Dam safety monitoring system and method based on Internet of Things

Abstract

The present invention belongs to the technical field of water conservancy projects. The present invention provides a dam safety monitoring system and method based on the Internet of Things, including a vibration energy supply module, in which a cylindrical blunt body structure and a piezoelectric ceramic sensor are arranged on the water-facing surface of the dam, and the blunt body structure is used to guide the water flow to form a Karman vortex street, so as to monitor the water flow velocity of the dam in real time; the Karman vortex street shedding frequency is analyzed based on the FFT spectrum, and is matched with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data; a multi-dimensional observation and acquisition module uses the generated Karman vortex street shedding frequency to convert the kinetic energy of the dam water into electrical energy, and the converted electrical energy is used to power a preset multi-dimensional sensor, and multi-dimensional observation data of the dam is collected based on the multi-dimensional sensor; a three-dimensional hopping communication module senses and identifies the power frequency and harmonic components of the surrounding environment in real time based on the dual-frequency matching data, generates a dynamic avoidance frequency point table to complete frequency hopping switching, thereby enhancing the reliability and foresight of dam safety monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy engineering, and more particularly to a dam safety monitoring system and method based on the Internet of Things. Background Art

[0002] Patent publication number CN111221287A discloses a dam safety monitoring system and method based on the Internet of Things. The system includes a cloud-side architecture, a pipe-side architecture, and an end-side architecture. It uses the "cloud-pipe-edge-end" system architecture to realize wireless network transmission between components in the system and cloud platform data services, effectively simplifying the structural setting of the safety monitoring system, and can effectively reduce the difficulty of data monitoring construction and maintenance investment during the construction and operation of the dam, thereby reducing labor costs.

[0003] The existing dam safety monitoring systems and methods have the following main problems:

[0004] Existing technologies fail to consider the temperature-dependent dynamic viscosity of water. Dynamic viscosity is sensitive to temperature, and dam water temperature varies with season, day, and depth, leading to measurement errors introduced by fixed viscosity values. Traditional methods also fail to account for factors such as sediment content and pollutant concentration. When water carries sediment or pollutants, its dynamic viscosity increases, leading to systematic deviations in dam flow velocity calculations. Piezoelectric ceramic sensors experience sensitivity loss under long-term water flow shock, but existing technologies fail to consider dynamic compensation of sensor status and viscosity parameters. Existing technologies generally use a fixed Strouhal number as a key parameter for water velocity calculations, assuming constant dynamic viscosity and fluid properties. This ignores the impact of environmental variables such as water temperature, sediment content, and pollutant concentration on dynamic viscosity.

[0005] The dynamic viscosity of water changes significantly with temperature, and the water temperature of the dam is affected by seasonal changes, the temperature difference between day and night, and the water depth, resulting in dynamic fluctuations in viscosity. In addition, when the sediment and pollutant content in the water increases, the dynamic viscosity of the water increases, further affecting the Reynolds number of the fluid, causing the Strouhal number to no longer maintain a fixed value. Piezoelectric ceramic sensors will experience sensitivity attenuation under long-term water flow impact, further exacerbating measurement errors. The Strouhal number actually changes with the Reynolds number, which in turn depends on the dynamic viscosity of the fluid; when the water temperature changes or the sediment content and pollutants in the water increase, the viscosity of the water will change, thereby affecting the Reynolds number, and thus causing the Strouhal number to deviate from the original set value;

[0006] In existing dam safety monitoring systems, Internet of Things technology is widely used to collect and transmit multi-dimensional observation data, but the following outstanding technical problems still exist: existing technologies mostly use fixed time windows to time align the collected data, which cannot fully adapt to the uneven distribution of observation point timestamps in multi-dimensional data, easily leading to alignment errors, and thus affecting the reliability of comprehensive data analysis. Existing alignment methods are mainly based on temporal proximity and lack comprehensive consideration of the spatial location information of observation points, resulting in physically unreasonable alignment results that may mask key monitoring signals. When observation points are sparsely distributed, it is difficult for existing systems to achieve complete alignment, resulting in monitoring gaps in certain areas and affecting global safety assessments. For areas with dense observation points, existing random numbers fail to dynamically adjust the alignment strategy according to data density, resulting in inefficient alignment or over-alignment, thereby wasting computing resources or reducing monitoring quality.

[0007] In view of this, the present invention proposes a dam safety monitoring system and method based on the Internet of Things to solve the above problems. Summary of the Invention

[0008] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a dam safety monitoring system based on the Internet of Things, comprising:

[0009] The vibration energy supply module installs a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing surface of the dam. The blunt-body structure guides the water flow to form a Karman vortex street, and monitors the water flow velocity of the dam in real time. The Karman vortex street shedding frequency is analyzed based on the FFT spectrum and matched with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data.

[0010] The multi-dimensional observation and acquisition module uses the generated Karman vortex street shedding frequency to convert the kinetic energy of the dam water into electrical energy. The converted electrical energy is used to power the preset multi-dimensional sensors, and the multi-dimensional observation data of the dam is collected based on the multi-dimensional sensors.

[0011] The three-dimensional hopping communication module, based on dual-frequency matching data, senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table to complete frequency hopping switching; it uses a seepage path prediction mechanism to predict the active seepage area in the next n time periods based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain node spatial location information;

[0012] The multi-source asynchronous alignment module uses the spatial location information of the nodes to control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals. It uses an elastic time window to dynamically align the dam's multi-dimensional observation data to obtain the aligned dam's multi-dimensional observation data.

[0013] The seepage stress early warning module calculates the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the preset gradient change rate threshold is exceeded, it automatically identifies the migration of the stress concentration area and generates an early warning signal.

[0014] Preferably, the method for real-time monitoring of dam water flow velocity comprises:

[0015] A cylindrical bluff structure is installed on the water-facing side of the dam to guide water flow around and form a Karman vortex street. A piezoelectric ceramic sensor is installed on the cylindrical bluff structure to detect in real time the periodic vibration signal generated by the water flow acting on the cylindrical bluff structure. The diameter and length of the cylindrical bluff structure are designed according to the preset target water flow velocity and scale, and it is made of water-resistant material.

[0016] The electro-ceramic sensor converts the monitored periodic vibration signal into an electrical signal, and collects the vibration frequency signal caused by the Karman vortex street shedding in real time; based on the relationship between the Karman vortex street shedding frequency and the diameter and Strouhal number of the cylindrical blunt body structure, the dam water flow velocity is calculated in real time.

