A method and system for predicting the freeze resistance of a pipe of an automatic sprinkler system

Abstract

The application provides a pipeline anti-freezing prediction method and system of an automatic water spray fire extinguishing system. In a cold environment monitoring area, ice crystal growth vibration signals are captured by an acoustic emission sensor, and real-time icing rate data is synchronously associated. A multi-scale long short-term memory network is input, and a short-time span memory unit and a long-time span memory unit are used to output composite freezing features. The transmission power and modulation rate of a LoRaWAN node are adaptively adjusted, high-priority composite freezing features and corresponding pipe segment geographic location codes are transmitted to a remote control center, ice crystal dynamic parameters and material stress parameters in the composite freezing features are matched based on a pipeline material ice expansion coefficient database, the rupture risk probability of a specific pipe segment is calculated, and a pipeline electric heat tracing gradient warming strategy corresponding to the rupture risk probability is activated. The technical scheme provided by the application accurately predicts and effectively reduces the rupture risk of the pipeline of the automatic water spray fire extinguishing system in a cold environment.

Description

Technical Field

[0001] The present application relates to the field of data prediction technology, and in particular to a method and system for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system. Background Art

[0002] In wet automatic sprinkler systems operating in cold environments, ice formation caused by phase changes in the fluid inside the pipes can cause material deformation and even rupture, posing a safety hazard. Traditional antifreeze measures struggle to monitor the dynamic process of ice crystal growth in real time, particularly the low-frequency cumulative effects of microsecond-scale stress changes and material fatigue deformation caused by instantaneous ice crystal rupture. Therefore, an intelligent monitoring and control technology is needed that can accurately capture the dynamic characteristics of ice formation inside pipes, assess rupture risks, and dynamically adjust protection strategies to ensure the safe operation of fire protection pipelines.

[0003] One current technical solution to this problem involves combining a static threshold control system with an electric heating system based on temperature sensors. This solution uses temperature sensors placed on the pipe surface to monitor the ambient temperature. When the temperature falls below a preset threshold (e.g., 4°C), the electric heating system activates to uniformly heat the pipe to prevent freezing. Some solutions also incorporate insulation wrapping to slow heat loss and periodic pressure testing and water release to reduce the risk of freezing.

[0004] Existing solutions rely on temperature thresholds to trigger passive heating, but are unable to sense dynamic characteristics such as ice crystal growth rate and material stress changes. For example, temperature sensors cannot capture the instantaneous vibration signals of ice crystal breakage or the low-frequency oscillations of material fatigue accumulation, making it impossible to assess the actual ice thickness and pipeline deformation risk. Uniform heating lacks targeted regulation of high-risk pipe sections, potentially resulting in energy waste. Static threshold control cannot predict critical icing states, which may result in the pipeline entering a high-risk stage when protective measures are activated, increasing the risk of rupture in automatic sprinkler system pipes in cold environments. Summary of the Invention

[0005] The present application provides a method and system for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system, which are used to solve the problem of high rupture risk of pipes in wet automatic sprinkler fire extinguishing systems in cold environments in the prior art.

[0006] In a first aspect, the present application provides a method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system, comprising:

[0007] Acoustic emission sensors are used to collect ice crystal growth vibration signals caused by internal fluid phase changes on the walls of sprinkler pipes, and these signals are then correlated with real-time freezing rate data collected by temperature sensors mounted on the pipe surfaces.

[0008] The ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long short-term memory network. The short-time span memory unit captures the microsecond pulse waveform caused by the instantaneous rupture of ice crystals. The long-time span memory unit tracks the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipeline material, and outputs a composite freezing feature.

[0009] Adaptively adjust the transmit power and modulation rate of the LoRaWAN node based on the spatial distribution of the pipeline network and the priority of the composite frozen feature, encode the high-priority composite frozen feature and the corresponding pipeline segment geographic location into an emergency data frame, and preferentially transmit it to the remote control center via multi-hop routing;

[0010] Based on the ice expansion coefficient database of pipeline materials pre-stored in the remote control center, the ice crystal dynamic parameters and material stress parameters in the composite freezing characteristics are matched, the rupture risk probability of a specific pipe section at a critical ice thickness is calculated, the pipeline antifreeze performance is predicted based on the rupture risk probability, and the pipeline heating strategy corresponding to the rupture risk probability is activated according to the antifreeze performance prediction result.

[0011] Optionally, the ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long short-term memory network, and the microsecond pulse waveform caused by the instantaneous rupture of ice crystals is captured by a short-time span memory unit, while the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipeline material is tracked by a long-time span memory unit, and the composite freezing feature is output, including:

[0012] Before the ice crystal growth vibration signal is input into the multi-scale long short-term memory network, the ice crystal growth vibration signal is subjected to vibration component elimination processing of non-phase change interference, and the vibration component coaxial with the axial stress direction of the pipeline is retained;

[0013] Inputting the vibration component into the time-frequency decomposition layer to separate the high-frequency sub-signal representing the instantaneous breakage of ice crystals and the low-frequency sub-signal reflecting the plastic deformation of the material, and converting the real-time freezing rate data into freezing dynamic coefficients aligned with the timestamps of each sub-signal;

[0014] The high-frequency sub-signal is input into a short-time span memory unit for microsecond waveform contour tracking to capture the phase mutation points between adjacent ice crystal breakage events. The low-frequency sub-signal and the icing dynamic coefficient are input into a long-time span memory unit for oscillation mode correlation to establish an implicit correlation between the change in pipe cross-section ellipticity and the stress relaxation rate.

