Icebreaking method based on combination of high-pressure gas and resonance device

Abstract

The invention relates to the field of polar region icebreaking, in particular to an icebreaking method based on the combination of high-pressure gas and a resonance device, and the method comprises the following steps: building an ice layer three-dimensional mechanical property model and a high-pressure gas layered injection channel, forming a gas cavity, and pre-judging the deviation trend of the inherent frequency of the ice layer; and dynamically matching the excitation frequency with the inherent frequency of the ice layer, starting the resonance device according to the matched excitation frequency, and adjusting the periodic fluctuation of the excitation frequency to strengthen the fatigue damage of the ice layer. By forming the air cavity, reducing the bearing capacity of the ice layer and improving the inherent frequency of the ice layer, the problem that an existing resonance device is difficult to carry out low-frequency excitation is solved, and by pre-judging the deviation direction and the change amplitude of the inherent frequency and by taking the deviation of the excitation frequency and the real-time inherent frequency as a target, the vibration excitation frequency is accurately measured. And determining an initial excitation frequency and an adjustment interval in combination with a deviation trend pre-judgment result, and realizing dynamic adjustment of the excitation frequency through a frequency deviation compensation amount.

Description

Technical Field

[0001] This invention relates to the field of polar icebreaking, and more specifically to an icebreaking method based on a combination of high-pressure gas and a resonance device. Background Technology

[0002] Current icebreaking technologies mainly include mechanical impact icebreaking, thermal icebreaking, and single resonance icebreaking, but all have significant technical bottlenecks. Mechanical impact icebreaking relies on rigid impact to break the ice layer, which not only consumes a lot of energy but also easily damages the operating equipment. Furthermore, it is difficult to accurately apply force to areas with weak mechanical properties in the ice layer, thus limiting the icebreaking efficiency. Thermal icebreaking, on the other hand, suffers from low energy utilization efficiency and poor controllability of the icebreaking range, making it unsuitable for large-scale rapid icebreaking needs.

[0003] Therefore, vibration-based icebreaking is more common. However, to induce resonance in sea ice, the required excitation frequency must be very low, usually below 1Hz. This poses a significant challenge to the design of high-power low-frequency exciters. Furthermore, the natural frequency of ice in the underwater environment is easily affected by water flow, water pressure, and uneven mechanical properties. Traditional technologies lack effective means to predict this shift trend, making it difficult to maintain a continuous match between the excitation frequency and the natural frequency of the ice. The resonance effect decays rapidly, and the strong damping characteristics of the water medium cause the vibration energy to dissipate quickly. A single resonance method is insufficient to achieve effective energy accumulation and is unlikely to quickly induce ice fatigue failure. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing an ice-breaking method based on a combination of high-pressure gas and a resonance device.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: an ice-breaking method based on a combination of high-pressure gas and a resonance device, the method comprising: Step 1: Deploy an underwater autonomous vehicle equipped with a multi-dimensional sensor array to the target ice-breaking area to collect data synchronously. Based on the collected data, build a three-dimensional mechanical property model of the ice layer and a high-pressure gas layered injection channel. High-pressure gas is injected through the layered injection channel to form a gas cavity. Step 2: Collect the vibration response signal of the ice layer under the air cavity support state, extract the dynamic modal parameters of the ice layer, and establish a coupling relationship model with the natural frequency change of the ice layer based on the obtained air cavity pressure change and volume change, so as to predict the shift trend of the natural frequency of the ice layer. Step 3: Based on the dynamic modal parameters and the predicted offset trend, the excitation frequency is dynamically matched with the natural frequency of the ice layer. The resonance device is activated according to the matched excitation frequency, and the periodic fluctuation of the excitation frequency is adjusted to enhance the fatigue failure of the ice layer. The crack propagation state and air cavity of the ice layer are continuously monitored, and high-pressure gas is added until the ice layer is completely broken and a stable ice-breaking channel is formed.

[0006] In a preferred embodiment, in step one, an underwater autonomous vehicle equipped with a multi-dimensional sensor array is deployed to the target ice-breaking area, and the multi-dimensional sensor array scans and detects the entire target ice-breaking area through a preset navigation path planning. After the autonomous underwater vehicle reaches the target ice-breaking area, it activates the multi-dimensional sensor array on board the autonomous underwater vehicle to acquire data including mechanical parameters and environmental parameters. After acquiring the collected data, the ice thickness distribution is used as the basic spatial coordinate reference. Mechanical parameters such as internal crack and bubble distribution data, elastic modulus data, and density data under the same coordinate are mapped and matched, and environmental parameters such as water flow velocity under ice, water pressure, and water temperature at the corresponding coordinate are synchronously associated to establish a three-dimensional associated coordinate dataset of spatial coordinates, mechanical parameters, and environmental parameters. Based on the spatial dimensional features of the three-dimensional associated coordinate dataset, an initial three-dimensional mesh framework is constructed. The mesh accuracy of the initial three-dimensional mesh framework is determined according to the resolution of the ice thickness distribution to ensure that the framework can accurately cover the entire spatial range of the target ice-breaking area. The mechanical and environmental parameters in the three-dimensional associated dataset are filled into the grid cells of the initial three-dimensional mesh frame according to their corresponding spatial coordinates. The filled mesh data is then fused and optimized using a data association analysis algorithm to eliminate data redundancy and conflicts, thereby generating a complete three-dimensional mechanical property model of the ice layer. The spatial coordinates of the mechanically weak areas of the ice layer, the initial range of the natural frequency of the ice layer, and the ice layer thickness distribution data are then extracted from the three-dimensional mechanical property model of the ice layer.

