Distributed intelligent monitoring method and system based on multi-node edge computing device

Abstract

The application relates to the technical field of distributed intelligent monitoring, and discloses a distributed intelligent monitoring method and system based on a multi-node edge computing device, which comprises the following steps: arranging M*N monitoring nodes at a fixed interval on a target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds a preset acceleration threshold, recording the corresponding trigger time, and calculating the real-time wheel speed of a vehicle based on the trigger time and the fixed interval; obtaining the main frequency of fluctuation based on the vertical acceleration corresponding to the trigger time, and associating the main frequency of fluctuation with the real-time wheel speed of the vehicle, determining the wave number of fluctuation first, and then determining the vertical displacement of an ice layer and the curvature of the ice layer based on the wave number of fluctuation; calculating the corrected wave speed based on the vertical displacement of the ice layer and the fixed interval, so as to obtain a damping coefficient; and combining the curvature of the ice layer and prior material parameters to calculate the maximum bending stress of the corresponding monitoring node.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed intelligent monitoring, more particularly, it relates to a distributed intelligent monitoring method and system based on multi-node edge computing devices. BACKGROUND

[0002] In cold regions, the demand for ice surface as a temporary traffic channel is widespread. Existing ice layer safety monitoring technologies include monitoring ice layer thickness, temperature field distribution and static stress by pre-embedded temperature sensors, strain gauges or sonar devices, and triggering an early warning when the parameters exceed the threshold. The existing ice layer safety monitoring technology is only applicable to static load or low dynamic load scenarios and cannot reflect the dynamic fluctuation characteristics of the ice layer when the vehicle is running.

[0003] When a vehicle is running on the ice surface, the concentrated load of the wheels will cause elastic bending vibration of the ice surface, and at the same time form a gravity wave in the water body under the ice. The gravity wave propagates along the ice surface forward, consistent with the direction and speed of the vehicle movement.

[0004] The core defect of the existing ice layer safety monitoring technology is that:

[0005] The existing ice layer safety monitoring technology does not establish a monitoring model based on the dynamic fluctuation characteristics of the ice layer when the vehicle is running, and cannot identify the high-risk state of stress amplification when the vehicle speed approaches the critical wave speed, resulting in safety risks when the vehicle is running. SUMMARY

[0006] The present application provides a distributed intelligent monitoring method and system based on multi-node edge computing devices to solve the technical problems raised in the background art.

[0007] In a first aspect, a distributed intelligent monitoring method based on multi-node edge computing devices, characterized in that it comprises:

[0008] Step 1: Arranging MxN monitoring nodes at a fixed interval on the target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds the preset acceleration threshold, recording the corresponding trigger time, and calculating the real-time wheel speed of the vehicle based on the trigger time and the fixed interval;

[0009] Step 2: Obtain the fluctuation main frequency based on the vertical acceleration corresponding to the trigger time, and associate the fluctuation main frequency with the real-time wheel speed of the vehicle to determine the fluctuation wave number first, and then determine the ice layer vertical displacement and ice layer curvature based on the fluctuation wave number;

[0010] Step 3: Calculate the corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain the damping coefficient; combine the ice layer curvature and the prior material parameters to calculate the maximum bending stress of the corresponding monitoring node;

[0011] Step 4, each monitoring node exchanges the corrected wave speed and the maximum bending stress, and iteratively processes the global average wave speed and the global average bending stress of the target monitoring ice surface by using a distributed alternating direction multiplier method;

[0012] Step 5, each monitoring node calculates a critical wave speed based on the global average wave speed and the prior material parameters, compares the global average bending stress with a preset bending stress threshold to generate a safety factor, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed;

[0013] Step 6, each monitoring node fuses the speed margin, the safety factor and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk.

[0014] Further, the real-time wheel speed of the vehicle is calculated based on the trigger time and the fixed interval, comprising:

[0015] The trigger time when the recorded vertical acceleration of the i-th monitoring node is greater than a preset acceleration threshold;

[0016] The trigger time of the adjacent monitoring node of the i-th monitoring node is differentially processed to obtain a differential time;

[0017] The fixed interval is divided by the differential time of the trigger time to obtain the real-time wheel speed of the vehicle.

[0018] Further, the fluctuation main frequency is obtained based on the vertical acceleration corresponding to the trigger time, and the fluctuation wave number is determined by correlating the fluctuation main frequency with the real-time wheel speed of the vehicle, comprising:

[0019] A time window of the vertical acceleration of the i-th monitoring node is taken with the trigger time as the center;

[0020] The time-frequency analysis is performed on the vertical acceleration in the time window to obtain the fluctuation main frequency;

[0021] The fluctuation main frequency is correlated with the real-time wheel speed of the vehicle to output the corresponding fluctuation wave number.

[0022] Further, the ice layer vertical displacement and the ice layer curvature are determined based on the fluctuation wave number, comprising:

[0023] The prior material parameters include: ice layer density, ice layer Poisson's ratio, ice layer elastic modulus, ice layer bending stiffness of each monitoring node and ice layer thickness of each monitoring node;

[0024] The fluctuation wave number of each monitoring node and the prior material parameters are brought into the vibration equation as follows:

[0025]

[0026] wherein, denotes the second derivative of the ice layer vertical displacement of the i-th monitoring node with respect to the triggering time, denotes the ice layer density, denotes the ice layer bending stiffness of the i-th monitoring node, denotes the ice layer thickness of the monitoring node of the i-th monitoring node, denotes the wave number of the i-th monitoring node, denotes the ice layer vertical displacement of the i-th monitoring node; the ice layer vertical displacement is obtained by numerically integrating the vibration equation;

[0027] the wave number and the ice layer vertical displacement are squared to obtain the ice layer curvature.

