireless power transmission coil 1 to ensure the uniform distribution of the fiber optic gratings and improve the overall sensitivity and sensing ability of the coil. In addition, corrosion-resistant insulating material (insulating wrapping layer 6) is used to wrap the wireless power transmission coils 1, which can also ensure the insulation between the wireless power transmission coils 1 and the resistance to seawater corrosion, while ensuring the safety of the fiber optic gratings.

[0072] The grating sensor is used to acquire multiple dynamic time-series temperature stress signals of the wireless power transmission coil 1, and transmits them to the online monitoring device 8 through the fiber optic demodulator 7.

[0073] The dynamic time-series temperature-stress signal includes the dynamic time-series temperature signal and the dynamic time-series stress signal.

[0074] It should be noted that after the grating sensor transmits the multiple dynamic time-series temperature signals and multiple dynamic time-series stress signals acquired by the wireless power transmission coil 1 to the fiber optic demodulator 7, the fiber optic demodulator 7 will convert the multiple dynamic time-series temperature signals and multiple dynamic time-series stress signals acquired by the grating sensor in the form of optical signals into multiple dynamic time-series temperature signals and multiple dynamic time-series stress signals in the form of electrical signals, and then transmit them to the online monitoring device 8.

[0075] In this embodiment, a high-sensitivity fiber Bragg grating (grating sensor, optical fiber) is used as the sensing element, which can respond with high precision to minute temperature and stress changes. The fiber Bragg gratings are distributed along the surface of the wireless power transfer coil 1 to form a distributed sensing network, realizing real-time monitoring of the entire interior of the wireless power transfer coil 1, thereby obtaining more comprehensive data and helping to accurately identify the location and extent of abnormal situations.

[0076] The online monitoring device 8 is used to perform averaging calculation on multiple dynamic time-series temperature stress signals, output the average dynamic time-series temperature stress signal and compare it with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multi-modal one-dimensional model for fault prediction and the coil monitoring result is output.

[0077] The online monitoring device 8 includes a data acquisition module and a data processing module.

[0078] Please see Figure 3 The pre-defined multimodal one-dimensional model is a multimodal one-dimensional CNN model, consisting of a signal processing module, a convolutional network, and an output module. The signal processing module processes the dynamic time-series mean temperature signal and the dynamic time-series mean stress signal through Fourier transform, first-order differentiation, second-order differentiation, and Hilbert transform. The convolutional network consists of four cascaded convolutional layers, an adaptive max-pooling layer, and a fully connected layer. The output module consists of a data fusion layer and a classifier. The data fusion layer is used to perform data fusion processing on the input temperature and stress data.

[0079] Furthermore, the training process of the pre-set multimodal one-dimensional model is as follows: Multiple historical temperature and stress signal data accumulated after the system has been running for a period of time are used. These historical temperature and stress signal data include both normal and abnormal results. These data are labeled as normal or abnormal and then placed into the untrained pre-set multimodal one-dimensional model. Features are extracted from the convolutional modules in the pre-set multimodal one-dimensional model and classified. Key normal / abnormal classification criteria are obtained after feature extraction, thereby determining the parameters corresponding to each module in the pre-set multimodal one-dimensional model. This yields the parameters of the pre-set multimodal one-dimensional model (convolutional kernel, classification criteria, etc.), resulting in a trained pre-set multimodal one-dimensional model.

[0080] Furthermore, firstly, the data acquisition module acquires various dynamic time-series temperature signals and stress signals. Then, the data processing module performs averaging operations on each dynamic time-series temperature and stress signal to generate dynamic time-series average temperature and stress signals. The dynamic time-series average temperature and stress signals are then compared to preset temperature and stress thresholds. If the dynamic time-series average temperature signal is greater than or equal to the preset temperature threshold, or the dynamic time-series average stress signal is lower than or equal to the preset stress threshold, the comparison is performed. If the value signal is greater than or equal to the preset temperature signal threshold, an early warning command is generated and the alarm device in the online monitoring device 8 responds to the command to trigger an alarm. The above data is retained, and the preset multimodal one-dimensional model is further optimized. If the dynamic time-series average temperature signal is less than the preset temperature signal threshold and the dynamic time-series average stress signal is less than the preset stress signal threshold, the preset multimodal one-dimensional model is used to predict the faults of the dynamic time-series average temperature signal and the dynamic time-series average stress signal, outputting one-dimensional temperature sequence data and one-dimensional stress sequence data, and outputting the coil monitoring results.

[0081] As a further improvement, the online monitoring device 8 also includes an alarm module;

[0082] The alarm module is connected to the data processing module;

[0083] The alarm module is used to trigger an alarm in response to the coil warning command sent by the data processing module.

[0084] It should be noted that the coil monitoring results output by the preset multimodal one-dimensional model include normal coil monitoring results and abnormal coil monitoring results. That is, the coil monitoring results output by the preset multimodal one-dimensional model are either normal coil monitoring results or abnormal coil monitoring results. When the coil monitoring result is an abnormal coil monitoring result, the data processing module will generate a coil warning command, and the alarm module will respond to the coil warning command sent by the data processing module to issue an alarm.