[0007] In a preferred embodiment, step one, after constructing a three-dimensional mechanical property model of the ice layer, involves planning the spatial layout of the layered injection channels, including the following specific steps: The three-dimensional mechanical property model of the ice layer is called to extract the ice layer thickness distribution data, spatial coordinates and range data of mechanically weak areas in the whole region. Abnormal data is removed through the central control unit to form a standardized thickness-spatial coordinate dataset. Based on a thickness-spatial coordinate dataset, thickness extrema and abrupt change points are determined through data traversal. Thickness extrema include the maximum and minimum thickness points. Thickness abrupt change points refer to points where the thickness difference between adjacent spatial coordinate points exceeds a preset threshold, which is determined by the standard deviation of the ice layer thickness data. In other words, the thickness fluctuates significantly with the thickness of adjacent grid cells. A spatial clustering algorithm is used to perform spatial clustering on the ice layer with thickness extrema as cluster centers and thickness abrupt change points as cluster boundaries. Combined with the extracted range data of mechanically weak regions, the mechanically weak regions are completely assigned to the same cluster unit, ensuring that the ice layer thickness characteristics and mechanical properties are consistent within the same cluster unit. The mechanically weak regions are treated as a whole, ultimately resulting in several cluster units. Each cluster unit is an independent injection layer. The spatial coordinate range and depth interval of each injection layer are output, completing the injection layer division. The spatial clustering algorithm refers to a basic algorithm for grouping based on the spatial distribution characteristics of data. In this application, it is used to group regions with similar thickness characteristics into one class.

[0008] In a preferred embodiment, step one further includes: calling the spatial coordinate range data of each injection layer, extracting the preset coordinate data of the channel corresponding to each injection layer, combining the ice thickness distribution data in the three-dimensional mechanical property model of the ice layer, determining the reference point and termination reference point of the channel corresponding to each injection layer, calculating the optimal path of each set of reference points and termination reference points through a laser positioning algorithm, and completing the construction of the channel corresponding to each injection layer through laser drilling of the vehicle, which serves as a high-pressure gas layered injection channel. The high-pressure gas is injected through the layered injection channel to form a gas cavity.

[0009] In a preferred embodiment, step two is based on the distribution location of the air cavity, the spatial coordinates and range data of the mechanically weak area, and coordinate alignment is performed. The spatial coordinate system of the three-dimensional mechanical property model of the ice layer is used as a unified reference. The excitation center set of the resonance device is determined by superimposing the spatial coordinates, and strain gauges and acceleration sensors are deployed in the excitation center set to collect vibration response signals, including strain signals generated by ice layer vibration and acceleration signals of ice layer vibration, in real time. Spectral analysis is performed on the vibration response signal to extract its frequency components, peak amplitude, and phase information, forming a hybrid vibration response signal. Based on the blind source separation algorithm, the mixed vibration response signal is whitened. By subtracting the signal mean, solving the covariance matrix and performing eigenvalue decomposition, the correlation between the various dimensions of the signal is eliminated, and the whitened vibration signal is obtained. The whitened vibration signal is then iteratively calculated using a nonlinear activation function to obtain several candidate modal signals after the initial separation. For each candidate mode signal, perform a lag product operation on the time series to obtain the autocorrelation value at different lag times. Take the maximum autocorrelation value as the autocorrelation coefficient to characterize the periodicity of the signal itself. Perform a cross product operation on the time series of any two candidate mode signals to obtain the cross-correlation value. Take the cross-correlation value with the largest absolute value as the cross-correlation coefficient to characterize the correlation between different candidate mode signals. Candidate mode signals with cross-correlation coefficients and autocorrelation coefficients exceeding the average value are retained. After preliminary decoupling, spectral analysis is performed on the retained candidate mode signals to extract the main frequency of each candidate mode signal, i.e., the frequency corresponding to the peak amplitude in the spectrum. The initial range of the natural frequency of the ice layer obtained in step one is called, and the main frequency of each candidate mode signal is matched with this range. Candidate mode signals whose main frequency falls within the initial range of the natural frequency are selected, thereby eliminating interference signals that exceed the range, such as high-frequency equipment noise and low-frequency water flow vibration signals. Finally, independent signals of each mode related to the vibration of the ice layer itself are obtained.

[0010] In a preferred embodiment, based on the separated modal independent signals, the main frequency of each modal independent signal is taken as the real-time natural frequency; The half-power bandwidth method is adopted to determine the frequency bandwidth corresponding to the half-power point by the spectrum diagram of the independent signal of each mode. The damping ratio is calculated by combining the real-time natural frequency. The calculation logic of the damping ratio is to divide the frequency bandwidth corresponding to the half-power point by twice the real-time natural frequency. Using the initial range data of the natural frequency of the ice layer, the real-time natural frequency is range-checked, and abnormal parameters that exceed the reasonable range are eliminated. The damping ratio is cross-checked with the mechanical parameters of the three-dimensional mechanical property model of the ice layer, and finally the dynamic modal parameter set is obtained. The damping ratio is verified by associating the elastic modulus and density data in the three-dimensional mechanical property model of the ice layer. The verification is completed by referring to the property relationship between the elastic modulus and density on ice and the damping ratio.

[0011] In a preferred embodiment, step two calculates the volume change by using real-time morphological data of the air cavity obtained by ultrasonic tomography and the volume reference value at the initial formation of the air cavity, and calculates the pressure change by comparing the real-time pressure value with the pressure reference value at the stable formation of the air cavity. The volume change and pressure change are aligned and adapted in time series to form a time series dataset of volume change and pressure change. The time series data of the time series dataset are sorted and aligned by timestamp, and duplicate time series points are removed. Then, the volume change column data and the pressure change column data are concatenated row by row to generate a three-dimensional input matrix containing timestamp, volume change and pressure change. Using the timestamps of the three-dimensional input matrix as the reference index, the dynamic modal parameter set is traversed and matched in time sequence to select the target data subset that completely corresponds to all timestamps of the three-dimensional input matrix. Based on the target data subset, the real-time natural frequency values ​​corresponding to each timestamp are extracted. Based on the reference value of the natural frequency of the ice layer when the air cavity is initially formed, the change in the natural frequency of the ice layer corresponding to each time sequence is obtained by difference operation, which is the real-time natural frequency minus the reference value of the natural frequency, forming a sequence of natural frequency changes that is completely aligned with the time sequence of the three-dimensional input matrix. The volume change column, pressure change column, and natural frequency change sequence of the three-dimensional input matrix are concatenated row by row to construct a four-dimensional structured air cavity parameter change-ice layer natural frequency change correlation dataset, which is composed of timestamp-volume change-pressure change-ice layer natural frequency change. Based on the feature dimensions of the correlation dataset, a general data-driven model suitable for small sample and nonlinear relationship fitting, such as BP neural network and support vector regression model, is selected as the basic model architecture. The correlation dataset is randomly divided into a validation set and a training set. A coupling relationship model is constructed through the regression model. Using a time window covering at least two complete fluctuation cycles as the predicted duration range, the coupling relationship model calculates the predicted value of the natural frequency of the ice layer corresponding to each time node through forward reasoning. The predicted values ​​of the natural frequencies of all time nodes are extracted and arranged in ascending order by timestamp to form a sequence of natural frequency changes of the ice layer within the preset duration. The rate of change of the natural frequency of each time node is calculated by the first derivative, and the difference of the natural frequencies of adjacent time nodes is calculated to obtain the change amplitude of each time period. A complete prediction curve of the natural frequency offset trend is generated as the offset trend.