[0028] Further, a corrected wave speed is calculated based on the ice layer vertical displacement and the fixed interval to obtain a damping coefficient, including:

[0029] A triggering time window H is set to extract the ice layer vertical displacement and the ice layer vertical displacement amplitude of the monitoring node within the time window H with the triggering time of the monitoring node as the starting point;

[0030] The ice layer vertical displacement peak value of the i-th monitoring node within the time window H and the ice layer vertical displacement peak value of the adjacent monitoring node of the i-th monitoring node within the time window H are determined respectively, the arrival time of the ice layer vertical displacement peak value is recorded respectively, and the arrival time difference is calculated; the ratio of the fixed interval and the arrival time difference is taken as the corrected wave speed;

[0031] The ice layer vertical displacement amplitude peak value of the i-th monitoring node within the time window H and the ice layer vertical displacement amplitude peak value of the adjacent monitoring node of the i-th monitoring node within the time window H are determined respectively, and the damping coefficient of the i-th monitoring node is calculated as follows:

[0032] ,

[0033] wherein, denotes the fixed interval.

[0034] Further, the maximum bending stress of the corresponding monitoring node is calculated in combination with the ice layer curvature and the prior material parameters, including:

[0035] The maximum bending stress of the i-th monitoring node is calculated as follows:

[0036]

[0037] wherein, a maximum bending stress of the i-th monitoring node, an ice layer elastic modulus, an ice layer curvature of the i-th monitoring node.

[0038] Further, each monitoring node exchanges the corrected wave speed and the maximum bending stress, and iteratively processes by using a distributed alternating direction multiplier method to obtain a global average wave speed and a global average bending stress of the target monitoring ice surface, including:

[0039] Step 41, loading local variables and dual variables of the alternating direction multiplier; wherein the corrected wave speed and the maximum bending stress are respectively taken as the local variables of the alternating direction multiplier, and the corresponding dual variables are initialized to be 0;

[0040] Step 42, at the k-th iteration, k≥0, based on the dual variables and of the k-th iteration, the local variables are updated to obtain;

[0041] Step 43, obtaining the average value of the M×N monitoring nodes as the global average value of the k+1-th iteration; wherein, ;

[0042] Step 44, based on the global average value of the k+1-th iteration, the dual variables are updated to obtain;

[0043] Step 45, repeating steps 42 to 44 for a preset number of times to obtain the global average wave speed and the global average bending stress of the target monitoring ice surface.

[0044] Further, each monitoring node calculates a critical wave speed based on the global average wave speed and the prior material parameters, compares the global average bending stress with a preset bending stress threshold to generate a safety factor, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed, including:

[0045] The calculation formula of the critical wave speed is as follows:

[0046] ,

[0047] ,

[0048] wherein, represents the critical wave speed of the i-th monitoring node, a critical wave number of the i-th monitoring node, a Poisson's ratio of the ice layer, a gravitational acceleration; wherein a global average wave speed is taken as an initial value of the critical wave number, and the critical wave number is iteratively converged to obtain the critical wave number;

[0049] a ratio of the global average bending stress to a preset bending stress threshold is taken as a safety factor ;

[0050] a ratio of the vehicle real-time wheel speed to the critical wave speed is taken as a speed margin of the corresponding monitoring node .

[0051] Further, each monitoring node fuses the speed margin, the safety factor, and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of a target monitoring ice surface according to the global average risk, including:

[0052] linearly fusing the speed margin and the damping coefficient, and the safety factor of each monitoring node according to a preset weight to obtain a local risk indicator of each monitoring node;

[0053] taking an average value of the local risk indicators of the MxN monitoring nodes to obtain the global average risk;

[0054] when the global average risk is greater than or equal to a preset global risk threshold, the target monitoring ice surface is closed; otherwise, the target monitoring ice surface is kept open.

[0055] In a second aspect, a distributed intelligent monitoring system based on a multi-node edge computing device is applied to any of the distributed intelligent monitoring methods based on a multi-node edge computing device, and includes:

[0056] a data acquisition module, configured to arrange MxN monitoring nodes at a fixed interval on a target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds a preset acceleration threshold, record a corresponding trigger time, and calculate a vehicle real-time wheel speed based on the trigger time and the fixed interval;

[0057] a data processing module, configured to obtain a wave fluctuation main frequency based on the vertical acceleration corresponding to the trigger time, and associate the wave fluctuation main frequency with the vehicle real-time wheel speed to determine a wave fluctuation wave number first, and then determine an ice layer vertical displacement and an ice layer curvature based on the wave fluctuation wave number;

[0058] a bending stress module, configured to calculate a corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain a damping coefficient, and calculate a maximum bending stress of the corresponding monitoring node in combination with the ice layer curvature and prior material parameters;

[0059] ​The iterative averaging module is used to exchange the corrected wave velocity and the maximum bending stress among the monitoring nodes, and then use the distributed alternating direction multiplier method to iteratively process and obtain the global average wave velocity and global average bending stress of the target monitored ice surface.

[0060] The risk factor module calculates the critical wave velocity based on the global average wave velocity and the prior material parameters at each monitoring node, compares the global average bending stress with the preset bending stress threshold to generate a safety factor, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave velocity.

[0061] The risk assessment module integrates the speed margin, the safety factor, and the damping coefficient of each monitoring node according to preset weights to obtain the global average risk, and outputs a management plan for the target monitored ice surface based on the global average risk.