[0017] Preferably, the method for acquiring the dual-frequency matching data includes:

[0018] Based on the electrical signal monitored by the piezoelectric ceramic sensor, the fast Fourier transform is used to perform spectrum analysis, convert the electrical signal into a frequency domain signal, and extract the vortex shedding frequency characteristic signal; the piezoelectric ceramic sensor has a preset natural resonant frequency;

[0019] The vortex shedding frequency obtained by spectrum analysis is matched with the preset natural resonant frequency. When the matching rate between the vortex shedding frequency and the preset natural resonant frequency reaches the preset matching rate threshold, the sensor signal amplitude is enhanced and a frequency resonance enhancement effect occurs, thereby obtaining dual-frequency matching data; the dual-frequency matching data includes the vortex shedding frequency, the resonant frequency of the piezoelectric ceramic sensor, the frequency matching rate, the signal amplitude and the resonance enhancement effect characteristics.

[0020] Preferably, the multi-dimensional observation data of the dam includes dam water flow parameter data, structural vibration parameter data, stress and strain data, seepage parameter data and environmental monitoring parameter data.

[0021] Preferably, the method for generating a dynamic avoidance frequency table to complete frequency hopping switching includes:

[0022] Through spectrum scanning technology, the electromagnetic environment of the current communication frequency band and adjacent communication frequency bands is monitored, and the amplitude and frequency characteristics of the harmonic components of the power frequency and integer multiples of the power frequency are collected. By comparing the preset power frequency and harmonic frequency templates with the real-time spectrum data, the location and intensity of various power frequency and harmonic signals in the environment are determined, and the frequency bands that interfere with the communication signal are screened out.

[0023] Based on the identification results, an avoidance frequency table is automatically generated. The avoidance frequency table lists the frequency points that need to be avoided and the bandwidth range of the frequency points. According to the avoidance frequency table, the preset frequency hopping algorithm is adjusted, and non-interference frequency bands are intelligently selected for signal transmission to achieve frequency hopping switching; the spectrum environment is continuously monitored, the avoidance frequency table is dynamically updated, and the frequency hopping strategy is adjusted.

[0024] Preferably, the method for obtaining the node spatial location information includes:

[0025] Train and build a prediction model for active seepage areas, including an input layer, an LSTM layer, and an output layer. The model's input layer is used to input historical dam flow velocities; the model's output layer is used to output active seepage areas within the next n time periods. The prediction model for active seepage areas is an LSTM model.

[0026] Nodes corresponding to active seepage areas are marked and defined as key nodes. In the communication protocol, key nodes in active seepage areas are preferentially selected as data relay paths. Based on the connection relationship between nodes and the location information of known nodes, a multidimensional scaling analysis algorithm is used to infer the spatial position of unknown nodes and obtain the spatial location information of the nodes.

[0027] Preferably, the method of controlling each node to generate a synchronization pulse based on a rubidium atomic clock and a GPS signal includes:

[0028] The spatial location information of each node is obtained through a multi-dimensional scaling analysis algorithm. Each node obtains a global unified time reference by receiving GPS signals. The time information transmitted by the GPS signal enables each node to achieve preliminary coarse synchronization, thereby establishing a unified time reference.

[0029] The rubidium atomic clock is used as the node's local stable clock source to generate a stable oscillation frequency and corresponding time signal. The time signal generated by the rubidium atomic clock is used in conjunction with the GPS reference time to correct and compensate the local time.

[0030] Based on the spatial location information of the node, the physical distance from each node to the reference synchronization anchor node is calculated, and the propagation delay compensation amount is calculated according to the propagation speed of electromagnetic waves in air or water. The node adjusts the clock signal output by the rubidium atomic clock according to its spatial location information.

[0031] Dynamic correction based on environmental monitoring data is introduced to adjust the propagation delay compensation to cope with changes in the propagation speed of electromagnetic waves in different environmental media; each node uses its own spatial position information, GPS signal and rubidium atomic clock output to generate a unified time reference and compensated synchronous pulse.

[0032] Preferably, the method for acquiring the aligned multi-dimensional observation data of the dam includes:

[0033] Preset time alignment benchmark for dam multi-dimensional observation data , as the time alignment reference point for the multidimensional observation data; for each dimension of the dam multidimensional observation data, set the initial time window width , the initial time window is Centered on Search nearby for the observation data point closest to the reference time in the observation data of this dimension;

[0034] And dynamically adjust the range of the time window according to the deviation of the observation data points in each dimension of observation data relative to the time alignment reference point; in the time alignment process, the timestamp of the observation data points in each observation data dimension is calculated in real time. Relative to the reference point Deviation ;in, Indicates the The timestamp of each observation data point; Represents the index of the observed data point; when the deviation is greater than the preset deviation threshold, the time window is dynamically expanded and the expansion ratio is set according to the expansion function;

[0035] Within the dynamically adjusted time window, for each dimension of observation data, filter and benchmark points The observation data points whose time difference is less than the preset time difference threshold are used as the observation values ​​of the dimension at the time alignment reference point; the time window width is adjusted according to the density of the aligned observation data points, and the number of observation data points is counted. If at the reference point If the number of observation data points in any nearby dimension is greater than or equal to the preset threshold, the observation value of the time-aligned reference point is calculated by weighted averaging; the weight of the weighted average is determined by the time difference and spatial distance difference between the observation data point and the reference point;

[0036] If at the reference point If the number of observation data points in any dimension of the recent observation data is less than the preset threshold, the observation value of the time alignment reference point is estimated by linear interpolation; the observation values ​​of each dimension in the multidimensional observation data under the time alignment reference point are uniformly aligned to form an aligned multidimensional observation data set.

[0037] Preferably, the method for automatically identifying the migration of stress concentration areas and generating early warning signals includes:

[0038] Based on the aligned multi-dimensional observation data of the dam, the stress tensor data at each observation data point is extracted. For each observation data point, the eigenvalue decomposition is performed to obtain the eigenvector and eigenvalue of the stress tensor data, from which the maximum principal stress direction vector and maximum principal stress value are determined. Each observation data point is selected and compared with adjacent observation data points. For each pair of adjacent observation data points, the angle between the maximum principal stress direction vectors of the two points is calculated to obtain the direction change angle.

[0039] The local rate of change of the maximum principal stress gradient is calculated by calculating the ratio of the direction change angle to the physical distance between the two points. The rate of change of the principal stress gradient in the direction of each observation data point is calculated and averaged for all adjacent points in the neighborhood of the observation data point. A threshold value for the rate of change of the principal stress gradient is preset. For each observation data point, if the rate of change of the principal stress gradient in the direction exceeds the preset threshold value, the observation data point is marked as a stress concentration abnormal point.