[0012] In a preferred embodiment, step three is based on the real-time natural frequency and the natural frequency offset trend prediction curve in the dynamic modal parameter set. The least squares fitting frequency matching algorithm is used to construct an optimization objective function with the goal of minimizing the deviation between the excitation frequency and the real-time natural frequency. After obtaining the optimal solution representing the initial excitation frequency, the initial excitation frequency and adjustment range of the resonance device are formed according to the initial offset direction of the natural frequency offset trend prediction curve. Each resonance device is transported to the corresponding excitation center one by one. In some other specific embodiments, the mechanical locking mechanism fixes the device to the ice surface to ensure that the device does not move during the excitation process. The predicted natural frequency of the ice layer output by the coupling relationship model is calculated by subtracting the real-time natural frequency from the dynamic modal parameter set to obtain the basic deviation. The natural frequency change rate and the change amplitude of each time period are combined for weighted correction to generate the dynamically adjusted frequency deviation compensation. The excitation frequency output of the resonance device is adjusted based on the frequency deviation compensation amount. The output excitation frequency always tracks the real-time natural frequency deviation of the ice layer. The damping ratio data in the dynamic modal parameters is called to back-determine the fluctuation amplitude and fluctuation period of the excitation frequency. The periodic fluctuation of the excitation frequency is adjusted to conform to the determined fluctuation amplitude and fluctuation period, thereby strengthening the fatigue failure of the ice layer.

[0013] In a preferred embodiment, after calling the damping ratio data in the dynamic modal parameters, step three synchronously extracts the real-time natural frequency from the dynamic modal parameter set and combines it with the mechanical parameters of the three-dimensional mechanical property model of the ice layer to construct an associated parameter group; The product of half-power bandwidth and damping ratio in the associated parameter group is divided by the product of elastic modulus and density. The resulting value is used as the initial value of the fluctuation amplitude. The peak amplitude of strain signal and peak amplitude of acceleration signal in vibration response signal are used as verification benchmarks. The theoretical vibration amplitude corresponding to the initial value of fluctuation amplitude is compared with the actual collected peak amplitude to calculate the deviation. A correction coefficient is introduced based on the deviation to obtain the quantitative correlation between fluctuation amplitude and damping ratio. That is, the product of half-power bandwidth and damping ratio is divided by the product of elastic modulus and density, and the resulting value is then multiplied by the correction coefficient to obtain the fluctuation amplitude. Based on the principle of vibration energy accumulation, within one cycle, the energy input corresponding to the fluctuation amplitude can offset the energy dissipation caused by the damping ratio. The product of 2π and the real-time natural frequency, divided by the quotient of the product of the damping ratio and the fluctuation amplitude, is taken as the fluctuation period.

[0014] In a preferred embodiment, in step three, after the ice layer breaks, effective passage cross-section data is collected, and the collection time is set to be no less than one complete fluctuation cycle, which serves as the basis for judging whether the ice layer is completely broken, i.e., the minimum value of the effective passage cross-section data exceeds the preset threshold for the complete breakage of the ice layer.

[0015] The beneficial effects of this invention are as follows: by forming an air cavity, the bearing capacity of the ice layer is reduced, and on the other hand, the natural frequency of the ice layer is increased, solving the problem that current resonance devices are difficult to excite at low frequencies. In addition, by predicting the offset direction and change amplitude of the natural frequency, and aiming to minimize the deviation between the excitation frequency and the real-time natural frequency, the initial excitation frequency and adjustment range are determined by combining the offset trend prediction results. The excitation frequency is dynamically adjusted through the frequency deviation compensation, ensuring that it always tracks the natural frequency of the ice layer, achieving the purpose of adjusting the excitation frequency output by the resonance device in real time, and always being in resonance with the enhanced natural frequency of the ice layer. At this time, the vibration amplitude of the ice layer will reach the maximum and the energy transfer efficiency will be the highest. Attached Figure Description

[0016] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] As attached Figure 1 As shown, this embodiment provides an ice-breaking method based on a combination of high-pressure gas and a resonance device, comprising the following steps: Step 1: Deploy an underwater autonomous vehicle equipped with a multi-dimensional sensor array to the target ice-breaking area to collect data synchronously. Based on the collected data, build a three-dimensional mechanical property model of the ice layer and a high-pressure gas layered injection channel. High-pressure gas is injected through the layered injection channel to form a gas cavity. This step consists of the following sub-steps: Step 1-1: Deploy an underwater autonomous vehicle equipped with a multi-dimensional sensor array to the target ice-breaking area. Through preset navigation path planning, the multi-dimensional sensor array scans and detects the entire target ice-breaking area. Steps 1-2: After the autonomous underwater vehicle reaches the target ice-breaking area, activate the multi-dimensional sensor array on board the autonomous underwater vehicle to acquire data including mechanical parameters and environmental parameters. In some specific practical applications, the underwater autonomous vehicle travels at a constant speed along a planned path. The ultrasonic tomography module continuously emits ultrasonic signals and receives reflected signals from inside the ice layer to obtain data on the distribution of internal cracks and bubbles. The electromagnetic induction thickness detection module measures the thickness data of different locations in the ice layer in real time through the principle of electromagnetic induction, forming ice layer thickness distribution data. The elastic wave propagation velocity test module emits elastic waves into the ice layer, records the propagation time and path of the elastic waves in the ice layer, and calculates and obtains core mechanical parameters such as elastic modulus and density.