[0062] The beneficial effects of this invention are as follows: by coupling vehicle speed, ice wave velocity, and bending stress, a dynamic assessment of the safety status of the ice surface is achieved throughout the entire process of vehicle real-time driving. This breaks through the technical bottleneck of existing static monitoring technology, which can only reflect the static load state and cannot identify the stress amplification induced by the critical wave velocity. Thus, predictive early warning and intelligent management of vehicle driving safety are realized, improving the reliability of temporary ice surface passage in cold regions. Attached Figure Description

[0063] Figure 1 This is a flowchart of the distributed intelligent monitoring method based on multi-node edge computing devices according to the present invention. Detailed Implementation

[0064] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0065] like Figure 1 As shown, the distributed intelligent monitoring method based on multi-node edge computing devices includes:

[0066] Step 1: Arrange M×N monitoring nodes at fixed intervals on the target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds the preset acceleration threshold, record the corresponding trigger time, and calculate the real-time wheel speed of the vehicle based on the trigger time and the fixed interval.

[0067] In one embodiment of the present invention, calculating the real-time wheel speed of the vehicle based on the triggering time and the fixed interval includes:

[0068] The trigger time when the vertical acceleration of the i-th monitoring node exceeds the preset acceleration threshold;

[0069] The trigger times of the monitoring nodes adjacent to the i-th monitoring node are differentially processed to obtain the differential time;

[0070] Divide the difference between the fixed interval and the trigger time to obtain the real-time wheel speed of the vehicle.

[0071] In detail, step 1 is as follows:

[0072] To quickly and accurately obtain the real-time speed of a vehicle when it enters the ice, this invention employs a method based on trigger time and fixed intervals to estimate the vehicle speed online, as follows:

[0073] Several monitoring nodes are arrayed at fixed intervals on both sides of the road for driving on the ice. Each monitoring node is numbered sequentially along the direction of vehicle travel. The horizontal distance between adjacent nodes was calibrated by technicians during installation and stored in the local memory of the monitoring nodes, denoted as . The unit is meters.

[0074] Each monitoring node is equipped with a vertical acceleration sensor to capture the instantaneous impact vibrations generated when a vehicle load arrives. Under normal conditions, the sensor continuously collects the vertical acceleration at the monitoring node at a fixed sampling frequency. ,in, Indicates the first One monitoring node, This is a time variable, measured in seconds. An acceleration threshold is set by experts. When a vehicle wheel passes a monitoring node, the vertical acceleration at the monitoring node will experience a short-term surge due to the vehicle load.

[0075] If the vertical acceleration at this moment satisfy, Then determine the first A monitoring node is triggered, and this moment is recorded as the trigger time. .

[0076] The vehicles successively ran over the first The monitoring node and the first One monitoring node. Record the trigger time when a monitoring node is triggered. ;No. Record the trigger time when a monitoring node is triggered. .but ;

[0077] Further,

[0078] .

[0079] Step 2, obtaining a fluctuation main frequency based on the vertical acceleration corresponding to the trigger time, and associating the fluctuation main frequency with the real-time wheel speed of the vehicle to determine a fluctuation wave number, and then determining an ice layer vertical displacement and an ice layer curvature based on the fluctuation wave number;

[0080] In an embodiment of the present application, the fluctuation main frequency is obtained based on the vertical acceleration corresponding to the trigger time, and the fluctuation main frequency is associated with the real-time wheel speed of the vehicle to determine the fluctuation wave number, including:

[0081] Taking the trigger time as the center, a time window of the vertical acceleration of the i th monitoring node is intercepted;

[0082] Performing time-frequency analysis on the vertical acceleration in the time window to obtain the fluctuation main frequency;

[0083] Associating the fluctuation main frequency with the real-time wheel speed of the vehicle to output the corresponding fluctuation wave number.

[0084] In detail, when the vehicle is driving on the ice surface, the concentrated load generated by the vehicle wheel will cause the ice surface to be instantaneously bent at the position under the vehicle wheel, and a bending-gravity wave coupled with the ice layer will be formed in the water body under the ice layer. The wave propagates forward along the ice surface and synchronizes with the vehicle; that is, when the vehicle advances, the phase of the wave advances forward at the interface of the ice layer relative to the water body at the same speed (approximately equal to the vehicle speed). The main space-time characteristics of the bending-gravity wave are:

[0085] Main frequency : indicating the basic oscillation frequency of the fluctuation in the time dimension, reflecting the periodic characteristics of the ice surface panel vibration;

[0086] Wave speed : the speed in the ice layer and the water body medium below when propagating, under the driving of the vehicle load, the wave speed is approximately equal to the real-time wheel speed of the vehicle ;

[0087] Fluctuation wave number : describing the distribution characteristics of the wave in the spatial dimension (the i th monitoring node), , and satisfying the dispersion relation:

[0088] , is the ice layer vibration frequency.

[0089] Step 21, taking the trigger time as the center to intercept an acceleration time window, as follows:

[0090] At the triggering time , a symmetric time window with width is intercepted, i.e. the vertical acceleration segment between is reserved, denoted as:

[0091] ,

[0092] Since the load exerted by the vehicle wheels on the ice surface is the strongest at the triggering time, the vibration response of the ice layer usually lasts for a short time interval after the triggering. The vertical acceleration segment with width is intercepted to reserve the pre-shock and subsequent oscillation information before and after the triggering time, so as to avoid confusion of multiple continuous triggering events.

[0093] Step 22, the continuous wavelet transform is performed on the vertical acceleration segment , and the wavelet coefficients are calculated; the modulus square of the wavelet coefficients is taken to obtain the energy spectrum; at the triggering time , the scale corresponding to the maximum value of the energy spectrum is searched, and the wave frequency is calculated by using the inverse ratio relationship between the mother wavelet center frequency and the scale.