[0085] In this embodiment, the online monitoring device 8 should include a data acquisition module, a data processing module, and an alarm module. The data acquisition module is responsible for real-time acquisition of dynamic time-series temperature and stress signals transmitted via a distributed sensor network, i.e., through a grating sensor, optical fiber, and optical fiber demodulator 7. The data processing module processes the acquired data (dynamic time-series temperature and stress signals) in real-time through real-time threshold comparison and fault judgment based on a preset multimodal one-dimensional model, identifies possible anomalies, and performs corresponding data predictions to provide more accurate monitoring results. The preset multimodal one-dimensional model built into the data processing module is constructed based on previously collected temperature and strain parameters, enabling fault prediction. The online monitoring device 8 also integrates an alarm module, which can promptly respond to commands and issue alarm signals once an anomaly is detected inside the wireless power transmission coil 1, transmitting the signals to the ground control center via underwater communication technology.

[0086] As a further improvement, the online monitoring device 8 includes a data acquisition module and a data processing module;

[0087] The data acquisition module is connected to the data processing module;

[0088] The data acquisition module is used to transmit multiple dynamic time-series temperature stress signals sent by the fiber optic demodulator to the data processing module;

[0089] The data processing module is used to perform mean calculation on multiple dynamic time-series temperature stress signals, output the mean dynamic time-series temperature stress signal and compare it with the preset temperature stress signal threshold. If the mean dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the mean dynamic time-series temperature stress signal is input to the preset multi-modal one-dimensional model for fault prediction and outputs the coil monitoring results.

[0090] The dynamic time-series temperature-stress signal includes both dynamic time-series temperature signal and dynamic time-series stress signal.

[0091] As a further improvement, the dynamic time-series temperature stress mean signal includes a dynamic time-series temperature mean signal and a dynamic time-series stress mean signal; the preset temperature stress signal threshold includes a preset temperature signal threshold and a preset stress signal threshold; the preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module; the data processing module is specifically used for:

[0092] The mean values ​​of each dynamic time-series temperature signal and each dynamic time-series stress signal are calculated to generate the mean dynamic time-series temperature signal and the mean dynamic time-series stress signal, and then compared with the preset temperature signal threshold and the preset stress signal threshold.

[0093] If the average dynamic temperature signal is less than the preset temperature signal threshold and the average dynamic stress signal is less than the preset stress signal threshold, the signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the average dynamic temperature signal and the average dynamic stress signal respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data.

[0094] A convolutional network is used to map one-dimensional temperature sequence data and one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data.

[0095] The output module fuses and predicts temperature mapping data and stress mapping data to output coil monitoring results.

[0096] As a further improvement, the one-dimensional temperature sequence data includes Fourier temperature amplitude, Fourier temperature phase, first-order differential temperature signal, second-order differential temperature signal, and Hilbert transform temperature signal; the one-dimensional stress sequence data includes Fourier stress amplitude, Fourier stress phase, first-order differential stress signal, second-order differential stress signal, and Hilbert transform stress signal; the data processing procedure of the signal processing module is as follows:

[0097] The input dynamic time-series average temperature signal is subjected to Fourier transform, first-order derivative, second-order derivative and Hilbert transform respectively, and the corresponding Fourier transform temperature signal, first-order derivative temperature signal, second-order derivative temperature signal and Hilbert transform temperature signal are output.

[0098] The input dynamic time-series mean stress signal is subjected to Fourier transform, first-order differential, second-order differential and Hilbert transform respectively, and the corresponding Fourier transform stress signal, first-order differential stress signal, second-order differential stress signal and Hilbert transform stress signal are output.

[0099] Based on the Fourier transform temperature signal, determine the Fourier temperature amplitude and Fourier temperature phase.

[0100] Based on the Fourier transform stress signal, the Fourier stress amplitude and Fourier stress phase are determined.

[0101] It should be noted that the signal processing module performs Fourier transform, first-order differential, second-order differential, and Hilbert transform on both the input dynamic time-series average temperature signal and dynamic time-series average stress signal, respectively, thereby outputting the corresponding Fourier transform temperature signal, first-order differential temperature signal, second-order differential temperature signal, and Hilbert transform temperature signal, as well as the corresponding Fourier transform stress signal, first-order differential stress signal, second-order differential stress signal, and Hilbert transform stress signal. Then, based on the Fourier transform temperature signal, the Fourier temperature amplitude and Fourier temperature phase are determined; based on the Fourier transform stress signal, the Fourier stress amplitude and Fourier stress phase are determined. The first-order differential processing is specifically as follows:

[0102]

[0103] Where f'(t) is the first-order differential signal; f(t) is the dynamic time-series mean signal; and t is time.

[0104] The process of finding the second derivative is as follows:

[0105]

[0106] Where f(t) is the second-order differential signal; f(t) is the dynamic time-series mean signal; and t is time.