[0019] In addition to the more common polar data collection methods mentioned above, it also includes the synchronous collection of environmental parameters such as subglacial water flow velocity, water pressure, and water temperature. Steps 1-3: After acquiring the collected data, using the ice layer thickness distribution as the basic spatial coordinate reference, the mechanical parameters under the same coordinate, including the distribution data of internal cracks and bubbles, elastic modulus data, density data, etc., are mapped and matched, and the corresponding environmental parameters, including the water flow velocity under the ice, water pressure, water temperature, etc., are synchronously associated to establish a three-dimensional associated coordinate dataset of spatial coordinates-mechanical parameters-environmental parameters. Steps 1-4: Based on the spatial dimensional features of the three-dimensional associated coordinate dataset, an initial three-dimensional mesh framework is constructed. The mesh accuracy of the initial three-dimensional mesh framework is determined according to the resolution of the ice thickness distribution to ensure that the framework can accurately cover the entire spatial range of the target ice-breaking area. Steps 1-5: Fill the grid cells of the initial three-dimensional mesh frame with the various mechanical and environmental parameters in the three-dimensional associated dataset according to their corresponding spatial coordinates. After the data is filled, the mesh data is fused and optimized by the data association analysis algorithm to eliminate data redundancy and conflicts, and a complete three-dimensional mechanical property model of the ice layer is generated. The spatial coordinates of the mechanically weak areas of the ice layer, the initial range of the natural frequency of the ice layer, and the ice layer thickness distribution data are extracted from the three-dimensional mechanical property model of the ice layer.

[0020] The initial range of natural frequencies of the ice layer is obtained based on environmental and mechanical parameters, and there are two common methods for obtaining it: Determining the initial range of natural frequencies under waterless conditions: When there is no water beneath sea ice (i.e., the ice layer is directly suspended in the air or on land), the vibration of sea ice can generally be regarded as a problem of vibration of an elastic thin plate. The natural frequencies are mainly determined by factors such as the thickness, elastic modulus, and density of the ice layer. The natural frequencies of sea ice can be expressed by the following formula: in: It is the natural frequency (Hz). It is the thickness of the ice layer (in meters). It is the elastic modulus (Pa) of the ice layer. The density of sea ice (kg / m³) 3 ); Determining the initial range of natural frequencies in ice-water conditions: When there is water beneath sea ice, the buoyancy and support of the water on the ice layer affect the natural frequencies. In this case, the natural frequencies depend not only on the properties of the ice layer itself, but also on factors such as the depth and density of the water, and the interaction between the ice and water interfaces.

[0021] In this case, a coupled vibration model of the sea ice-water system can be used. A common approximation is: in: It is the natural frequency (Hz). It is the thickness of the ice layer (in meters). It is the elastic modulus (Pa) of the ice layer. The density of sea ice (kg / m³) 3 ), It is the density of water (kg / m³) 3 ).

[0022] After constructing a three-dimensional mechanical property model of the ice layer, the spatial layout of the layered injection channels needs to be planned, which involves the following steps: Steps 1-6: Call the three-dimensional mechanical property model of the ice layer, extract the ice layer thickness distribution data of the whole region, the spatial coordinates and range data of the mechanically weak areas, and remove abnormal data through the central control unit to form a standardized thickness-spatial coordinate dataset. Steps 1-7: Based on the thickness-spatial coordinate dataset, thickness extreme points and thickness mutation points are determined through data traversal. Thickness extreme points include the maximum thickness point and the minimum thickness point. Thickness mutation points refer to points where the thickness difference between adjacent spatial coordinate points exceeds a preset threshold, and the threshold is determined by the standard deviation of the ice layer thickness data, i.e., points where the thickness fluctuates significantly with the thickness of adjacent grid cells. Using a spatial clustering algorithm, the ice layer is spatially clustered with the thickness extreme points as cluster centers and the thickness mutation points as cluster boundaries. Combined with the extracted range data of mechanically weak areas, the mechanically weak areas are completely assigned to the same cluster unit, ensuring that the ice layer thickness characteristics and mechanical properties are consistent within the same cluster unit. The mechanically weak areas are treated as a whole, and finally several cluster units are obtained. Each cluster unit is an independent injection layer. The spatial coordinate range and depth interval of each injection layer are output to complete the injection layer division. The spatial clustering algorithm refers to a basic algorithm for grouping based on the spatial distribution characteristics of data. In this application, it is used to group areas with similar thickness characteristics into one class.

[0023] Steps 1-8: Call the spatial coordinate range data of each injection layer, extract the preset coordinate data of the channel corresponding to each injection layer, combine the ice thickness distribution data in the three-dimensional mechanical property model of the ice layer, determine the reference point and termination reference point of the channel corresponding to each injection layer, calculate the optimal path of each set of reference points and termination reference points through the laser positioning algorithm, and complete the construction of the channel corresponding to each injection layer through laser drilling of the vehicle, which serves as the high-pressure gas layered injection channel. The high-pressure gas is injected through the layered injection channel to form a gas cavity.

[0024] The purpose of forming the air cavity in this application is to change the ice layer from pure ice or ice-water coupling to ice-air cavity support, aiming to reduce the load-bearing capacity of the ice layer and increase its natural frequency. In a real marine environment, the excitation frequency required to induce resonance in sea ice must be very low, typically below 1 Hz, which poses a significant challenge to the design of high-power low-frequency resonance devices. Furthermore, water has extremely strong damping properties, meaning energy dissipates rapidly. To maintain a sufficiently large vibration amplitude to achieve fatigue failure, a much greater energy input is required than in air.

[0025] Therefore, this application utilizes the principle of resonance to induce fatigue failure of the ice layer through energy accumulation using a small periodic load. The principle of fatigue failure lies in the fact that when a material is subjected to cyclic loads (repeated stress) far below its ultimate strength, its performance gradually deteriorates and eventually fractures. Ice, as a material, also exhibits fatigue characteristics. If a periodic force (such as vibration) is applied to the ice layer, even if this force itself is insufficient to crush the ice layer in one go, stress concentration will occur at microscopic defects (such as cracks and bubbles) inside the ice, leading to the gradual initiation and propagation of microcracks. Therefore, the resonance effect is the key to the amplifier and efficiency of this application. After the air cavity is built, the natural frequency of the ice layer changes. How to dynamically match the natural frequency of the ice layer and the resulting offset trend is discussed below.