[0094] Step 23, under the action of the vehicle load, the ice layer and the underlying water form a coupled thin plate-fluid system, and the bending-gravity wave propagating in the system satisfies the following linear dispersion relation (linear infinite depth approximation):

[0095] ,

[0096] wherein denotes the gravitational acceleration, denotes the bending stiffness of the ice layer, denotes the density of the ice layer, denotes the thickness of the ice layer at the i-th monitoring node;

[0097] However, when the vehicle real-time wheel speed rolls on the ice surface, the actually observed wave mainly presents the propagation characteristic of “moving at the same speed as the vehicle”, and has strong excitation strength, so that the wave approximately satisfies the simplified relationship under the driving of “moving load”:

[0098] i.e. the vibration frequency of the ice layer under the driving of the load is roughly proportional to the spatial wave number , and the proportional coefficient is approximately the vehicle speed. Based on this approximation, the corresponding formula between the wave number and the frequency and the wheel speed can be directly obtained without solving the more complex dispersion equation, and the wave number of the wave at the i-th monitoring node is , and the linear approximation formula is:

[0099] ​ The embodiment has the following remarkable beneficial effects:

[0100] The dynamic coupling between the vehicle speed, the ice layer fluctuation frequency and the wave number is realized, and the high-risk state of the sharp increase of the stress in the ice layer when the vehicle speed approaches or exceeds the critical wave speed can be accurately identified.

[0101] All the calculations are completed locally on the edge monitoring node, and the processing delay is hundreds of milliseconds, which fully meets the real-time requirements of vehicle driving monitoring.

[0102] By using the linear approximation formula, the calculation complexity caused by directly solving the high-order dispersion equation is avoided, and the demand for the computing power of the embedded edge device is reduced.

[0103] In detail, step 2 is specifically as follows:

[0104] In an embodiment of the present application, the ice layer vertical displacement and the ice layer curvature are determined based on the fluctuation wave number, and the determination includes:

[0105] The prior material parameters include: ice layer density, ice layer Poisson's ratio, ice layer elastic modulus, ice layer bending stiffness of each monitoring node and ice layer thickness of each monitoring node;

[0106] The fluctuation wave number of each monitoring node and the prior material parameters are brought into the vibration equation as follows:

[0107] , wherein, represents the second-order derivative of the ice layer vertical displacement corresponding to the i-th monitoring node with respect to the triggering time, represents the ice layer density, represents the ice layer bending stiffness of the i-th monitoring node, represents the ice layer thickness of the monitoring node of the i-th monitoring node, represents the fluctuation wave number of the i-th monitoring node, represents the ice layer vertical displacement of the i-th monitoring node; the ice layer vertical displacement is obtained by numerically integrating the vibration equation;

[0108] The fluctuation wave number and the ice layer vertical displacement are squared to obtain the ice layer curvature.

[0109] In detail, step 2 further includes:

[0110] In order to directly invert the ice layer vertical displacement and the curvature based on the fluctuation wave number, the prior material parameters of the ice layer are introduced, and these parameters and the wave number are brought into the ice layer vibration equation, and finally the ice layer vertical displacement at each monitoring node is obtained by numerical solution, and the ice layer curvature is further calculated, which is specifically as follows:

[0111] Ice layer density: refers to the material density of the ice body material itself, with the unit of kg / m³. The typical value is 917 kg / m³.

[0112] Ice layer Poisson's ratio: refers to the ratio between the lateral strain and the longitudinal strain of the ice body material. The typical value is Poisson's ratio 0.3.

[0113] Ice layer elastic modulus: refers to the stiffness coefficient of the ice body material in the elastic deformation stage, with the unit of Pa. The typical value is 9 GPa.

[0114] Ice layer thickness of each monitoring node: refers to the ice layer thickness at the location of the i-th monitoring node, with the unit of meters. It is determined based on the measurement of the sonar unit of the monitoring node.

[0115] Ice layer bending stiffness of each monitoring node: the bending stiffness of the ice layer at the i-th monitoring node, with the unit of Nm, as the thin plate bending elastic coefficient.

[0116] In the coupling system of the ice surface and the underlying water body, the ice layer acts as an elastic thin plate with a thickness much smaller than the length dimension, and its bending dynamics can be approximately described by the Kirchhoff-Love thin plate theory. The free vibration equation of the elastic thin plate derived from this theory contains four-order spatial derivative terms and two-order time derivative terms.

[0117] Based on the simplified vibration model of bending-gravity waves in the thin plate-fluid system, the wave number detected at each monitoring node and the prior material parameters are substituted into the transient vibration equation of the node for numerical solution, and the vertical displacement of the ice layer is obtained, and finally the curvature of the ice layer is calculated by multiplying the wave number and the displacement. The specific process is as follows:

[0118] At the i-th monitoring node, the wave number recorded by the i-th monitoring node is combined with the prior material parameters. Then the transient bending vibration of the node can be approximately described by a one-dimensional simplified thin plate vibration differential equation as follows:

[0119] wherein, represents the vertical displacement of the ice layer corresponding to the i-th monitoring node, represents the second-order derivative of the vertical displacement of the ice layer at the i-th monitoring node with respect to the triggering time , i.e. the vertical acceleration; represents the fourth power of the wave number, reflecting the influence of the high-order derivative term in the thin plate bending differential equation on the bending shape. The physical meaning is that when the wave number increases, the spatial curvature increases, and the energy distribution of the corresponding vibration mode will change accordingly.

[0120] It should be noted that the vertical displacement of the ice layer at each monitoring node is obtained by numerical solution, and the numerical solution algorithm can be: fourth-order Runge-Kutta method.

[0121] In the bending mechanics of thin plates, the local curvature of the ice layer... This can be approximated as the second spatial derivative of the displacement. However, in actual measurements, a continuous displacement curve of the ice layer in the spatial direction is not obtained; instead, wavenumbers are used. This indirectly characterizes the spatial distribution features.