[0040] When it is detected that there are continuous observation data points in an adjacent area that meet the condition of exceeding the preset gradient change rate threshold, the area is identified as a potential stress concentration area; by analyzing the position changes of the stress concentration area at different time points, it is determined whether the stress concentration area has a migration trend; after identifying the stress concentration area and judging whether it has a migration trend, an early warning signal is automatically generated.

[0041] The IoT-based dam safety monitoring method includes:

[0042] S1. Install a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing side of the dam. Use the blunt-body structure to guide the water flow to form a Karman vortex street and monitor the water flow velocity in the dam in real time. Use FFT spectrum analysis to analyze the Karman vortex street shedding frequency and match it with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data.

[0043] S2. Using the generated vortex shedding frequency, the kinetic energy of the dam water is converted into electrical energy, and the converted electrical energy is used to power a preset multi-dimensional sensor, which then collects multi-dimensional observation data of the dam;

[0044] S3: Based on dual-frequency matching data, it senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table, and completes frequency hopping switching. It also uses a seepage path prediction mechanism to predict the active seepage area in the next n periods of time based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain the spatial location information of the nodes.

[0045] S4. Using the spatial location information of the nodes, control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals; use the elastic time window to dynamically align the dam multidimensional observation data to obtain the aligned dam multidimensional observation data;

[0046] S5. Calculate the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the gradient change rate exceeds the preset threshold, automatically identify the migration of the stress concentration area and generate an early warning signal.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] By introducing a dynamic correction model for water viscosity, temperature, and concentration, and combining it with the Reynolds number to correct the Strouhal number, the limitations of the constant viscosity assumption in traditional Karman vortex street water velocity calculations are overcome, significantly improving the accuracy of flow velocity monitoring in complex hydrological environments. Combined with water viscosity correction and sensor status monitoring, dynamic compensation for the sensitivity attenuation of piezoelectric ceramic sensors during long-term operation is achieved, further enhancing the stability and reliability of flow velocity monitoring in long-term operation. The Reynolds number-based Strouhal number correction mechanism enables flow velocity calculations to accurately reflect the current physical state of the water, significantly reducing deviations caused by environmental changes. Real-time response to environmental changes ensures that the monitoring system continuously outputs high-precision water velocity monitoring results under variable operating conditions, providing more reliable real-time data support for dam safety monitoring. By acquiring real-time water temperature and pollutant concentration sensor data, the water velocity calculation results are dynamically corrected, reducing system upgrade and deployment costs and enhancing water resource management and risk warning capabilities.

[0049] By dynamically adjusting the time window width, the time window can be automatically expanded or contracted according to the deviation of the observation points during the alignment process; it effectively adapts to the differences in time distribution in different observation dimensions, significantly improves the time alignment accuracy of multi-dimensional observation data, and reduces the analysis deviation caused by alignment errors. By introducing an expansion factor and an expansion function based on the deviation of the observation points, this solution can achieve adaptive adjustment of the time window; whether facing dense or sparse distribution of observation points, it can flexibly adjust the time window according to the characteristics of the actual observation data, avoiding the shortcomings of fixed time windows, and ensuring that the system can maintain efficient and accurate data alignment under different monitoring conditions.

[0050] When expanding the time window, a feedback adjustment mechanism for observation point density is implemented to prevent over-expansion in densely populated areas, which can lead to decreased alignment accuracy and wasted computational resources. By dynamically adjusting the expansion factor, the system's computational complexity and resource consumption can be reduced while maintaining high-precision alignment, improving overall system efficiency. A multidimensional observation data alignment and fusion mechanism based on time alignment reference points naturally adapts to the access and fusion of multidimensional and multi-source observation data, ensuring the system's excellent adaptability and processing capabilities in complex multidimensional data environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1This is a schematic structural diagram of the dam safety monitoring system based on the Internet of Things of the present invention;

[0052] Figure 2 This is a flow chart of the dam safety monitoring method based on the Internet of Things of the present invention;

[0053] Figure 3 This is a flow chart of the method for real-time monitoring of dam water flow velocity provided by the present invention. DETAILED DESCRIPTION

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0055] Example 1

[0056] See also Figure 1 and Figure 3 As shown, this embodiment 1 further illustrates the dam safety monitoring system based on the Internet of Things proposed by the present invention, including:

[0057] With the widespread application of the Internet of Things (IoT) technology, dam safety monitoring systems have gradually achieved real-time collection and transmission of multi-dimensional observation data, providing important support for dam operational status assessment and early warning. However, existing technologies still have several technical deficiencies that restrict the accuracy and reliability of dam safety monitoring.

[0058] Existing methods for monitoring dam water velocity use a cylindrical blunt-body structure to guide water flow around and form vortex streets. These structures, combined with piezoelectric ceramic sensors, detect the periodic vibration signals generated by the water flow in real time, and calculate the water velocity using the Strouhal number. The Strouhal number is typically assumed to be a fixed value, failing to account for the effects of changes in the water's dynamic viscosity and fluid properties. However, dam water viscosity is highly sensitive to temperature, which fluctuates dynamically with the seasons, diurnal temperature fluctuations, and water depth, leading to significant variations in viscosity. Furthermore, increased sediment content and pollutant concentrations in the water further increase its dynamic viscosity, thereby affecting the fluid's Reynolds number. This causes the Strouhal number to no longer maintain a fixed value, easily introducing systematic bias and reducing the accuracy of velocity calculations. Existing technologies fail to incorporate real-time corrections for changes in water viscosity, rendering water velocity monitoring results unreliable under complex hydrological conditions.

[0059] Piezoelectric ceramic sensors will experience sensitivity degradation under long-term water flow impact. Existing systems generally lack compensation mechanisms for dynamic changes in sensor status and environmental factors (such as temperature, sand content, and pollutant concentration), further exacerbating the accumulation of errors in monitoring data.

[0060] Existing IoT-based dam safety monitoring systems mostly use a fixed time window approach to time-align the collected multi-dimensional observation data. However, since the timestamps of observation points are often unevenly distributed, the fixed time window alignment approach is difficult to fully adapt to the temporal characteristics of the observation data, which can easily lead to alignment errors and affect the effectiveness of comprehensive data analysis. At the same time, existing alignment methods are mainly based on temporal proximity and ignore the spatial location information of the observation points. This may cause the alignment results to be physically irrational, or even mask key monitoring signals and reduce the accuracy of global safety assessments. In areas where observation points are sparsely distributed, the existing system has difficulty achieving complete data alignment, resulting in monitoring gaps; and in areas where observation points are densely distributed, a dynamic alignment strategy based on data density is not adopted, which can easily lead to inefficient alignment or over-alignment, wasting computing resources or reducing monitoring quality.