[0026] In some other specific embodiments, the preparation and purification of dry inert gas are completed by starting the high-pressure gas generation system, while the flow control valves and pressure sensors of each injection pipe are preheated to ensure stable operation of the equipment; According to the planned timing, adjustment commands are sent to the flow control valves of the injection pipes at each level, and injection is carried out according to the initial injection parameters. During the injection process, the pressure sensors at the ports of each injection pipe collect the pressure data of the gas chamber in real time. Through the pressure-flow closed-loop control algorithm, the real-time pressure data is compared with the pressure reference control point. If the real-time pressure is lower than the reference value, the output command increases the opening of the corresponding flow control valve. If the real-time pressure is higher than the reference value, the output command decreases the valve opening. The ultrasonic tomography module is reused to scan the key monitoring points of gas cavity formation in each layer in real time to obtain data on gas diffusion range and gas cavity morphology. By comparing the air cavity morphology data obtained from ultrasonic scanning with the preset air cavity range, if the air cavity covers all key monitoring points and has no obvious diffusion blind area, the air cavity is determined to be fully formed; if there are uncovered areas, the air injection parameters of the corresponding area injection pipe are adjusted.

[0027] The ice layer's support state changes are monitored using a three-dimensional mechanical property model. When the ice layer completely transforms from ice-water coupling support to ice-air cavity support, and the air cavity pressure at each layer stabilizes at the benchmark control point for more than a preset time, the dynamic formation of the air cavity is determined to be complete.

[0028] Step 2: Collect the vibration response signal of the ice layer under the air cavity support state, extract the dynamic modal parameters of the ice layer, and establish a coupling relationship model with the natural frequency change of the ice layer based on the obtained air cavity pressure change and volume change, so as to predict the shift trend of the natural frequency of the ice layer. This step consists of the following sub-steps: Step 2-1: Based on the distribution location of the air cavity, the spatial coordinates and range data of the mechanically weak area, coordinate alignment is performed. Using the spatial coordinate system of the three-dimensional mechanical property model of the ice layer as a unified benchmark, the excitation center set of the resonance device is determined by superimposing the spatial coordinates. Strain gauges and acceleration sensors are deployed in the excitation center set to collect vibration response signals, including strain signals generated by ice layer vibration and acceleration signals of ice layer vibration, in real time. Step 2-2: Perform spectral analysis on the vibration response signal to extract the frequency components, peak amplitude, and phase information of the vibration response signal, forming a hybrid vibration response signal; Steps 2-3: Based on the blind source separation algorithm, the mixed vibration response signal is whitened. By subtracting the signal mean, solving the covariance matrix and performing eigenvalue decomposition, the correlation between the various dimensions of the signal is eliminated, and the whitened vibration signal is obtained. The whitened vibration signal is then iteratively calculated using a nonlinear activation function to obtain several candidate modal signals after the initial separation. Steps 2-4: Perform lag product operation on the time series of each candidate mode signal to obtain the autocorrelation value at different lag times. Take the maximum autocorrelation value as the autocorrelation coefficient to characterize the periodicity of the signal itself. Perform cross product operation on the time series of any two candidate mode signals to obtain the cross-correlation value. Take the cross-correlation value with the largest absolute value as the cross-correlation coefficient to characterize the correlation between different candidate mode signals. Steps 2-5: Retain candidate mode signals whose cross-correlation coefficients and autocorrelation coefficients exceed the average mean. After completing the initial decoupling, perform spectral analysis on the retained candidate mode signals to extract the main frequency of each candidate mode signal, i.e., the frequency corresponding to the peak amplitude in the spectrum. Call the initial range of the ice layer's natural frequency obtained in Step 1 and match the main frequency of each candidate mode signal with this range. Filter out candidate mode signals whose main frequency falls within the initial range of the natural frequency, thereby eliminating interference signals that exceed the range, such as high-frequency equipment noise and low-frequency water flow vibration signals. Finally, obtain the independent signals of each mode related to the vibration of the ice layer itself. Steps 2-6: Based on the separated modal independent signals, take the main frequency of each modal independent signal as the real-time natural frequency; and use the half-power bandwidth method to determine the frequency bandwidth corresponding to the half-power point through the spectrum diagram of each modal independent signal, and calculate the damping ratio in combination with the real-time natural frequency. The calculation logic of the damping ratio is to divide the frequency bandwidth corresponding to the half-power point by twice the real-time natural frequency. Steps 2-7: Using the initial range data of the ice layer's natural frequency, perform range verification on the real-time natural frequency, eliminate abnormal parameters that exceed the reasonable range, and cross-validate the damping ratio with the mechanical parameters of the three-dimensional mechanical property model of the ice layer to finally obtain the dynamic modal parameter set. Among them, the damping ratio is verified by associating the elastic modulus and density data in the three-dimensional mechanical property model of the ice layer. The verification is completed by referring to the property relationship between the elastic modulus and density on ice and the damping ratio. Steps 2-8: The volume change is calculated by comparing the real-time morphological data of the air cavity obtained by ultrasonic tomography with the volume reference value when the air cavity is initially formed. The pressure change is calculated by comparing the real-time pressure value with the pressure reference value when the air cavity is stably formed. Steps 2-9: Perform time-series alignment and format adaptation on the volume change and pressure change to form a time-series dataset of volume change and pressure change. Sort and align the time-series data of the time-series dataset by timestamp, remove duplicate time-series points, and then concatenate the volume change column data and the pressure change column data row by row to generate a three-dimensional input matrix containing timestamps, volume change and pressure change. Steps 2-10: Using the timestamps of the three-dimensional input matrix as the reference index, perform time-series traversal matching on the dynamic modal parameter set, filter out the target data subset that completely corresponds to all timestamps of the three-dimensional input matrix, extract the real-time natural frequency values ​​corresponding to each timestamp based on the target data subset, and use the reference value of the natural frequency of the ice layer when the air cavity is initially formed as the basis, through difference operation, which refers to the real-time natural frequency minus the reference value of the natural frequency, obtain the change in the natural frequency of the ice layer corresponding to each time series, forming a sequence of natural frequency changes that is completely aligned with the time series of the three-dimensional input matrix; Step 2-11: Concatenate the volume change column, pressure change column, and natural frequency change sequence of the three-dimensional input matrix row by row to construct a four-dimensional structured air cavity parameter change-ice layer natural frequency change related dataset of timestamp-volume change-pressure change-ice layer natural frequency change. Based on the feature dimensions of the related dataset, select a general data-driven model suitable for small sample and nonlinear relationship fitting, such as BP neural network or support vector regression model, as the basic model architecture. Randomly divide the related dataset into a validation set and a training set, and construct a coupling relationship model through a regression model. Step 2-12: Set the time window covering at least two complete fluctuation cycles as the prediction duration range. The coupling relationship model calculates the predicted value of the natural frequency of the ice layer corresponding to each time node through forward reasoning. Extract the predicted value of the natural frequency of all time nodes, arrange them in ascending order of timestamp to form the natural frequency change sequence of the ice layer within the preset duration, calculate the rate of change of the natural frequency of each time node through the first derivative, and calculate the natural frequency difference between adjacent time nodes to obtain the change amplitude of each time period. Generate a complete natural frequency offset trend prediction curve as the offset trend.