[0122] The second spatial derivative is expressed as:

[0123] ,

[0124] in, This indicates the spatial coordinates (in meters) along the longitudinal direction of the ice surface, with the direction aligned with the vehicle's travel direction. It represents the second-order partial derivative with respect to spatial coordinate x, describing the degree of curvature of the ice layer's vertical displacement in the spatial coordinate x direction;

[0125] ice curvature and wave number and vertical displacement of ice layer There is an approximate relationship:

[0126] ,

[0127] Step 3: Calculate the corrected wave velocity based on the vertical displacement of the ice layer and the fixed spacing to obtain the damping coefficient; combine the ice layer curvature and prior material parameters to calculate the maximum bending stress of the corresponding monitoring node;

[0128] In one embodiment of the present invention, the corrected wave velocity is calculated based on the vertical displacement of the ice layer and the fixed spacing to obtain the damping coefficient, including:

[0129] Set a trigger time window H to extract the vertical displacement of the ice layer and the amplitude of the vertical displacement of the ice layer within the time window H, starting from the trigger time of the monitoring node.

[0130] The peak value of the vertical displacement of the ice layer at the i-th monitoring node within the time window H, and the peak value of the vertical displacement of the ice layer at the adjacent monitoring node of the i-th monitoring node within the time window H are determined respectively. The arrival time of the peak value of the vertical displacement of the ice layer is recorded respectively, and the arrival time difference is calculated. The ratio of the fixed spacing to the arrival time difference is used as the corrected wave velocity.

[0131] Determine the peak value of the vertical displacement amplitude of the ice layer at the i-th monitoring node within the time window H. And the peak value of the vertical displacement amplitude of the ice layer at the monitoring nodes adjacent to the i-th monitoring node within the time window H. And calculate the damping coefficient of the i-th monitoring node. ,as follows:

[0132] wherein, represents a fixed interval.

[0133] In detail, the detection node defines a time period with the time as the center when recording the triggering time . The absolute value (amplitude) of the ice layer vertical displacement at each time in the time period is obtained, and the maximum value in the absolute value is taken as the vertical displacement amplitude peak value, When the vehicle travels on the ice layer, a time sequence of bending-gravity waves will be excited, and the bending-gravity waves will experience two significant physical processes during the forward propagation: one is that the wave propagation speed is not strictly equal to the speed corresponding to the time difference of the triggering time recorded by the node at the triggering time of the vehicle, but slightly lags behind the instantaneous load of the vehicle; the second is that the wave will continuously decrease in amplitude due to the viscous dissipation of the ice layer itself and the water body below during the propagation process. Based on these two points, we need to introduce a modified wave speed and a damping coefficient to more accurately depict the real propagation characteristics of the wave, as follows:

[0134] When the vehicle tire passes through the monitoring node, the acceleration sensor of the monitoring node records the transient impact of the vehicle load on the ice surface, at which time a triggering time can be determined. However, the bending-gravity wave does not directly reach the adjacent node at the triggering time, but needs to travel a certain propagation path, during which the mutual coupling, speed conversion and energy redistribution between the water body and the ice layer occur, resulting in the peak displacement time of the wave reaching the adjacent node being slightly later than the triggering time. That is, using only the difference in triggering time to estimate the wave speed will ignore this small but critical time domain lag. Within the time window H starting from the triggering time, the ice layer vertical displacement of each node time sequence is extracted, and the ice layer vertical displacement (amplitude) with the maximum absolute value and the corresponding time are found. This time is the strongest and most easily recognizable moment of the bending-gravity wave after the medium dissipation and propagation delay, that is, the time when the bending-gravity wave really reaches the monitoring node and produces the maximum displacement. The arrival time difference represents the actual time taken by the bending-gravity wave to complete one propagation between two adjacent monitoring nodes and reach the peak value. Therefore, the modified wave speed not only compensates for the phase lag effect, but also uses the most significant amplitude information of the ice layer vibration, so it is more accurate than simply using the triggering time difference to estimate the wave speed.

[0135] When the vehicle tire passes through the monitoring node, the acceleration sensor of the monitoring node records the transient impact of the vehicle load on the ice surface, at which time a triggering time can be determined. However, the bending-gravity wave does not directly reach the adjacent node at the triggering time, but needs to travel a certain propagation path, during which the mutual coupling, speed conversion and energy redistribution between the water body and the ice layer occur, resulting in the peak displacement time of the wave reaching the adjacent node being slightly later than the triggering time. That is, using only the difference in triggering time to estimate the wave speed will ignore this small but critical time domain lag. Within the time window H starting from the triggering time, the ice layer vertical displacement of each node time sequence is extracted, and the ice layer vertical displacement (amplitude) with the maximum absolute value and the corresponding time are found. This time is the strongest and most easily recognizable moment of the bending-gravity wave after the medium dissipation and propagation delay, that is, the time when the bending-gravity wave really reaches the monitoring node and produces the maximum displacement. The arrival time difference represents the actual time taken by the bending-gravity wave to complete one propagation between two adjacent monitoring nodes and reach the peak value. Therefore, the modified wave speed not only compensates for the phase lag effect, but also uses the most significant amplitude information of the ice layer vibration, so it is more accurate than simply using the triggering time difference to estimate the wave speed.

[0136] ​In detail, after the bending-gravity wave originates from the vehicle's loading point, it transfers energy within the ice layer due to elastic deformation, while simultaneously dissipating energy in the water layer beneath the ice due to viscous drag. This energy dissipation mechanism in the two-layer medium causes the amplitude of the bending-gravity wave to gradually decrease with horizontal propagation distance. Therefore, the bending-gravity wave decays exponentially along the propagation distance, with the damping coefficient quantifying the logarithmic decay rate of the wave amplitude per unit distance. The peak amplitude detected at each monitoring node was And at an interval of The The peak attenuation measured at each monitoring node is The damping coefficient is the factor that reduces the bending-gravity wave by a certain amount per unit length. The exponential rate is multiples of the damping coefficient. The damping coefficient characterizes the bending-gravity wave from the first... The monitoring node to the first The energy lost by each monitoring node can reflect the combined effects of the viscosity of the ice layer and water body, microcrack dissipation, and other factors on the attenuation of fluctuations.