[0061] Therefore, in order to effectively solve the above problems, the present invention proposes a dam safety monitoring system based on the Internet of Things, comprising:

[0062] The vibration energy supply module installs a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing surface of the dam. The blunt-body structure guides the water flow to form a Karman vortex street, and monitors the water flow velocity of the dam in real time. The Karman vortex street shedding frequency is analyzed based on the FFT spectrum and matched with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data.

[0063] The multi-dimensional observation and acquisition module uses the generated Karman vortex street shedding frequency to convert the kinetic energy of the dam water into electrical energy. The converted electrical energy is used to power the preset multi-dimensional sensors, and the multi-dimensional observation data of the dam is collected based on the multi-dimensional sensors.

[0064] The three-dimensional hopping communication module, based on dual-frequency matching data, senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table to complete frequency hopping switching; it uses a seepage path prediction mechanism to predict the active seepage area in the next n time periods based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain node spatial location information;

[0065] The multi-source asynchronous alignment module uses the spatial location information of the nodes to control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals. It uses an elastic time window to dynamically align the dam's multi-dimensional observation data to obtain the aligned dam's multi-dimensional observation data.

[0066] The seepage stress early warning module calculates the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the preset gradient change rate threshold is exceeded, it automatically identifies the migration of the stress concentration area and generates an early warning signal.

[0067] Methods for real-time monitoring of dam water flow velocity include:

[0068] A cylindrical bluff structure is installed on the water-facing side of the dam to guide water flow around and form a Karman vortex street. A piezoelectric ceramic sensor is installed on the cylindrical bluff structure to detect in real time the periodic vibration signal generated by the water flow acting on the cylindrical bluff structure. The diameter and length of the cylindrical bluff structure are designed according to the preset target water flow velocity and scale, and it is made of water-resistant material.

[0069] The electro-ceramic sensor converts the monitored periodic vibration signal into an electrical signal, and collects the vibration frequency signal caused by the Karman vortex street shedding in real time; based on the relationship between the Karman vortex street shedding frequency and the diameter and Strouhal number of the cylindrical blunt body structure, the dam water flow velocity is calculated in real time.

[0070] The water flow velocity at the dam is ;in, represents the Karman vortex shedding frequency; represents the Strouhal number; Indicates the water flow velocity at the dam; represents the diameter of the cylindrical blunt body structure;

[0071] However, existing technologies fail to consider the temperature-dependent dynamic viscosity of water. Dynamic viscosity is sensitive to temperature, and dam water temperature varies with season, day, and depth, leading to measurement errors introduced by fixed viscosity values. Traditional methods also fail to consider factors such as the amount of sediment in the water and the concentration of pollutants. When water carries sediment or pollutants, the dynamic viscosity increases, leading to systematic deviations in the calculation of dam flow velocity. Piezoelectric ceramic sensors experience sensitivity loss under long-term water flow impact, but existing technologies fail to consider the dynamic compensation of sensor status and viscosity parameters. Existing technologies generally use a fixed Strouhal number as a key parameter for water velocity calculations, assuming that the dynamic viscosity and fluid properties of the water are constant, while ignoring the effects of environmental variables such as water temperature, sediment content, and pollutant concentration on the dynamic viscosity of the water.

[0072] The dynamic viscosity of water varies significantly with temperature, and the water temperature of the dam is affected by seasonal changes, the temperature difference between day and night, and the water depth, resulting in dynamic fluctuations in viscosity. In addition, when the sediment and pollutant content in the water increases, the dynamic viscosity of the water increases, further affecting the Reynolds number of the fluid, causing the Strouhal number to no longer maintain a fixed value. Therefore, the traditional Karman vortex street velocity calculation method based on a fixed Strouhal number is difficult to guarantee high accuracy under actual working conditions, and is prone to introducing systematic measurement errors, affecting the reliability of dam safety monitoring. At the same time, piezoelectric ceramic sensors will experience sensitivity attenuation under long-term water flow impact. Traditional methods lack a dynamic compensation mechanism for sensor status and environmental changes, further exacerbating measurement errors. The Strouhal number actually varies with the Reynolds number, which in turn depends on the dynamic viscosity of the fluid; when the water temperature changes or the sediment content and pollutants in the water increase, the viscosity of the water will change, thereby affecting the Reynolds number, and thus causing the Strouhal number to deviate from the original set value;

[0073] To this end, the Reynolds number and water viscosity correction are introduced to improve the accuracy and reliability of the dam water flow velocity calculation based on the Karman vortex street; the Reynolds number is calculated as: ;in, Indicates the density of water; Represents the dynamic viscosity function of water, which changes with temperature changes occur;

[0074] The dynamic viscosity function of water is: ;in, Indicates the preset reference water dynamic viscosity; represents the concentration influence coefficient; Indicates the concentration of pollutants; represents the temperature influence coefficient; Indicates water temperature;

[0075] Compared with existing technologies, the advantages are as follows: by introducing a dynamic correction model for water viscosity, temperature, and concentration, and combining it with the Reynolds number to correct the Strouhal number, the limitations of the constant viscosity assumption in traditional Karman vortex street water velocity calculations are overcome, significantly improving the accuracy of flow velocity monitoring in actual complex hydrological environments; by combining water viscosity correction and sensor status monitoring, dynamic compensation for the sensitivity attenuation of piezoelectric ceramic sensors during long-term operation is achieved, further improving the stability and reliability of flow velocity monitoring in long-term operation. The Strouhal number correction mechanism based on the Reynolds number enables flow velocity calculations to accurately reflect the current physical state of the water body, significantly reducing deviations caused by environmental changes; by responding to environmental changes in real time, ensuring that the monitoring system continuously outputs high-precision water velocity monitoring results under variable working conditions, providing more reliable real-time data support for dam safety monitoring; by acquiring water temperature and pollutant concentration sensor data in real time, the water velocity calculation results are dynamically corrected, reducing system upgrade and deployment costs and enhancing water resource management and risk warning capabilities.