[0107] The specific process of Hilbert transform is as follows:

[0108]

[0109] Among them, f * f(t) is the Hilbert transform signal; f(τ) is the integral variable signal corresponding to the dynamic time-series mean signal; τ is the integral variable; t is time; f(t) is the dynamic time-series mean signal.

[0110] The formulas for calculating Fourier amplitude and Fourier phase are as follows:

[0111]

[0112] Among them, A m (ω) is the Fourier amplitude; φ m (ω) is the Fourier phase; F(ω) is the Fourier transform signal; f(t) is the dynamic time-series mean signal; j is the imaginary unit; t is time; ω is the frequency; Re(ω) is the real part of the Fourier transform signal; Im(ω) is the imaginary part of the Fourier transform signal.

[0113] As a further improvement, the convolutional network includes four cascaded convolutional layers, an adaptive max pooling layer, and a fully connected layer. The data processing procedure of the convolutional network is as follows: the four cascaded convolutional layers perform convolution operations on one-dimensional temperature sequence data and one-dimensional stress sequence data respectively, generating one-dimensional temperature convolutional data and one-dimensional stress convolutional data; the adaptive max pooling layer performs pooling operations on the one-dimensional temperature convolutional data and one-dimensional stress convolutional data respectively, outputting one-dimensional temperature pooled data and one-dimensional stress pooled data; the one-dimensional temperature pooled data and one-dimensional stress pooled data are input into the fully connected layer for feature mapping, generating temperature mapping data and stress mapping data.

[0114] One-dimensional temperature convolution data includes Fourier convolution temperature amplitude, Fourier convolution temperature phase, first-order differential convolution temperature signal, second-order differential convolution temperature signal, and Hilbert transform convolution temperature signal.

[0115] One-dimensional stress convolution data includes Fourier convolution stress amplitude, Fourier convolution stress phase, first-order differential convolution stress signal, second-order differential convolution stress signal, and Hilbert transform convolution stress signal.

[0116] It should be noted that the four cascaded convolutional layers (convolutional layer 1, convolutional layer 2, convolutional layer 3, and convolutional layer 4) perform convolution operations on the Fourier temperature amplitude, Fourier temperature phase, first-order differential temperature signal, second-order differential temperature signal, and Hilbert transform temperature signal, respectively, outputting the corresponding Fourier convolutional temperature amplitude, Fourier convolutional temperature phase, first-order differential convolutional temperature signal, second-order differential convolutional temperature signal, and Hilbert transform convolutional temperature signal. Similarly, the four cascaded convolutional layers perform convolution operations on the Fourier stress amplitude, Fourier stress phase, first-order differential stress signal, second-order differential stress signal, and Hilbert transform stress signal, respectively, outputting the corresponding Fourier convolutional stress amplitude, Fourier convolutional stress phase, first-order differential convolutional stress signal, second-order differential convolutional stress signal, and Hilbert transform convolutional stress signal.

[0117] Furthermore, the Fourier convolution temperature amplitude, Fourier convolution temperature phase, first-order differential convolution temperature signal, second-order differential convolution temperature signal, and Hilbert transform convolution temperature signal are pooled by an adaptive maximum pooling layer, respectively, and the corresponding Fourier pooling temperature amplitude, Fourier pooling temperature phase, first-order differential pooling temperature signal, second-order differential pooling temperature signal, and Hilbert transform pooling temperature signal are output. Among them, the Fourier pooling temperature amplitude, Fourier pooling temperature phase, first-order differential pooling temperature signal, second-order differential pooling temperature signal, and Hilbert transform pooling temperature signal constitute one-dimensional temperature pooling data.

[0118] An adaptive maximum pooling layer is used to pool the Fourier convolution stress amplitude, Fourier convolution stress phase, first-order differential convolution stress signal, second-order differential convolution stress signal, and Hilbert transform convolution stress signal, respectively, and outputs the corresponding Fourier pooled stress amplitude, Fourier pooled stress phase, first-order differential pooled stress signal, second-order differential pooled stress signal, and Hilbert transform pooled stress signal. Among them, the Fourier pooled stress amplitude, Fourier pooled stress phase, first-order differential pooled stress signal, second-order differential pooled stress signal, and Hilbert transform pooled stress signal constitute one-dimensional stress pooling data.

[0119] Furthermore, the Fourier pooling temperature amplitude, Fourier pooling temperature phase, first-order differential pooling temperature signal, second-order differential pooling temperature signal, and Hilbert transform pooling temperature signal are respectively characterized by a fully connected layer, and the corresponding Fourier mapped temperature amplitude, Fourier mapped temperature phase, first-order differential mapped temperature signal, second-order differential mapped temperature signal, and Hilbert transform mapped temperature signal are output. Among them, the Fourier mapped temperature amplitude, Fourier mapped temperature phase, first-order differential mapped temperature signal, second-order differential mapped temperature signal, and Hilbert transform mapped temperature signal constitute temperature mapping data.