[0137] therefore, Simplifying, we get: .

[0138] In one embodiment of the present invention, the maximum bending stress of the corresponding monitoring node is calculated by combining the ice layer curvature and prior material parameters, including:

[0139] The maximum bending stress at the i-th monitoring node is calculated as follows:

[0140] ,in, This represents the maximum bending stress at the i-th monitoring node. This represents the elastic modulus of the ice layer. Let represent the curvature of the ice layer at the i-th monitoring node.

[0141] In detail, when ice is subjected to vehicle loads, it approximates as a thin, elastic sheet under bending, undergoing bending deformation in the load area. In this bending state, the fibers within the ice are subjected to both tension and compression. For example, the fibers at the top of the ice are stretched along the horizontal plane, while the fibers at the bottom are compressed. The fibers at the top and bottom of the ice often experience the greatest stress; therefore, the maximum bending stress represents the limit of internal force that the weakest fiber within the ice can withstand when bent.

[0142] In the classic theory of elastic thin-plate bending, there is a very intuitive linear relationship between bending stress and curvature. For a sheet of thickness... If the ice layer curvatures at a certain point... The bending deformation, according to Hooke's Law and thin plate mechanics, is such that the thickness is... ice curvature At that time, located The maximum bending stress is greatest at the i-th monitoring node, expressed as:

[0143] ,

[0144] Step 4: Each monitoring node exchanges the corrected wave velocity and the maximum bending stress, and then uses the distributed alternating direction multiplier method to iterate and obtain the global average wave velocity and global average bending stress of the target monitored ice surface.

[0145] In one embodiment of the present invention, each monitoring node exchanges the corrected wave velocity and the maximum bending stress, and obtains the global average wave velocity and global average bending stress of the target monitored ice surface after iterative processing using the distributed alternating direction multiplier method, including:

[0146] Step 41: Load the local and dual variables of the alternating direction multipliers; among them, the corrected wave velocity will be... and maximum bending stress Local variables, respectively, used as alternating direction multipliers Initialize the corresponding dual variables All are 0;

[0147] Step 42, in the k-th iteration, k≥0, based on the dual variable of the k-th iteration. Update to get local variables The details are as follows:

[0148] ,

[0149] in, Let each represent a local variable obtained by the i-th monitoring node in the (k+1)-th iteration. Let each represent the dual variable of the i-th monitoring node in the k-th iteration. The penalty factor represents the penalty factor in the distributed alternating direction multiplier method.

[0150] Step 43, obtain the data from M×N monitoring nodes. average These are respectively used as the global average values ​​for the (k+1)th iteration; where, ;

[0151] Step 44: Based on the global average value of the (k+1)th iteration, update the dual variable. The details are as follows:

[0152] , ,

[0153] wherein, respectively represent the dual variables of the k+1th iteration;

[0154] Step 45, the steps 42 to 44 are repeated for a preset number of times to obtain the global average wave speed and the global average bending stress of the target monitoring ice surface.

[0155] It should be noted that the distributed alternating direction multiplier method is prior art, and thus will not be described again.

[0156] Step 5, each monitoring node calculates a critical wave speed based on the global average wave speed and the prior material parameters, generates a safety factor by comparing the global average bending stress with a preset bending stress threshold, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed;

[0157] In an embodiment of the present application, each monitoring node calculates a critical wave speed based on the global average wave speed and the prior material parameters, generates a safety factor by comparing the global average bending stress with a preset bending stress threshold, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed, including:

[0158] The calculation formula of the critical wave speed is as follows:

[0159] ,

[0160] wherein, represents the critical wave speed of the i th monitoring node, represents the critical wave number of the i th monitoring node, represents the Poisson's ratio of the ice layer, represents the acceleration of gravity; wherein the global average wave speed is taken as the initial value of the critical wave number, and the iteration converges to obtain the critical wave number;

[0161] The ratio of the global average bending stress to the preset bending stress threshold is taken as the safety factor .

[0162] The ratio of the real-time wheel speed of the vehicle to the critical wave speed is taken as the speed margin of the corresponding monitoring node .

[0163] In detail, when the vehicle is running on the ice layer, the ice layer has bending-gravity waves. When the bending-gravity waves propagate on the ice layer, if the vehicle running speed is lower than a certain critical value, the motion of the bending-gravity waves is mainly in the form of gravity waves; once the speed exceeds the critical value, the bending of the ice layer (that is, the rigidity of the ice body itself as a solid thin plate) begins to dominate, causing a sudden change in the propagation mode of the bending-gravity waves in the ice. The critical wave speed represents the speed that can make the bending-gravity waves in the ice just change from gravity wave dominant to bending of the ice layer dominant. When the actual running speed of the vehicle approaches or exceeds the critical wave speed, the bending deformation inside the ice plate will be concentrated and amplified, causing the ice layer to be prone to cracking.

[0164] The safety factor represents the ratio between the maximum bending stress borne by the ice layer and the preset maximum allowable bending stress of the ice layer, which is set by experts. If the safety factor is greater than 1, the risk is higher, and if the safety factor is less than or equal to 1, the risk of the ice layer is lower.

[0165] The speed margin represents the ratio of the real-time wheel speed of the vehicle to the critical wave speed. If the speed margin is greater than 1, the risk of the ice layer is higher, and if the speed margin is less than or equal to 1, the risk of the ice layer is lower.

[0166] Step 6, each monitoring node fuses the speed margin, the safety factor and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk.