[0076] The method for obtaining dual-frequency matching data includes:

[0077] Based on the electrical signal monitored by the piezoelectric ceramic sensor, the fast Fourier transform is used to perform spectrum analysis, convert the electrical signal into a frequency domain signal, and extract the vortex shedding frequency characteristic signal; the piezoelectric ceramic sensor has a preset natural resonant frequency;

[0078] It should be noted that the time-domain signal refers to the electrical signal directly collected by the piezoelectric ceramic sensor, which reflects the time-varying vibration of the bluff body structure caused by the water flow. Specifically, when the water flows around the bluff body and a Karman vortex street is formed, periodic vibrations are generated in the cylindrical bluff body. These mechanical vibrations are converted into corresponding voltage signals through the electromechanical coupling effect of the piezoelectric ceramic sensor. This voltage signal is a waveform that continuously changes with time, that is, a time-domain signal, which describes the change in vibration intensity over time. By performing a fast Fourier transform on this time-domain electrical signal, a frequency-domain signal can be obtained, and the main frequency components, such as the Karman vortex street shedding frequency, can be identified. Therefore, the time-domain signal is derived from the vibration electrical signal collected by the sensor in real time and is the basic data for the entire spectrum analysis.

[0079] The vortex shedding frequency obtained by spectrum analysis is matched with the preset natural resonant frequency. When the matching rate between the vortex shedding frequency and the preset natural resonant frequency reaches the preset matching rate threshold, the sensor signal amplitude is enhanced and a frequency resonance enhancement effect occurs, thereby obtaining dual-frequency matching data; the dual-frequency matching data includes the vortex shedding frequency, the resonant frequency of the piezoelectric ceramic sensor, the frequency matching rate, the signal amplitude and the resonance enhancement effect characteristics.

[0080] The multi-dimensional observation data of the dam include dam water flow parameter data, structural vibration parameter data, stress and strain data, seepage parameter data and environmental monitoring parameter data.

[0081] Water flow parameter data include dam water velocity, dam water pressure, dam water level, dam water turbulence and dam water temperature; structural vibration parameter data include dam structure vibration frequency, vibration acceleration and vibration displacement; stress and strain data include stress and strain changes in various parts of the dam; seepage parameter data include seepage rate and seepage pressure;

[0082] Environmental monitoring parameter data include atmospheric temperature, humidity, rainfall, wind speed and wind direction; multi-dimensional sensors include flow rate sensors, pressure sensors, water level sensors, strain sensors, temperature and humidity sensors, rain gauges and wind speed and direction sensors.

[0083] The method for generating a dynamic avoidance frequency table to complete frequency hopping switching includes:

[0084] Through spectrum scanning technology, the electromagnetic environment of the current communication frequency band and adjacent communication frequency bands is monitored, and the amplitude and frequency characteristics of the harmonic components of the power frequency and integer multiples of the power frequency are collected. By comparing the preset power frequency and harmonic frequency templates with the real-time spectrum data, the location and intensity of various power frequency and harmonic signals in the environment are determined, and the frequency bands that interfere with the communication signal are screened out.

[0085] Based on the identification results, an avoidance frequency table is automatically generated. The avoidance frequency table lists the frequency points that need to be avoided and the bandwidth range of the frequency points. According to the avoidance frequency table, the preset frequency hopping algorithm is adjusted, and non-interference frequency bands are intelligently selected for signal transmission to achieve frequency hopping switching; the spectrum environment is continuously monitored, the avoidance frequency table is dynamically updated, and the frequency hopping strategy is adjusted.

[0086] Methods for obtaining node spatial location information include:

[0087] Train and build a prediction model for active seepage areas, including an input layer, an LSTM layer, and an output layer. The model's input layer is used to input historical dam flow velocities; the model's output layer is used to output active seepage areas within the next n time periods. The prediction model for active seepage areas is an LSTM model.

[0088] Nodes corresponding to active seepage areas are marked and defined as key nodes. In the communication protocol, key nodes in active seepage areas are preferentially selected as data relay paths. Based on the connection relationship between nodes and the location information of known nodes, a multidimensional scaling analysis algorithm is used to infer the spatial position of unknown nodes and obtain the spatial location information of the nodes.

[0089] The method for controlling each node to generate a synchronization pulse based on a rubidium atomic clock and a GPS signal includes:

[0090] The spatial location information of each node is obtained through a multi-dimensional scaling analysis algorithm. Each node obtains a global unified time reference by receiving GPS signals. The time information transmitted by the GPS signal enables each node to achieve preliminary coarse synchronization, thereby establishing a unified time reference.

[0091] The rubidium atomic clock is used as the node's local stable clock source to generate a stable oscillation frequency and corresponding time signal. The time signal generated by the rubidium atomic clock is used in conjunction with the GPS reference time to correct and compensate the local time.

[0092] Based on the spatial location information of the node, the physical distance from each node to the reference synchronization anchor node is calculated, and the propagation delay compensation amount is calculated according to the propagation speed of electromagnetic waves in air or water. The node adjusts the clock signal output by the rubidium atomic clock according to its spatial location information.

[0093] Dynamic correction based on environmental monitoring data is introduced to adjust the propagation delay compensation to cope with changes in the propagation speed of electromagnetic waves in different environmental media; each node uses its own spatial position information, GPS signal and rubidium atomic clock output to generate a unified time reference and compensated synchronous pulse.

[0094] The method for obtaining aligned multi-dimensional observation data of the dam includes:

[0095] Preset time alignment benchmarks for dam multi-dimensional observation data , as the time alignment reference point for the multidimensional observation data; for each dimension of the dam multidimensional observation data, set the initial time window width , the initial time window is Centered on Search nearby for the observation data point closest to the reference time in the observation data of this dimension;

[0096] And dynamically adjust the range of the time window according to the deviation of the observation data points in each dimension of observation data relative to the time alignment reference point; in the time alignment process, the timestamp of the observation data points in each observation data dimension is calculated in real time. Relative to the reference point Deviation ;in, Indicates the The timestamp of each observation data point; Represents the index of the observed data point; when the deviation is greater than the preset deviation threshold, the time window is dynamically expanded and the expansion ratio is set according to the expansion function;

[0097] The extension function is: ;in, Indicates the width of the time window after dynamic expansion; Represents the expansion factor, which is used to control the expansion degree of the time window; Indicates the The deviation of each observation data point;

[0098] Within the dynamically adjusted time window, for each dimension of observation data, filter and benchmark points The observation data points whose time difference is less than the preset time difference threshold are used as the observation values ​​of the dimension at the time alignment reference point; the time window width is adjusted according to the density of the aligned observation data points, and the number of observation data points is counted. If at the reference point If the number of observation data points in any nearby dimension is greater than or equal to the preset threshold, the observation value of the time-aligned reference point is calculated by weighted averaging; the weight of the weighted average is determined by the time difference and spatial distance difference between the observation data point and the reference point;

[0099] If at the reference point If the number of observation data points in any dimension of the recent observation data is less than the preset threshold, the observation value of the time alignment reference point is estimated by linear interpolation; the observation values ​​of each dimension in the multidimensional observation data under the time alignment reference point are uniformly aligned to form an aligned multidimensional observation data set.