[0120] The Fourier pooling stress amplitude, Fourier pooling stress phase, first-order differential pooling stress signal, second-order differential pooling stress signal, and Hilbert transform pooling stress signal are feature-mapped through a fully connected layer, and the corresponding Fourier mapped stress amplitude, Fourier mapped stress phase, first-order differential mapped stress signal, second-order differential mapped stress signal, and Hilbert transform mapped stress signal are output. The Fourier mapped stress amplitude, Fourier mapped stress phase, first-order differential mapped stress signal, second-order differential mapped stress signal, and Hilbert transform mapped stress signal constitute the stress mapping data.

[0121] As a further modification, the output module includes a data fusion layer and a classifier; the data processing procedure of the output module is as follows: the temperature mapping data and stress mapping data are fused using a multi-parameter fusion algorithm through the data fusion layer to generate temperature-stress fused data; the temperature-stress fused data is classified and predicted using a classifier to output the coil monitoring results.

[0122] It should be noted that the data fusion layer employs a multi-parameter fusion algorithm to fuse Fourier-mapped temperature amplitude, Fourier-mapped temperature phase, first-order differential-mapped temperature signal, second-order differential-mapped temperature signal, and Hilbert transform-mapped temperature signal, as well as Fourier-mapped stress amplitude, Fourier-mapped stress phase, first-order differential-mapped stress signal, second-order differential-mapped stress signal, and Hilbert transform-mapped stress signal, generating temperature-stress fused data. The specific processing steps of the multi-parameter fusion algorithm are as follows:

[0123] Y PCA =DD T X;

[0124] Among them, Y PCA This represents temperature stress fusion data; D is the encoding matrix, and D must satisfy: I is the identity matrix; D T X is the transpose of the encoding matrix; X is the modal information matrix composed of Fourier-mapped temperature amplitude, Fourier-mapped temperature phase, first-order differential-mapped temperature signal, second-order differential-mapped temperature signal, and Hilbert-transform-mapped temperature signal, as well as Fourier-mapped stress amplitude, Fourier-mapped stress phase, first-order differential-mapped stress signal, second-order differential-mapped stress signal, and Hilbert-transform-mapped stress signal; ||·|| F It is the Frobenius norm (matrix 2-norm).

[0125] Furthermore, when the pre-set multimodal one-dimensional model training is completed, all parameters in the classifier of the pre-set multimodal one-dimensional model have been adjusted, and it can classify whether the input temperature and stress fusion data is abnormal or normal, thereby outputting the coil monitoring results. For example, during the monitoring process, the collected temperature and stress data are processed by a multi-parameter fusion algorithm through a pre-network, and the fused data is output. The classifier with configured parameters classifies and predicts the data. The built-in parameters of the classifier will automatically determine whether the fused data is abnormal or normal, thereby outputting normal or abnormal results (coil monitoring results).

[0126] As a further improvement, the data processing module is also used to generate a coil warning command when the coil monitoring result shows an abnormality.

[0127] It should be noted that when the classifier determines that the input temperature stress fusion data is abnormal, that is, when the output coil monitoring result is an abnormal coil monitoring result, a coil early warning command is generated and the alarm module is triggered to sound an alarm.

[0128] As a further improvement, the grating sensor includes a temperature grating sensor 3 and a stress grating sensor 5; the optical fiber includes a temperature optical fiber 2 and a stress optical fiber 4; the stress optical fiber 4 is wound around the wireless power transmission coil 1 along the length of the wireless power transmission coil, and the temperature optical fiber 2 is spirally wound around the wireless power transmission coil 1; the temperature grating sensor 3 and the stress grating sensor 5 are arranged adjacent to each other on the surface of the wireless power transmission coil 1; each temperature grating sensor 3 is interconnected through the temperature optical fiber 2, and each stress grating sensor 5 is interconnected through the stress optical fiber 4; each temperature grating sensor 3, each stress grating sensor 5, the temperature optical fiber 2, the stress optical fiber 4, and the wireless power transmission coil 1 are disposed within an insulating wrapping layer 6; both ends of the temperature optical fiber 2 and both ends of the stress optical fiber 4 are respectively connected to an optical fiber demodulator 7; the temperature grating sensor 3 is used to acquire the dynamic time-series temperature signal of the wireless power transmission coil 1 and transmits it to the data acquisition module through the optical fiber demodulator 7; the stress grating sensor 5 is used to acquire the dynamic time-series stress signal of the wireless power transmission coil 1 and transmits it to the data acquisition module through the optical fiber demodulator 7.

[0129] It should be noted that, based on the structure of the wireless power transmission coil 1 itself, fiber optic gratings (temperature grating sensor 3 and stress grating sensor 5, and optical fibers including temperature optical fiber 2 and stress optical fiber 4) are alternately wound around the coil, ensuring the uniform distribution of the fiber optic gratings and improving the overall sensitivity and sensing capability of the coil.