[0029] The natural frequency change rate of each time node is calculated using the first derivative. When the change rate is positive, it is determined to be a positive offset, and when it is negative, it is determined to be a negative offset, thus clarifying the direction of change of the natural frequency.

[0030] Step 3: Based on the dynamic modal parameters and the predicted offset trend, the excitation frequency is dynamically matched with the natural frequency of the ice layer. The resonance device is activated according to the matched excitation frequency, and the periodic fluctuation of the excitation frequency is adjusted to enhance the fatigue failure of the ice layer. The crack propagation state and air cavity of the ice layer are continuously monitored, and high-pressure gas is added until the ice layer is completely broken and a stable ice-breaking channel is formed.

[0031] This step consists of the following sub-steps: Step 3-1: Based on the real-time natural frequency and the natural frequency offset trend prediction curve in the dynamic modal parameter set, the least squares fitting frequency matching algorithm is used to construct an optimization objective function with the goal of minimizing the deviation between the excitation frequency and the real-time natural frequency. After obtaining the optimal solution representing the initial excitation frequency, the initial excitation frequency and adjustment range of the resonance device are formed according to the initial offset direction of the natural frequency offset trend prediction curve. Each resonance device is transported to the corresponding excitation center one by one. In some other specific embodiments, the mechanical locking mechanism fixes the device to the ice surface to ensure that the device does not move during the excitation process. Step 3-2: Perform difference calculation between the predicted natural frequency of the ice layer output by the coupling relationship model and the real-time natural frequency in the dynamic modal parameter set to obtain the basic deviation. Combine the natural frequency change rate and the change amplitude of each time period for weighted correction to generate the dynamically adjusted frequency deviation compensation amount. In summary, the rate of change of the natural frequency and the amplitude of change in each time period are used as core quantitative indicators. They are obtained by extracting the predicted value of the natural frequency of the ice layer corresponding to each time node based on the natural frequency shift trend prediction curve, and obtaining the instantaneous rate of change of each node through the first derivative calculation. The absolute value of the rate of change is taken as the core quantitative indicator, which intuitively reflects the speed of the natural frequency shift. At the same time, with a preset sampling period as the time interval, the difference between the predicted values ​​of the natural frequency of two adjacent time nodes is calculated. Similarly, the absolute value of the difference is taken as the quantitative indicator of the amplitude of change, which represents the size of the span of the natural frequency shift. Statistical analysis was conducted on the absolute values ​​of the rate of change and the absolute values ​​of the magnitude of change for all time series nodes to identify the maximum and minimum values ​​of the two indicators. Linear normalization was used to map the absolute values ​​of the rate of change and the absolute values ​​of the magnitude of change for each time series node to the standard interval [0,1]. The weighting coefficients for the rate of change are calculated by using a standardized rate of change index and preset linear fitting coefficients, which include slope coefficients and intercept coefficients. The weighting coefficients for the magnitude of change are calculated by using a standardized magnitude of change index and another set of preset linear fitting coefficients. These linear fitting coefficients are obtained through extensive simulation experiments of ice layer natural frequency shift and training with actual icebreaking data. When the standardized rate of change index reaches its maximum value of 1, the corresponding weighting coefficient reaches the preset upper limit value; when the standardized magnitude of change index reaches its maximum value of 1, its corresponding weighting coefficient also reaches the preset upper limit value, thus completing the weighting correction. Step 3-3: Adjust the excitation frequency output of the resonance device based on the frequency deviation compensation amount. The output excitation frequency always tracks the real-time natural frequency deviation of the ice layer. Call the damping ratio data in the dynamic modal parameters to back-determine the fluctuation amplitude and fluctuation period of the excitation frequency. Adjust the periodic fluctuation of the excitation frequency to conform to the determined fluctuation amplitude and fluctuation period to enhance the fatigue failure of the ice layer.