[0100] Considering the influence of density on observation data points, in order to avoid over-expansion in dense areas of observation points, the expansion factor is adjusted by the expansion adjustment function, which is: ; represents the adjusted expansion factor; where, Indicates the density of observation data points in the current time window; A factor that adjusts the density feedback strength.

[0101] Methods for automatically identifying the migration of stress concentration areas and generating early warning signals include:

[0102] Based on the aligned multi-dimensional observation data of the dam, the stress tensor data at each observation data point is extracted. For each observation data point, the eigenvalue decomposition is performed to obtain the eigenvector and eigenvalue of the stress tensor data, from which the maximum principal stress direction vector and maximum principal stress value are determined. Each observation data point is selected and compared with adjacent observation data points. For each pair of adjacent observation data points, the angle between the maximum principal stress direction vectors of the two points is calculated to obtain the direction change angle.

[0103] The local rate of change of the maximum principal stress gradient is calculated by calculating the ratio of the direction change angle to the physical distance between the two points. The rate of change of the principal stress gradient in the direction of each observation data point is calculated and averaged for all adjacent points in the neighborhood of the observation data point. A threshold value for the rate of change of the principal stress gradient is preset. For each observation data point, if the rate of change of the principal stress gradient in the direction exceeds the preset threshold value, the observation data point is marked as a stress concentration abnormal point.

[0104] When it is detected that there are continuous observation data points in an adjacent area that meet the condition of exceeding the preset gradient change rate threshold, the area is identified as a potential stress concentration area; by analyzing the position changes of the stress concentration area at different time points, it is determined whether the stress concentration area has a migration trend; after identifying the stress concentration area and judging whether it has a migration trend, an early warning signal is automatically generated.

[0105] The following problems existing in the existing technology are solved: In the existing dam safety monitoring system, the Internet of Things technology is generally used to collect and transmit multi-dimensional observation data, but the following outstanding technical problems still exist: the existing technology mostly uses a fixed time window to time align the collected data, which cannot fully adapt to the uneven distribution of observation point timestamps in multi-dimensional data, easily leading to alignment errors, and thus affecting the reliability of comprehensive data analysis. The existing alignment method is mainly based on temporal proximity and lacks comprehensive consideration of the spatial location information of the observation points, resulting in the alignment results being physically unreasonable and possibly masking key monitoring signals. When the observation points are sparsely distributed, it is difficult for the existing system to achieve complete alignment, resulting in monitoring gaps in certain areas, affecting the global safety assessment. For areas with dense observation points, the existing random numbers fail to dynamically adjust the alignment strategy according to the data density, resulting in inefficient alignment or over-alignment, thereby wasting computing resources or reducing monitoring quality.

[0106] Compared with the existing technology, the advantages are as follows: by dynamically adjusting the time window width, the time window can be automatically expanded or contracted according to the deviation of the observation points during the alignment process; it can effectively adapt to the differences in time distribution in different observation dimensions, significantly improve the time alignment accuracy of multi-dimensional observation data, and reduce the analysis deviation caused by alignment errors. By introducing an expansion factor and an expansion function based on the deviation of the observation points, this solution can achieve adaptive adjustment of the time window; whether facing dense or sparse distribution of observation points, it can flexibly adjust the time window according to the characteristics of the actual observation data, avoiding the shortcomings of fixed time windows, and ensuring that the system can maintain efficient and accurate data alignment under different monitoring conditions;

[0107] When expanding the time window, a feedback adjustment mechanism for observation point density is implemented to prevent over-expansion in densely populated areas, which can lead to decreased alignment accuracy and wasted computational resources. By dynamically adjusting the expansion factor, the system's computational complexity and resource consumption can be reduced while maintaining high-precision alignment, improving overall system efficiency. A multidimensional observation data alignment and fusion mechanism based on time alignment reference points naturally adapts to the access and fusion of multidimensional and multi-source observation data, ensuring the system's excellent adaptability and processing capabilities in complex multidimensional data environments.

[0108] The preset gradient change rate threshold is set by the staff by taking the average of multiple preset gradient changes as the preset gradient change rate threshold; similarly, the preset matching rate threshold, preset deviation threshold, preset time difference threshold and preset quantity threshold are set.

[0109] This embodiment overcomes the limitations of the constant viscosity assumption in traditional Karman vortex street water velocity calculations by introducing a dynamic correction model for water viscosity, temperature, and concentration, combined with the Reynolds number to correct the Strouhal number. This significantly improves the accuracy of velocity monitoring in complex hydrological environments. Combining water viscosity correction with sensor status monitoring, it dynamically compensates for the sensitivity degradation of piezoelectric ceramic sensors during long-term operation, further enhancing the stability and reliability of velocity monitoring in long-term operation. The Reynolds number-based Strouhal number correction mechanism enables velocity calculations to accurately reflect the current physical state of the water, significantly reducing deviations caused by environmental changes. It responds to environmental changes in real time, ensuring that the monitoring system continuously outputs high-precision water velocity monitoring results under variable operating conditions, providing more reliable real-time data support for dam safety monitoring. By acquiring real-time water temperature and pollutant concentration sensor data, it dynamically corrects water velocity calculation results, reducing system upgrade and deployment costs and enhancing water resource management and risk warning capabilities.

[0110] By dynamically adjusting the time window width, the time window can be automatically expanded or contracted according to the deviation of the observation points during the alignment process; it effectively adapts to the differences in time distribution in different observation dimensions, significantly improves the time alignment accuracy of multi-dimensional observation data, and reduces the analysis deviation caused by alignment errors. By introducing an expansion factor and an expansion function based on the deviation of the observation points, this solution can achieve adaptive adjustment of the time window; whether facing dense or sparse distribution of observation points, it can flexibly adjust the time window according to the characteristics of the actual observation data, avoiding the shortcomings of fixed time windows, and ensuring that the system can maintain efficient and accurate data alignment under different monitoring conditions.