[0130] Furthermore, the multiple grating sensors include multiple temperature grating sensors 3 and multiple stress grating sensors 5. The temperature grating sensors 3 are interconnected through temperature optical fibers 2, and the stress grating sensors 5 are interconnected through stress optical fibers 4. Both ends of the temperature optical fiber 2 and both ends of the stress optical fiber 4 are connected to the fiber demodulator 7, which enables the dynamic time-series temperature signal and dynamic time-series stress signal acquired by each temperature grating sensor 3 and each stress grating sensor 5 to be transmitted to the fiber demodulator 7.

[0131] In this embodiment, a fiber optic grating is wound around the wireless power transfer coil 1 to achieve real-time acquisition of temperature and stress data (dynamic time-series temperature and stress signals) of the wireless power transfer coil 1. The online monitoring device 8 acquires and processes the temperature and stress data. Based on a preset parameter threshold (preset temperature and stress signal threshold), the acquired temperature and stress data are compared in real time. If the acquired data is greater than or equal to the threshold, it is determined that the working state of the wireless power transfer coil 1 is not good, and an alarm is triggered by the alarm device in the online monitoring device 8 to remind maintenance personnel to handle the fault. At the same time, the acquired historical temperature and stress data are collected and a database is built. This database contains long-term temperature and stress maintenance data of the wireless power transfer coil 1. This data is used to train a preset multimodal one-dimensional model, resulting in a trained preset multimodal one-dimensional model, which is built into the data processing unit. When the wireless power transfer coil 1 is performing power transfer, the fiber optic grating acquires the temperature and stress data of the wireless power transfer coil 1 in real time, analyzes the dynamic time-domain data, and realizes early prediction of faults.

[0132] For example, please refer to Figure 4 By combining a fiber Bragg grating (fiber optic, grating sensor) with a wireless power transfer coil 1, specifically by arranging fiber Bragg grating sensors around the wireless power transfer coil 1, a high degree of coupling between the fiber Bragg grating sensors and the wireless power transfer coil 1 is achieved. This enables high-frequency, high-precision real-time monitoring of the coil's operating environment, providing real-time environmental monitoring for the underwater wireless coil power transfer system. The insulating wrapping layer 6 ensures the system's safety. Simultaneously, the wireless power transfer coil 1 supplies power to the online monitoring device 8, avoiding the dependence on external power sources found in traditional online monitoring equipment. This allows the online monitoring device 8 and the wireless power transfer coil system to operate synchronously, minimizing wireless monitoring time and improving online monitoring efficiency. The efficiency of the measurement is improved, and the temperature and stress data of the wireless power transfer coil 1 are collected in real time through fiber optic gratings. The data is analyzed in real time through the online monitoring device 8. If the data is abnormal, an alarm is triggered. Finally, based on the collected temperature and stress data, i.e., the historical dynamic time-varying stress and temperature data, the data is used as the training set to train the model, and a pre-trained pre-set multimodal one-dimensional model is obtained. By embedding the pre-set multimodal one-dimensional model into the online monitoring device 8, the fault prediction of the wireless power transfer coil 1 is realized. The pre-set multimodal one-dimensional model adopts a multimodal one-dimensional CNN model. The network front-end includes a time-frequency domain transformation algorithm, and the network back-end includes a multi-parameter fusion algorithm.

[0133] For comparison of technical effectiveness, existing technologies can be used as a reference. Current underwater wireless power transfer systems typically struggle to monitor the internal state of the coil in real time and cannot quickly capture changes in key parameters such as temperature and stress, resulting in a delayed understanding of the system's status. In underwater environments, due to extreme conditions such as high pressure, low temperature, and corrosion, temperature and stress changes may occur inside the coil, and current systems are unable to effectively adapt to and cope with these conditions, leading to insufficient system stability. Furthermore, due to the lack of highly coupled real-time monitoring methods, the system usually cannot issue alarm signals in a timely manner when encountering internal anomalies, preventing maintenance personnel from taking rapid measures, which may lead to equipment failure or damage.

[0134] Furthermore, traditional underwater energy transmission methods typically employ wired power transmission systems, which suffer from a series of significant drawbacks that hinder the widespread application of underwater equipment. Current underwater wired power transmission systems involve complex and costly cabling processes, are susceptible to corrosion from the marine environment, have high maintenance costs, limit the mobility of underwater equipment, and affect the efficiency of underwater operations. Moreover, due to the complexity of the underwater environment, the long-term operating conditions of underwater wireless power transfer coil systems are affected by various factors, such as high voltage, low temperature, and corrosion. These conditions can cause changes in temperature and stress within the coil, affecting system performance and stability. Simultaneously, the complexity of the underwater environment presents new requirements for underwater wireless power transfer coil systems and monitoring. First, a compact, highly coupled system is needed, capable of sensitively sensing changes in temperature and stress within the coil and promptly feeding back to the ground control center via a real-time data transmission system. Second, the solution should be able to adapt to extreme underwater conditions such as high pressure, low temperature, and corrosion to ensure the reliability and durability of the system in various underwater applications. Existing detection and monitoring systems are insufficient to meet these requirements and struggle to detect anomalies in the coil's internal state in a timely manner, introducing unpredictability into the stable operation of underwater wireless power transfer systems.