[0167] In an embodiment of the present application, each monitoring node fuses the speed margin, the safety factor and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk, including:

[0168] The speed margin and the damping coefficient, and the safety factor of each monitoring node are linearly fused according to a preset weight to obtain a local risk indicator of each monitoring node;

[0169] The average value of the local risk indicators of the monitoring nodes is taken to obtain a global average risk; When the global average risk is greater than or equal to a preset global risk threshold, the target monitoring ice surface is closed; otherwise, the target monitoring ice surface is kept open.

[0170] In detail, the calculation formula of the local risk indicator of the i-th monitoring node is as follows:

[0171]

[0172] , wherein, respectively represent a first preset weight, a second preset weight and a third preset weight, respectively represent a first preset weight, a second preset weight and a third preset weight, The sum of the values is 1, and each value is greater than 0.

[0173] The distributed intelligent monitoring system based on the multi-node edge computing device is applied to any of the distributed intelligent monitoring methods based on the multi-node edge computing device, and comprises:

[0174] The data acquisition module is configured to arrange MxN monitoring nodes at a fixed interval on the target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds a preset acceleration threshold, a corresponding trigger time is recorded, and a real-time wheel speed of the vehicle is calculated based on the trigger time and the fixed interval;

[0175] The data processing module is configured to obtain a fluctuation main frequency based on the vertical acceleration corresponding to the trigger time, associate the fluctuation main frequency with the real-time wheel speed of the vehicle, determine a fluctuation wave number first, and then determine an ice layer vertical displacement and an ice layer curvature based on the fluctuation wave number;

[0176] The bending stress module is configured to calculate a corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain a damping coefficient, and calculate a maximum bending stress of the corresponding monitoring node in combination with the ice layer curvature and a prior material parameter;

[0177] The iterative average module is configured to exchange the corrected wave speed and the maximum bending stress among the monitoring nodes, and obtain a global average wave speed and a global average bending stress of the target monitoring ice surface after iterative processing by using a distributed alternating direction multiplier method;

[0178] The risk factor module is configured to calculate a critical wave speed based on the global average wave speed and the prior material parameter, compare the global average bending stress with a preset bending stress threshold to generate a safety factor, and generate a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed;

[0179] The risk assessment module is configured to fuse the speed margin, the safety factor, and the damping coefficient according to a preset weight to obtain a global average risk, and output a management scheme of the target monitoring ice surface according to the global average risk.

[0180] The above describes the embodiments of the embodiments, but the embodiments are not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative and not limiting, and those skilled in the art can make many forms under the inspiration of the embodiments, which are all within the protection scope of the embodiments.