[0111] When expanding the time window, a feedback adjustment mechanism for observation point density is implemented to prevent over-expansion in densely populated areas, which can lead to decreased alignment accuracy and wasted computational resources. By dynamically adjusting the expansion factor, the system's computational complexity and resource consumption can be reduced while maintaining high-precision alignment, improving overall system efficiency. A multidimensional observation data alignment and fusion mechanism based on time alignment reference points naturally adapts to the access and fusion of multidimensional and multi-source observation data, ensuring the system's excellent adaptability and processing capabilities in complex multidimensional data environments.

[0112] Example 2

[0113] See also Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1. A dam safety monitoring method based on the Internet of Things is provided, including:

[0114] S1. Install a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing side of the dam. Use the blunt-body structure to guide the water flow to form a Karman vortex street and monitor the water flow velocity in the dam in real time. Use FFT spectrum analysis to analyze the Karman vortex street shedding frequency and match it with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data.

[0115] S2. Using the generated vortex shedding frequency, the kinetic energy of the dam water is converted into electrical energy, and the converted electrical energy is used to power a preset multi-dimensional sensor, which then collects multi-dimensional observation data of the dam;

[0116] S3: Based on dual-frequency matching data, it senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table, and completes frequency hopping switching. It also uses a seepage path prediction mechanism to predict the active seepage area in the next n periods of time based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain the spatial location information of the nodes.

[0117] S4. Using the spatial location information of the nodes, control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals; use the elastic time window to dynamically align the dam multidimensional observation data to obtain the aligned dam multidimensional observation data;

[0118] S5. Calculate the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the gradient change rate exceeds the preset threshold, automatically identify the migration of the stress concentration area and generate an early warning signal.

[0119] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0120] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. The dam safety monitoring system based on the Internet of Things is characterized by: include: The vibration energy supply module is equipped with a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing side of the dam. The blunt-body structure guides the water flow to form a Karman vortex street and monitors the water flow velocity of the dam in real time. Based on the relationship between the Karman vortex shedding frequency and the diameter and Strouhal number of the cylindrical blunt body structure, the dam water velocity is calculated in real time. A dynamic correction model based on water viscosity, temperature, and concentration is introduced to correct the Strouhal number in combination with the Reynolds number. The Reynolds number is calculated as: ;in, Indicates the density of water; represents the dynamic viscosity function of water; Indicates the water flow velocity at the dam; represents the diameter of the cylindrical blunt body structure; The dynamic viscosity function of water is: ;in, Indicates the preset reference water dynamic viscosity; represents the concentration influence coefficient; Indicates the concentration of pollutants; represents the temperature influence coefficient; Indicates water temperature; The Karman vortex shedding frequency is analyzed based on the FFT spectrum and matched with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data. The multi-dimensional observation and acquisition module uses the generated Karman vortex street shedding frequency to convert the kinetic energy of the dam water into electrical energy. The converted electrical energy is used to power the preset multi-dimensional sensors, and the multi-dimensional observation data of the dam is collected based on the multi-dimensional sensors. The three-dimensional hopping communication module, based on dual-frequency matching data, senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table to complete frequency hopping switching; it uses a seepage path prediction mechanism to predict the active seepage area in the next n time periods based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain node spatial location information; The multi-source asynchronous alignment module uses the spatial location information of the nodes to control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals. It uses an elastic time window to dynamically align the dam's multi-dimensional observation data to obtain the aligned dam's multi-dimensional observation data. The method for obtaining the aligned dam multi-dimensional observation data includes: Preset time alignment benchmark for dam multi-dimensional observation data , as the time alignment reference point for multidimensional observation data; for each dimension of observation data in the dam multidimensional observation data, set the initial time window width , the initial time window is Centered on Search nearby for the observation data point closest to the reference time in the observation data of this dimension; And dynamically adjust the range of the time window according to the deviation of the observation data points in each dimension of observation data relative to the time alignment reference point; in the time alignment process, the timestamp of the observation data points in each observation data dimension is calculated in real time. Relative to the reference point Deviation ;in, Indicates the The timestamp of each observation data point; Represents the index of the observed data point; when the deviation is greater than the preset deviation threshold, the time window is dynamically expanded and the expansion ratio is set according to the expansion function; Within the dynamically adjusted time window, for each dimension of observation data, filter and benchmark points The observation data points whose time difference is less than the preset time difference threshold are used as the observation values ​​of the dimension at the time alignment reference point; the time window width is adjusted according to the density of the aligned observation data points, and the number of observation data points is counted. If at the reference point If the number of observation data points in any nearby dimension is greater than or equal to the preset threshold, the observation value of the time-aligned reference point is calculated by weighted averaging; the weight of the weighted average is determined by the time difference and spatial distance difference between the observation data point and the reference point; If at the reference point If the number of observation data points in any dimension of the nearest observation data is less than the preset threshold, the observation value of the time alignment reference point is estimated by linear interpolation; the observation values ​​of each dimension in the multidimensional observation data under the time alignment reference point are uniformly aligned to form an aligned multidimensional observation data set; The seepage stress early warning module calculates the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the preset gradient change rate threshold is exceeded, it automatically identifies the migration of the stress concentration area and generates an early warning signal.

2. The dam safety monitoring system based on the Internet of Things according to claim 1 is characterized in that: The method for real-time monitoring of dam water flow velocity comprises: A cylindrical bluff structure is installed on the water-facing side of the dam to guide water flow around and form a Karman vortex street. A piezoelectric ceramic sensor is installed on the cylindrical bluff structure to detect in real time the periodic vibration signal generated by the water flow acting on the cylindrical bluff structure. The diameter and length of the cylindrical bluff structure are designed according to the preset target water flow velocity and scale, and it is made of water-resistant material. The piezoelectric ceramic sensor converts the monitored periodic vibration signal into an electrical signal and collects the vibration frequency signal caused by Karman vortex shedding in real time.

3. The dam safety monitoring system based on the Internet of Things according to claim 2 is characterized in that: The method for obtaining dual-frequency matching data includes: Based on the electrical signal monitored by the piezoelectric ceramic sensor, the fast Fourier transform is used to perform spectrum analysis, convert the electrical signal into a frequency domain signal, and extract the vortex shedding frequency characteristic signal; the piezoelectric ceramic sensor has a preset natural resonant frequency; The vortex shedding frequency obtained by spectrum analysis is matched with the preset natural resonant frequency. When the matching rate between the vortex shedding frequency and the preset natural resonant frequency reaches the preset matching rate threshold, the sensor signal amplitude is enhanced and a frequency resonance enhancement effect occurs, thereby obtaining dual-frequency matching data; the dual-frequency matching data includes the vortex shedding frequency, the resonant frequency of the piezoelectric ceramic sensor, the frequency matching rate, the signal amplitude and the resonance enhancement effect characteristics.