[0135] To address the aforementioned issues, this application proposes an underwater wireless power transfer coil monitoring system. By alternately winding fiber optic gratings and coils to form a highly coupled structure, and using a grating sensor to collect data from the wireless power transfer coil 1 in real time, it achieves real-time monitoring of key information such as internal temperature and stress. When abnormal changes occur in the coil, the system can issue an alarm signal in real time through the online monitoring device 8, enabling operators to take rapid action to ensure the reliability and stability of the underwater system. This further improves the underwater wireless power transfer coil system's ability to perceive its internal state. Furthermore, the high sensitivity of the fiber optic and grating sensors, combined with the online monitoring device 8, allows for more timely detection and response to abnormal conditions within the coil. In addition, the system is capable of adapting to extreme conditions such as high pressure, low temperature, and corrosion underwater, enabling reliable operation in various underwater environments and effectively solving the problems of traditional underwater wireless power transfer coil systems in monitoring and responding to internal anomalies.

[0136] Specifically, by alternately winding fiber Bragg gratings with coils, key parameters such as coil temperature and stress can be sensed at high frequency and with high precision. The online monitoring device 8 provides real-time feedback of monitoring results and promptly issues alarm signals when anomalies are detected, ensuring operators can take swift action to prevent potential risks. Simultaneously, the fiber Bragg grating (including temperature fiber, temperature grating sensor, and stress fiber, and stress grating sensor) is highly coupled with the wireless power transfer coil 1, ensuring transmission efficiency and insulation performance while better adapting to extreme conditions in the underwater environment. The application of fiber Bragg grating sensors helps to perceive changes inside the coil in real time, enabling system maintenance personnel to react more promptly, thereby improving the stability of the underwater wireless power transfer coil system. Through long-term real-time monitoring data accumulation, an operational status database of the underwater wireless power transfer coil is constructed. This database is used to train a pre-set multimodal one-dimensional model, resulting in a well-trained pre-set multimodal one-dimensional model, which is then integrated into the online monitoring device 8. This allows for further fault diagnosis, improving the accuracy of fault results, enabling preventative measures to be taken, reducing the risk of equipment damage and system failure, and improving maintenance efficiency.

[0137] In this embodiment of the invention, an underwater wireless power transfer coil monitoring system is provided. The system includes multiple grating sensors, optical fibers, a wireless power transfer coil, an optical fiber demodulator, an insulating sheath, and an online monitoring device. The optical fibers are wound around the wireless power transfer coil. Each grating sensor is equidistantly disposed on the surface of the wireless power transfer coil and interconnected via the optical fibers. Both ends of the optical fibers are connected to the optical fiber demodulator. Each grating sensor, optical fiber, and wireless power transfer coil is disposed within the insulating sheath. The optical fiber demodulator is embedded in the insulating sheath and connected to the wireless power transfer coil. The optical fiber demodulator is communicatively connected to the online monitoring device. First, the grating sensors acquire multiple dynamic time-series temperature stress signals from the wireless power transfer coil and transmit them to the online monitoring device via the optical fiber demodulator. Then, the online monitoring device performs an averaging operation on the multiple dynamic time-series temperature stress signals. The algorithm calculates and outputs a dynamic time-series average temperature stress signal, which is then compared to a preset temperature stress signal threshold. Finally, if the dynamic time-series average temperature stress signal is less than the preset temperature stress signal threshold, the dynamic time-series average temperature stress signal is input into a preset multi-modal one-dimensional model for fault prediction, and the coil monitoring result is output. This scheme, by comparing the dynamic time-series average temperature stress signal with a preset temperature stress signal threshold and combining the dynamic time-series average temperature stress signal input into a preset multi-modal one-dimensional model for fault prediction, can replace the traditional manual judgment method, promptly detect faults in the wireless power transmission coil, and achieve better monitoring results. At the same time, by performing threshold judgment and model fault judgment based on the dynamic time-series average temperature stress signal, the accuracy of the coil monitoring results is further improved.

[0138] Please see Figure 5 , Figure 5 This is a flowchart illustrating the steps of an underwater wireless power transfer coil monitoring method provided in Embodiment 2 of the present invention.

[0139] This invention provides a method for monitoring underwater wireless power transfer coils, comprising:

[0140] Step 501: When multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, and compared with a preset temperature stress signal threshold.

[0141] In this embodiment, when multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, and then compared with a preset temperature stress signal threshold.

[0142] Step 502: If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, then the average dynamic time-series temperature stress signal is input to the preset multimodal one-dimensional model. The preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module. The average dynamic time-series temperature stress signal includes the average dynamic time-series temperature signal and the average dynamic time-series stress signal.

[0143] In this embodiment, if the average dynamic time-series temperature stress signal is less than a preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multimodal one-dimensional model. The preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module. The average dynamic time-series temperature stress signal includes both the average dynamic temperature signal and the average dynamic time-series stress signal.

[0144] Step 503: The signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series mean temperature signal and the dynamic time-series mean stress signal respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data.