Claims

1. A distributed intelligent monitoring method based on a multi-node edge computing device, characterized in that, Comprise: Step 1, arrange MxN monitoring nodes at a fixed interval on the target monitoring ice surface; when the vertical acceleration collected by any monitoring node exceeds the preset acceleration threshold, record the corresponding trigger time, and calculate the vehicle real-time wheel speed based on the trigger time and the fixed interval; Step 2, obtain the main frequency of fluctuation based on the vertical acceleration corresponding to the trigger time, and associate the main frequency of fluctuation with the vehicle real-time wheel speed to determine the fluctuation wave number first, and then determine the ice layer vertical displacement and the ice layer curvature based on the fluctuation wave number; Step 3, calculate the corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain the damping coefficient; Combine the ice layer curvature and the prior material parameters to calculate the maximum bending stress of the corresponding monitoring node; Step 4, each monitoring node exchanges the corrected wave speed and the maximum bending stress, and obtains the global average wave speed and the global average bending stress of the target monitoring ice surface after iterative processing by using the distributed alternating direction multiplier method; Step 5, each monitoring node calculates the critical wave speed based on the global average wave speed and the prior material parameters, compares the global average bending stress with the preset bending stress threshold to generate a safety factor, and generates a speed margin based on the vehicle real-time wheel speed and the critical wave speed; Step 6, each monitoring node fuses the speed margin, the safety factor and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk. 2.The multi-node edge computing device based distributed intelligent monitoring method of claim 1, wherein, Calculate the vehicle real-time wheel speed based on the trigger time and the fixed interval, comprising: Record the trigger time of the vertical acceleration of the ith monitoring node greater than the preset acceleration threshold; Differential processing the trigger time of the adjacent monitoring nodes of the ith monitoring node to obtain a differential time; Divide the fixed interval by the differential time of the trigger time to obtain the vehicle real-time wheel speed. 3.The multi-node edge computing device based distributed intelligent monitoring method of claim 2, wherein, Obtain the main frequency of fluctuation based on the vertical acceleration corresponding to the trigger time, and associate the main frequency of fluctuation with the vehicle real-time wheel speed to determine the fluctuation wave number, comprising: Take the trigger time as the center to intercept a time window of the vertical acceleration of the ith monitoring node; Perform time-frequency analysis on the vertical acceleration in the time window to obtain the main frequency of fluctuation; Associate the main frequency of fluctuation with the vehicle real-time wheel speed to output the corresponding fluctuation wave number. 4.The multi-node edge computing device based distributed intelligent monitoring method of claim 3, wherein, Determine the ice layer vertical displacement and the ice layer curvature based on the fluctuation wave number, comprising: The prior material parameters include: ice layer density, ice layer Poisson's ratio, ice layer elastic modulus, ice layer bending stiffness of each monitoring node and ice layer thickness of each monitoring node; Bring the fluctuation wave number of each monitoring node and the prior material parameters into the vibration equation as follows: wherein, represents the second derivative of the ice layer vertical displacement of the i-th monitoring node with respect to the triggering time, represents the ice layer density, represents the ice layer bending stiffness of the i-th monitoring node, represents the ice layer thickness of the monitoring node of the i-th monitoring node, represents the wave number of the i-th monitoring node, represents the ice layer vertical displacement of the i-th monitoring node; the ice layer vertical displacement is obtained by numerically integrating the vibration equation; Perform a quadratic operation on the fluctuation wave number and the ice layer vertical displacement to obtain the ice layer curvature. 5.The multi-node edge computing device based distributed intelligent monitoring method of claim 4, wherein, Calculate the corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain the damping coefficient, comprising: Set a trigger time window H to extract the ice layer vertical displacement and the ice layer vertical displacement amplitude of the monitoring node within the time window H with the trigger time of the monitoring node as the starting point; respectively determine an ice layer vertical displacement peak value of the i th monitoring node in a time window H, and an ice layer vertical displacement peak value of a neighboring monitoring node of the i th monitoring node in the time window H, respectively record the arrival time of the ice layer vertical displacement peak value, and calculate a time difference of arrival; take a ratio of the fixed interval and the time difference of arrival as a corrected wave speed; respectively determine the ice layer vertical displacement amplitude peak value of the ith monitoring node within the time window H and the ice layer vertical displacement amplitude peak value of the adjacent monitoring node of the ith monitoring node within the time window H and calculate the damping coefficient of the ith monitoring node as follows: ， wherein represents a fixed interval. 6.The multi-node edge computing device based distributed intelligent monitoring method of claim 5, wherein, combine the ice layer curvature and the prior material parameters to calculate a maximum bending stress of the corresponding monitoring node, including: calculate the maximum bending stress of the i th monitoring node as follows: wherein, represents the maximum bending stress of the i-th monitoring node, represents the ice layer elastic modulus, represents the ice layer curvature of the i-th monitoring node. 7.The multi-node edge computing device based distributed intelligent monitoring method of claim 6, wherein, each monitoring node exchanges the corrected wave speed and the maximum bending stress, and iteratively processes by using a distributed alternating direction multiplier method to obtain a global average wave speed and a global average bending stress of the target monitoring ice surface, including: Step 41, load the local variable and the dual variable of the alternating direction multiplier; wherein the modified wave velocity and the maximum bending stress are respectively taken as the local variable of the alternating direction multiplier , the corresponding dual variable is initialized to 0; Step 42, at the kth iteration, k > 0, update the local variable based on the kth iteration of the dual variable and update the local variable ; Step 43, obtain the average value of the MxN monitoring nodes respectively as the global average value of the k+1 iteration; wherein, ;​ Step 44, update the dual variable based on the global average value of the k+1th iteration ; Step 45, repeat steps 42 to 44 for a predetermined number of times to obtain the global average wave speed and the global average bending stress of the target monitoring ice surface. 8.The multi-node edge computing device based distributed intelligent monitoring method of claim 7, wherein, Each monitoring node calculates a critical wave speed based on the global average wave speed and the prior material parameters, compares the global average bending stress with a preset bending stress threshold to generate a safety factor, and generates a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed, including: The calculation formula of the critical wave speed is as follows: ， ， wherein, represents the critical wave speed of the i-th monitoring node, represents the critical wave number of the i-th monitoring node, represents the Poisson's ratio of the ice layer, represents the gravitational acceleration; wherein the global average wave speed is taken as an initial value for the critical wave number, which is iteratively converged to obtain the critical wave number; a ratio of a global average bending stress to a preset bending stress threshold as a safety factor ; The ratio of the vehicle real-time wheel speed to the critical wave speed as the speed margin of the corresponding monitoring node . 9.The multi-node edge computing device based distributed intelligent monitoring method of claim 8, wherein, Each monitoring node fuses the speed margin, the safety factor and the damping coefficient according to a preset weight to obtain a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk, including: linearly fuse the speed margin and the damping coefficient, and the safety factor of each monitoring node according to a preset weight to obtain a local risk indicator of each monitoring node; Take the average of the local risk indicators of the monitoring nodes to obtain a global average risk; When the global average risk is greater than or equal to a preset global risk threshold, the target monitoring ice surface is closed; otherwise, the target monitoring ice surface is kept open.

10. A distributed intelligent monitoring system based on multi-node edge computing devices, applied in the distributed intelligent monitoring method based on multi-node edge computing devices of any one of claims 1-9, characterized in that, including: The data acquisition module is configured to arrange M×N monitoring nodes on the target monitoring ice surface at a fixed interval; when the vertical acceleration collected by any monitoring node exceeds a preset acceleration threshold, record the corresponding trigger time, and calculate the real-time wheel speed of the vehicle based on the trigger time and the fixed interval; The data processing module is configured to obtain a wave frequency based on the vertical acceleration corresponding to the trigger time, and associate the wave frequency with the real-time wheel speed of the vehicle, first determine a wave number, and then determine the ice layer vertical displacement and the ice layer curvature based on the wave number; The bending stress module is configured to calculate a corrected wave speed based on the ice layer vertical displacement and the fixed interval to obtain a damping coefficient; combine the ice layer curvature and the prior material parameters to calculate a maximum bending stress of the corresponding monitoring node; The iterative average module is configured to exchange the corrected wave speed and the maximum bending stress by each monitoring node, and iteratively process by using a distributed alternating direction multiplier method to obtain a global average wave speed and a global average bending stress of the target monitoring ice surface; The risk factor module is configured to calculate a critical wave speed based on the global average wave speed and the prior material parameters by each monitoring node, compare the global average bending stress with a preset bending stress threshold to generate a safety factor, and generate a speed margin based on the real-time wheel speed of the vehicle and the critical wave speed. The risk assessment module fuses the speed margin, the safety factor and the damping coefficient according to a preset weight, obtains a global average risk, and outputs a management scheme of the target monitoring ice surface according to the global average risk.

Images