[0032] Steps 3-4: After calling the damping ratio data in the dynamic modal parameters, the real-time natural frequencies are extracted from the dynamic modal parameter set and combined with the mechanical parameters of the three-dimensional mechanical property model of the ice layer to construct a set of related parameters. Based on the principle of energy conservation, the positive correlation between damping ratio and energy dissipation is clarified, that is, for every unit increase in damping ratio, a corresponding proportion of vibration energy needs to be added to maintain the effective vibration amplitude. Steps 3-5: Divide the product of half-power bandwidth and damping ratio in the associated parameter group by the product of elastic modulus and density, and use the resulting value as the initial value of the fluctuation amplitude. Use the peak amplitude of strain signal and the peak amplitude of acceleration signal in the vibration response signal as the verification benchmark. Compare the theoretical vibration amplitude corresponding to the initial value of the fluctuation amplitude with the actual collected peak amplitude to calculate the deviation. Based on the deviation, introduce a correction coefficient to obtain the quantitative correlation between the fluctuation amplitude and the damping ratio. That is, divide the product of half-power bandwidth and damping ratio by the product of elastic modulus and density, and then multiply the resulting value by the correction coefficient to obtain the fluctuation amplitude. Steps 3-6: Based on the principle of vibration energy accumulation, within one cycle, the energy input corresponding to the fluctuation amplitude can offset the energy dissipation caused by the damping ratio. The product of 2π and the real-time natural frequency is divided by the quotient of the product of the damping ratio and the fluctuation amplitude to obtain the fluctuation period. Therefore, this calculation logic ensures that the fluctuation period can meet the core requirement that the energy input corresponding to the fluctuation amplitude within a single cycle can exactly offset the energy dissipation caused by the damping ratio. At the same time, it forms a dynamic synergy with the real-time natural frequency of the ice layer, the energy dissipation state, and the fluctuation amplitude, avoiding energy superposition failure caused by period mismatch, and ensuring the continuous accumulation of vibration energy to strengthen the fatigue damage of the ice layer.

[0033] Steps 3-7: After the ice breaks, collect effective passage cross-section data and set the collection time to be no less than one complete fluctuation cycle as the basis for judging whether the ice is completely broken, that is, the minimum value of the effective passage cross-section data exceeds the preset threshold for the complete breakage of the ice.

Claims

1. An ice-breaking method based on a combination of high-pressure gas and a resonance device, characterized in that, The method includes: Step 1: Deploy an underwater autonomous vehicle equipped with a multi-dimensional sensor array to the target ice-breaking area to collect data synchronously. Based on the collected data, build a three-dimensional mechanical property model of the ice layer and a high-pressure gas layered injection channel. High-pressure gas is injected through the layered injection channel to form a gas cavity. Step 2: Collect the vibration response signal of the ice layer under the air cavity support state, extract the dynamic modal parameters of the ice layer, and establish a coupling relationship model with the natural frequency change of the ice layer based on the obtained air cavity pressure change and volume change, so as to predict the shift trend of the natural frequency of the ice layer. Step 3: Based on the dynamic modal parameters and the predicted offset trend, the excitation frequency is dynamically matched with the natural frequency of the ice layer. The resonance device is activated according to the matched excitation frequency, and the periodic fluctuation of the excitation frequency is adjusted to enhance the fatigue failure of the ice layer. The crack propagation state and air cavity of the ice layer are continuously monitored, and high-pressure gas is added until the ice layer is completely broken and a stable ice-breaking channel is formed.

2. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 1, characterized in that, In step one, by deploying an underwater autonomous vehicle equipped with a multi-dimensional sensor array to the target ice-breaking area, the multi-dimensional sensor array scans and detects the entire target ice-breaking area through a preset navigation path planning. After the autonomous underwater vehicle reaches the target ice-breaking area, it activates the multi-dimensional sensor array on board the autonomous underwater vehicle to acquire data including mechanical parameters and environmental parameters. After acquiring the collected data, the mechanical parameters under the same coordinate system are mapped and matched based on the ice thickness distribution as the basic spatial coordinate reference, and the environmental parameters at the corresponding coordinates are synchronously associated to establish a three-dimensional associated coordinate dataset. Based on the spatial dimensional characteristics of the three-dimensional associated coordinate dataset, an initial three-dimensional mesh framework is constructed. The mesh accuracy of the initial three-dimensional mesh framework is determined according to the resolution of the ice layer thickness distribution. The mechanical and environmental parameters in the three-dimensional associated dataset are filled into the grid cells of the initial three-dimensional mesh frame according to their corresponding spatial coordinates. The filled mesh data is then fused and optimized using a data association analysis algorithm to eliminate data redundancy and conflicts, thereby generating a complete three-dimensional mechanical property model of the ice layer. The spatial coordinates of the mechanically weak areas of the ice layer, the initial range of the natural frequency of the ice layer, and the ice layer thickness distribution data are then extracted from the three-dimensional mechanical property model of the ice layer.

3. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 1, characterized in that, Step one, after building a three-dimensional mechanical property model of the ice layer, involves planning the spatial layout of the layered injection channels, including the following specific steps: The three-dimensional mechanical property model of the ice layer is called to extract the ice layer thickness distribution data, spatial coordinates and range data of mechanically weak areas in the whole region. Abnormal data is removed through the central control unit to form a standardized thickness-spatial coordinate dataset. Based on the thickness-spatial coordinate dataset, thickness extrema and abrupt change points are determined through data traversal. Spatial clustering algorithm is used to perform spatial clustering on the ice layer with thickness extrema as cluster centers and thickness abrupt change points as cluster boundaries. Combined with the extracted range data of mechanically weak areas, the mechanically weak areas are completely assigned to the same cluster unit, resulting in several cluster units. Each cluster unit is an independent injection layer. The spatial coordinate range and depth interval of each injection layer are output, thus completing the injection layer division.

4. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 3, characterized in that, Step one further includes: calling the spatial coordinate range data of each injection layer, extracting the preset coordinate data of the channel corresponding to each injection layer, combining the ice thickness distribution data in the three-dimensional mechanical property model of the ice layer, determining the reference point and termination reference point of the channel corresponding to each injection layer, calculating the optimal path of each set of reference points and termination reference points through the laser positioning algorithm, and completing the construction of the channel corresponding to each injection layer through laser drilling of the aircraft, which serves as a high-pressure gas layered injection channel. The high-pressure gas is injected through the layered injection channel to form a gas cavity.

5. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 2, characterized in that, Step two involves aligning the coordinates based on the distribution location of the air cavity and the spatial coordinates and range data of the mechanically weak areas. The excitation center set of the resonance device is determined by superimposing the spatial coordinates, and strain gauges and acceleration sensors are deployed at the excitation center set to collect vibration response signals, including strain signals generated by ice layer vibration and acceleration signals of ice layer vibration, in real time. Spectral analysis is performed on the vibration response signal to extract its frequency components, peak amplitude, and phase information, forming a hybrid vibration response signal. Based on the blind source separation algorithm, the mixed vibration response signal is whitened to obtain the whitened vibration signal. The whitened vibration signal is then iteratively calculated using a nonlinear activation function to obtain several candidate mode signals after the initial separation. For each candidate mode signal time series, perform lag product operation to obtain autocorrelation values ​​at different lag times, and take the maximum autocorrelation value as the autocorrelation coefficient. Perform cross product operation on the time series of any two candidate mode signals to obtain cross-correlation values, and take the cross-correlation value with the largest absolute value as the cross-correlation coefficient. Candidate modal signals with cross-correlation coefficients and autocorrelation coefficients exceeding the average value are retained. After preliminary decoupling, spectral analysis is performed on the retained candidate modal signals to extract the main frequency of each candidate modal signal. The initial range of the natural frequencies of the ice layer obtained in step one is called, and the main frequency of each candidate modal signal is matched with the initial range of the natural frequencies of the ice layer. Candidate modal signals whose main frequencies fall within the initial range of the natural frequencies are selected, and finally, independent signals of each mode related to the vibration of the ice layer are obtained.

6. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 5, characterized in that, Based on the separated modal independent signals, the main frequency of each modal independent signal is taken as the real-time natural frequency; The half-power bandwidth method is adopted to determine the frequency bandwidth corresponding to the half-power point by using the spectrum diagram of the independent signal of each mode, and the damping ratio is calculated by combining the real-time natural frequency. Using the initial range data of the ice layer's natural frequency, the real-time natural frequency is range-checked, abnormal parameters that exceed the reasonable range are eliminated, and the damping ratio is cross-checked with the mechanical parameters of the three-dimensional mechanical property model of the ice layer, finally obtaining the dynamic modal parameter set.

7. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 1, characterized in that, Step 2 calculates the volume change based on the real-time morphological data of the air cavity and the volume reference value when the air cavity is initially formed, and calculates the pressure change by comparing the real-time pressure value with the pressure reference value when the air cavity is stably formed. The volume change and pressure change are aligned and adapted in time series to form a time series dataset of volume change and pressure change. The time series data of the time series dataset are sorted and aligned by timestamp, and duplicate time series points are removed. Then, the volume change column data and the pressure change column data are concatenated row by row to generate a three-dimensional input matrix containing timestamp, volume change and pressure change. Using the timestamps of the three-dimensional input matrix as the reference index, the dynamic modal parameter set is traversed and matched in time sequence to select the target data subset that completely corresponds to all timestamps of the three-dimensional input matrix. Based on the target data subset, the real-time natural frequency values ​​corresponding to each timestamp are extracted. Based on the reference value of the natural frequency of the ice layer when the air cavity is initially formed, the change in the natural frequency of the ice layer corresponding to each time sequence is obtained through difference calculation, forming a sequence of natural frequency changes that is completely aligned with the time sequence of the three-dimensional input matrix. The volume change, pressure change and natural frequency change sequences of the three-dimensional input matrix are concatenated row by row to construct a four-dimensional structured associated dataset. Based on the feature dimensions of the associated dataset, the dataset is randomly divided into a validation set and a training set. A coupling relationship model is constructed through a regression model. Using a time window covering at least two complete fluctuation cycles as the predicted duration range, the coupling relationship model calculates the predicted value of the natural frequency of the ice layer corresponding to each time node through forward reasoning. The predicted values ​​of the natural frequencies of all time nodes are extracted and arranged in ascending order by timestamp to form a sequence of natural frequency changes of the ice layer within the preset duration. The rate of change of the natural frequency of each time node is calculated by the first derivative, and the difference of the natural frequencies of adjacent time nodes is calculated to obtain the change amplitude of each time period. A complete prediction curve of the natural frequency offset trend is generated as the offset trend.

8. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 1, characterized in that, Step three is based on the real-time natural frequency and the natural frequency offset trend prediction curve in the dynamic modal parameter set. The least squares fitting frequency matching algorithm is used to construct an optimization objective function with the goal of minimizing the deviation between the excitation frequency and the real-time natural frequency. After obtaining the optimal solution representing the initial excitation frequency, the initial excitation frequency and adjustment range of the resonance device are formed according to the initial offset direction of the natural frequency offset trend prediction curve. The predicted natural frequency of the ice layer output by the coupling relationship model is calculated by subtracting the real-time natural frequency from the dynamic modal parameter set to obtain the basic deviation. The natural frequency change rate and the change amplitude of each time period are combined for weighted correction to generate the dynamically adjusted frequency deviation compensation. The excitation frequency output of the resonance device is adjusted based on the frequency deviation compensation amount. The output excitation frequency always tracks the real-time natural frequency deviation of the ice layer. The damping ratio data in the dynamic modal parameters is called to back-determine the fluctuation amplitude and fluctuation period of the excitation frequency. The periodic fluctuation of the excitation frequency is adjusted to meet the determined fluctuation amplitude and fluctuation period, thereby strengthening the fatigue failure of the ice layer.

9. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 8, characterized in that, In step three, after calling the damping ratio data in the dynamic modal parameters, the real-time natural frequency is extracted from the dynamic modal parameter set and combined with the mechanical parameters of the three-dimensional mechanical property model of the ice layer to construct a related parameter group. The product of half-power bandwidth and damping ratio in the associated parameter group is divided by the product of elastic modulus and density. The resulting value is used as the initial value of the fluctuation amplitude. The peak values ​​of strain signal amplitude and acceleration signal amplitude in the vibration response signal are used as verification benchmarks. The theoretical vibration amplitude corresponding to the initial value of the fluctuation amplitude is compared with the actual collected peak values ​​to calculate the deviation. The deviation is introduced as a correction coefficient to obtain the quantitative correlation between the fluctuation amplitude and the damping ratio, and thus obtain the fluctuation amplitude. Based on the principle of vibration energy accumulation, the product of 2π and the real-time natural frequency is divided by the quotient of the product of the damping ratio and the fluctuation amplitude to obtain the fluctuation period.

10. The ice-breaking method based on a combination of high-pressure gas and a resonance device according to claim 1, characterized in that, In step three, after the ice layer breaks, effective passage cross-section data is collected, and the collection time is set to be no less than one complete fluctuation cycle, which serves as the basis for judging whether the ice layer is completely broken.

Images