[0145] In this embodiment, the signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series average temperature signal and the dynamic time-series average stress signal, respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data.

[0146] Step 504: Use a convolutional network to perform data mapping on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively, to generate temperature mapping data and stress mapping data.

[0147] In this embodiment, a convolutional network is used to perform data mapping on one-dimensional temperature sequence data and one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data.

[0148] Step 505: The temperature mapping data and stress mapping data are fused and predicted through the output module, and the coil monitoring results are output.

[0149] In this embodiment, the output module fuses and predicts the temperature mapping data and stress mapping data to output the coil monitoring results.

[0150] In this embodiment of the invention, a method for monitoring underwater wireless power transfer coils is provided. When multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, which is then compared with a preset temperature stress signal threshold. If the dynamic time-series temperature stress average signal is less than the preset temperature stress signal threshold, the dynamic time-series temperature stress average signal is input to a preset multimodal one-dimensional model. The preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module. The dynamic time-series temperature stress average signal includes a dynamic time-series temperature average signal and a dynamic time-series stress average signal. The signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series temperature average signal and the dynamic time-series stress average signal, respectively, to output one-dimensional temperature sequence data and one-dimensional stress sequence data. The convolutional network is used to process the one-dimensional temperature sequence data. The method maps one-dimensional stress sequence data to generate temperature mapping data and stress mapping data. The output module then fuses and predicts the temperature mapping data and stress mapping data to output the coil monitoring results. This approach, by comparing the dynamic time-series average temperature and stress signal with a preset temperature and stress signal threshold, and by inputting the dynamic time-series average temperature and stress signal into a preset multimodal one-dimensional model for fault prediction, outputs the coil monitoring results. Compared to existing methods of scanning and monitoring wireless power transfer coils using underwater robots or manually determining whether wireless power transfer coils are faulty using monitoring instruments, this method can replace traditional manual judgment methods, promptly detect faults in wireless power transfer coils, and achieve better monitoring results. Furthermore, by sequentially comparing threshold judgments and model fault judgments based on the dynamic time-series average temperature and stress signal, the accuracy of the coil monitoring results is further improved.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0152] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0153] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An underwater wireless power transfer coil monitoring system, characterized in that, The system includes multiple grating sensors, optical fibers, wireless power transmission coils, an optical fiber demodulator, an insulating wrapping layer, and an online monitoring device; The optical fiber is wound around the wireless power transmission coil. Each of the grating sensors is equidistantly disposed on the surface of the wireless power transmission coil and interconnected with each other via the optical fiber. Both ends of the optical fiber are connected to the optical fiber demodulator, and each of the grating sensors, the optical fiber, and the wireless power transmission coil are disposed within the insulating wrapping layer; The fiber optic demodulator is embedded in the insulating wrapping layer and connected to the wireless power transmission coil. The fiber optic demodulator is communicatively connected to the online monitoring device; The grating sensor is used to acquire multiple dynamic time-series temperature stress signals of the wireless power transmission coil and transmit them to the online monitoring device through the fiber optic demodulator. The online monitoring device is used to perform averaging calculation on multiple dynamic time-series temperature stress signals, output the average dynamic time-series temperature stress signal and compare it with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multi-modal one-dimensional model for fault prediction and the coil monitoring result is output. The online monitoring device includes a data acquisition module and a data processing module; The data acquisition module is connected to the data processing module; The data acquisition module is used to transmit multiple dynamic time-series temperature stress signals received from the fiber optic demodulator to the data processing module. The data processing module is used to perform averaging calculation on multiple dynamic time-series temperature stress signals, output the average dynamic time-series temperature stress signal and compare it with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, the average dynamic time-series temperature stress signal is input to a preset multi-modal one-dimensional model for fault prediction and the coil monitoring result is output. The dynamic time-series temperature stress signal includes a dynamic time-series temperature signal and a dynamic time-series stress signal. The grating sensor includes a temperature grating sensor and a stress grating sensor; the optical fiber includes a temperature optical fiber and a stress optical fiber; The stress fiber is wound around the wireless power transmission coil along the length of the wireless power transmission coil, and the temperature fiber is spirally wound around the wireless power transmission coil. The temperature grating sensor and the stress grating sensor are disposed adjacent to each other on the surface of the wireless power transmission coil; Each of the temperature grating sensors is interconnected via the temperature optical fiber, and each of the stress grating sensors is interconnected via the stress optical fiber. Each of the temperature grating sensors, each of the stress grating sensors, the temperature optical fiber, the stress optical fiber, and the wireless power transmission coil are disposed within the insulating wrapping layer; Both ends of the temperature fiber and both ends of the stress fiber are connected to the fiber demodulator. The temperature grating sensor is used to acquire the dynamic time-series temperature signal of the wireless power transmission coil and transmit it to the data acquisition module through the fiber optic demodulator. The stress grating sensor is used to acquire the dynamic time-series stress signal of the wireless power transmission coil and transmit it to the data acquisition module through the fiber optic demodulator.