4. The dam safety monitoring system based on the Internet of Things according to claim 3 is characterized in that: The multi-dimensional observation data of the dam includes dam water flow parameter data, structural vibration parameter data, stress and strain data, seepage parameter data and environmental monitoring parameter data.

5. The dam safety monitoring system based on the Internet of Things according to claim 4 is characterized in that: The method for generating a dynamic avoidance frequency table to complete frequency hopping switching includes: Through spectrum scanning technology, the electromagnetic environment of the current communication frequency band and adjacent communication frequency bands is monitored, and the amplitude and frequency characteristics of the harmonic components of the power frequency and integer multiples of the power frequency are collected. By comparing the preset power frequency and harmonic frequency templates with the real-time spectrum data, the location and intensity of various power frequency and harmonic signals in the environment are determined, and the frequency bands that interfere with the communication signal are screened out. Based on the identification results, an avoidance frequency table is automatically generated. The avoidance frequency table lists the frequency points that need to be avoided and the bandwidth range of the frequency points. According to the avoidance frequency table, the preset frequency hopping algorithm is adjusted, and non-interference frequency bands are intelligently selected for signal transmission to achieve frequency hopping switching; the spectrum environment is continuously monitored, the avoidance frequency table is dynamically updated, and the frequency hopping strategy is adjusted.

6. The dam safety monitoring system based on the Internet of Things according to claim 5 is characterized in that: The method for obtaining the node spatial position information includes: Train and build a prediction model for active seepage areas, including an input layer, an LSTM layer, and an output layer. The model's input layer is used to input historical dam flow velocities; the model's output layer is used to output active seepage areas within the next n time periods. The prediction model for active seepage areas is an LSTM model. Nodes corresponding to active seepage areas are marked and defined as key nodes. In the communication protocol, key nodes in active seepage areas are preferentially selected as data relay paths. Based on the connection relationship between nodes and the location information of known nodes, a multidimensional scaling analysis algorithm is used to infer the spatial position of unknown nodes and obtain the spatial location information of the nodes.

7. The dam safety monitoring system based on the Internet of Things according to claim 6 is characterized in that: The method of controlling each node to generate a synchronization pulse based on a rubidium atomic clock and a GPS signal includes: The spatial location information of each node is obtained through a multi-dimensional scaling analysis algorithm. Each node obtains a global unified time reference by receiving GPS signals. The time information transmitted by the GPS signal enables each node to achieve preliminary coarse synchronization, thereby establishing a unified time reference. The rubidium atomic clock is used as the node's local stable clock source to generate a stable oscillation frequency and corresponding time signal. The time signal generated by the rubidium atomic clock is used in conjunction with the GPS reference time to correct and compensate the local time. Based on the spatial location information of the node, the physical distance from each node to the reference synchronization anchor node is calculated, and the propagation delay compensation amount is calculated according to the propagation speed of electromagnetic waves in air or water. The node adjusts the clock signal output by the rubidium atomic clock according to its spatial location information. Dynamic correction based on environmental monitoring data is introduced to adjust the propagation delay compensation to cope with changes in the propagation speed of electromagnetic waves in different environmental media; each node uses its own spatial position information, GPS signal and rubidium atomic clock output to generate a unified time reference and compensated synchronous pulse.

8. The dam safety monitoring system based on the Internet of Things according to claim 7 is characterized in that: The method for automatically identifying the migration of stress concentration areas and generating early warning signals includes: Based on the aligned multi-dimensional observation data of the dam, the stress tensor data at each observation data point is extracted. For each observation data point, the eigenvalue decomposition is performed to obtain the eigenvector and eigenvalue of the stress tensor data, from which the maximum principal stress direction vector and maximum principal stress value are determined. Each observation data point is selected and compared with adjacent observation data points. For each pair of adjacent observation data points, the angle between the maximum principal stress direction vectors of the two points is calculated to obtain the direction change angle. The local rate of change of the maximum principal stress gradient is calculated by calculating the ratio of the direction change angle to the physical distance between the two points. The rate of change of the principal stress gradient in the direction of each observation data point is calculated and averaged for all adjacent points in the neighborhood of the observation data point. A threshold value for the rate of change of the principal stress gradient is preset. For each observation data point, if the rate of change of the principal stress gradient in the direction exceeds the preset threshold value, the observation data point is marked as a stress concentration abnormal point. When it is detected that there are continuous observation data points in an adjacent area that meet the condition of exceeding the preset gradient change rate threshold, the area is identified as a potential stress concentration area; by analyzing the position changes of the stress concentration area at different time points, it is determined whether the stress concentration area has a migration trend; after identifying the stress concentration area and judging whether it has a migration trend, an early warning signal is automatically generated.

9. A dam safety monitoring method based on the Internet of Things, implemented by the dam safety monitoring system based on the Internet of Things according to any one of claims 1 to 8, characterized in that: include: S1. Install a cylindrical blunt-body structure and a piezoelectric ceramic sensor on the water-facing side of the dam. Use the blunt-body structure to guide the water flow to form a Karman vortex street and monitor the water flow velocity in the dam in real time. Use FFT spectrum analysis to analyze the Karman vortex street shedding frequency and match it with the resonant frequency of the piezoelectric ceramic sensor to obtain dual-frequency matching data. S2. Using the generated vortex shedding frequency, the kinetic energy of the dam water is converted into electrical energy, and the converted electrical energy is used to power a preset multi-dimensional sensor, which then collects multi-dimensional observation data of the dam; S3: Based on dual-frequency matching data, it senses and identifies the power frequency and harmonic components of the surrounding environment in real time, generates a dynamic avoidance frequency table, and completes frequency hopping switching. It also uses a seepage path prediction mechanism to predict the active seepage area in the next n periods of time based on the dam water velocity, and prioritizes nodes in the active seepage area as relay paths to obtain the spatial location information of the nodes. S4. Using the spatial location information of the nodes, control each node to generate synchronization pulses based on the rubidium atomic clock and GPS signals; use the elastic time window to dynamically align the dam multidimensional observation data to obtain the aligned dam multidimensional observation data; S5. Calculate the gradient change rate of the principal stress direction of the aligned dam multi-dimensional observation data. When the gradient change rate exceeds the preset threshold, automatically identify the migration of the stress concentration area and generate an early warning signal.

Images