2. The underwater wireless power transfer coil monitoring system according to claim 1, characterized in that, The dynamic time-series temperature stress average signal includes a dynamic time-series temperature average signal and a dynamic time-series stress average signal; the preset temperature stress signal threshold includes a preset temperature signal threshold and a preset stress signal threshold; the preset multimodal one-dimensional model includes a signal processing module, a convolutional network, and an output module; the data processing module is specifically used for: The mean values ​​of each dynamic time-series temperature signal and each dynamic time-series stress signal are calculated to generate a dynamic time-series temperature mean signal and a dynamic time-series stress mean signal, which are then compared with preset temperature signal thresholds and preset stress signal thresholds. If the dynamic time-series average temperature signal is less than a preset temperature signal threshold and the dynamic time-series average stress signal is less than a preset stress signal threshold, then the signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series average temperature signal and the dynamic time-series average stress signal respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data. A convolutional network is used to perform data mapping on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data; The output module fuses and predicts temperature mapping data and stress mapping data to output coil monitoring results.

3. The underwater wireless power transfer coil monitoring system according to claim 2, characterized in that, The one-dimensional temperature sequence data includes Fourier temperature amplitude, Fourier temperature phase, first-order differential temperature signal, second-order differential temperature signal, and Hilbert transform temperature signal; the one-dimensional stress sequence data includes Fourier stress amplitude, Fourier stress phase, first-order differential stress signal, second-order differential stress signal, and Hilbert transform stress signal; the data processing procedure of the signal processing module is specifically as follows: The input dynamic time-series average temperature signal is subjected to Fourier transform, first-order derivative, second-order derivative and Hilbert transform respectively, and the corresponding Fourier transform temperature signal, first-order derivative temperature signal, second-order derivative temperature signal and Hilbert transform temperature signal are output. The input dynamic time-series mean stress signal is subjected to Fourier transform, first-order differential, second-order differential and Hilbert transform respectively, and the corresponding Fourier transform stress signal, first-order differential stress signal, second-order differential stress signal and Hilbert transform stress signal are output. Based on the Fourier transform temperature signal, determine the Fourier temperature amplitude and Fourier temperature phase. Based on the Fourier transform stress signal, the Fourier stress amplitude and Fourier stress phase are determined.

4. The underwater wireless power transfer coil monitoring system according to claim 2, characterized in that, The convolutional network comprises four cascaded convolutional layers, an adaptive max-pooling layer, and a fully connected layer; the data processing procedure of the convolutional network is as follows: Four cascaded convolutional layers are used to perform convolution operations on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively to generate one-dimensional temperature convolutional data and one-dimensional stress convolutional data. The one-dimensional temperature convolutional data and the one-dimensional stress convolutional data are pooled by an adaptive maximum pooling layer, and the one-dimensional temperature pooled data and the one-dimensional stress pooled data are output. The one-dimensional temperature pooling data and the one-dimensional stress pooling data are input into the fully connected layer for feature mapping to generate temperature mapping data and stress mapping data.

5. The underwater wireless power transfer coil monitoring system according to claim 2, characterized in that, The output module includes a data fusion layer and a classifier; the data processing procedure of the output module is as follows: The temperature mapping data and the stress mapping data are fused using a multi-parameter fusion algorithm through the data fusion layer to generate temperature-stress fused data. A classifier is used to classify and predict the temperature stress fusion data, and the coil monitoring results are output.

6. The underwater wireless power transfer coil monitoring system according to claim 1, characterized in that, The data processing module is also used for: When the coil monitoring results show an abnormality, a coil warning command is generated.

7. The underwater wireless power transfer coil monitoring system according to claim 6, characterized in that, The online monitoring device also includes an alarm module; The alarm module is connected to the data processing module; The alarm module is used to issue an alarm in response to the coil warning command sent by the data processing module.

8. A method for monitoring underwater wireless power transfer coils, applied to the underwater wireless power transfer coil monitoring system as described in claim 1, characterized in that, include: When multiple dynamic time-series temperature stress signals are received, the average value of each dynamic time-series temperature stress signal is calculated to generate a dynamic time-series temperature stress average signal, and compared with a preset temperature stress signal threshold. If the average dynamic time-series temperature stress signal is less than the preset temperature stress signal threshold, then the average dynamic time-series temperature stress signal is input to the preset multimodal one-dimensional model, wherein the preset multimodal one-dimensional model includes a signal processing module, a convolutional network and an output module, and the average dynamic time-series temperature stress signal includes the average dynamic time-series temperature signal and the average dynamic time-series stress signal. The signal processing module uses a time-frequency domain transformation algorithm to perform signal transformation on the dynamic time-series average temperature signal and the dynamic time-series average stress signal, respectively, and outputs one-dimensional temperature sequence data and one-dimensional stress sequence data. A convolutional network is used to perform data mapping on the one-dimensional temperature sequence data and the one-dimensional stress sequence data respectively, generating temperature mapping data and stress mapping data; The output module fuses and predicts the temperature mapping data and the stress mapping data to output the coil monitoring results.

Images