[0015] Dynamic weight allocation is performed on the phase mutation point and the implicit association, and combined with the pre-stored Young's modulus parameters and thermal expansion coefficient parameters of the pipeline to generate a composite freezing feature including the ice crystal distribution gradient and the material yield strength attenuation trend.

[0016] Optionally, according to the spatial distribution of the pipeline network and the priority of the composite frozen feature, the transmission power and modulation rate of the LoRaWAN node are adaptively adjusted, and the high-priority composite frozen feature and the corresponding pipeline segment geographic location are encoded into an emergency data frame, which is preferentially transmitted to the remote control center via multi-hop routing, including:

[0017] The pipeline rupture risk probability and stress concentration factor in the composite freezing feature are analyzed, and the dynamic transmission path loss threshold of each LoRaWAN node is calculated by combining the distance parameters of adjacent nodes in the pipeline distribution topology and the historical signal interference parameters.

[0018] Dynamically allocate transmit power levels and orthogonal modulation indexes based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, and synchronously constrain the maximum adjustable power range in conjunction with the node residual energy parameter;

[0019] The high-priority composite freezing features and the corresponding pipe segment geographic locations are categorized and coded according to pipe diameter specifications. A frame header containing the pipe segment identifier and a checksum field for the ice crystal growth azimuth angle is constructed to generate an emergency data frame with a topology-aware tag.

[0020] Based on the pre-stored metal obstacle attenuation map in the pipeline distribution topology, a relay node sequence with minimum path fading is selected, and multi-hop relay transmission of the emergency data frame is achieved by alternately using ground reflection wave and direct wave propagation modes.

[0021] Optionally, dynamically allocating transmit power levels and orthogonal modulation indexes based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, and synchronously constraining the maximum adjustable power range in conjunction with the node residual energy parameter, includes:

[0022] Establishing a transmission power allocation model, converting the change rate of the ice crystal distribution gradient in the composite freezing feature and the node residual energy parameter through thermodynamic expansion margin to generate a power allocation reference value;

[0023] performing phase sensitivity calibration on the orthogonal modulation index according to the dynamic transmission path loss threshold, and triggering a compensation mechanism for shifting the orthogonal modulation index in the direction of anti-multipath interference when the dynamic transmission path loss threshold exceeds the theoretical attenuation value of the pipeline insulation layer, and outputting a time-varying compensation coefficient;

[0024] Injecting the time-varying compensation coefficient into the transmit power allocation model, triggering a transient overload transmit power pulse based on the power allocation reference value when the frost layer thickness exceeds a critical value of the sealing gap at the metal flange connection, and constraining the transient overload transmit power pulse to be within a maximum adjustable power range allowed by the node's residual energy parameter;

[0025] The orthogonal modulation rope after phase sensitivity calibration and the power level of the transient overload transmission power pulse are time-slot interleaved and coded to generate an adaptive communication parameter configuration table adapted to the Doppler frequency shift characteristics at the pipeline elbow.

[0026] Optionally, injecting the time-varying compensation coefficient into the transmit power allocation model, triggering a transient overload transmit power pulse based on the power allocation reference value when the frost layer thickness exceeds a critical value of the sealing gap at the metal flange connection, and constraining the transient overload transmit power pulse to be within a maximum adjustable power range allowed by the node residual energy parameter, includes:

[0027] The time-varying compensation coefficient is input into the transmission power allocation model, and the dynamic overload power margin parameter is generated by combining the real-time growth rate of the frost layer thickness and the phase change latent heat parameter of the pipeline metal material. The node corresponding to the pipe section where the frost layer thickness exceeds the critical value of the sealing gap automatically activates the margin parameter compensation channel;

[0028] Based on the rate of change of the ice crystal distribution gradient, the dynamic overload power margin parameter is subjected to azimuth weighting processing through the margin parameter compensation channel, and the waveform envelope of the transient overload transmission power pulse is phase modulated using the pipeline axial stress distribution mode to generate a pulse parameter set that matches the check field of the ice crystal growth azimuth angle;

[0029] Based on the creep parameters of the sealing material at the metal flange connection and the bolt preload attenuation coefficient, a time-domain energy distribution constraint condition is established so that the pulse parameter set forms a non-uniform energy density distribution within the maximum adjustable power range allowed by the node residual energy parameter;

[0030] The non-uniform energy density distribution and the dynamic overload power margin parameter are time-interleaved to generate a transient overload emission instruction set carrying a stress concentration factor correction parameter at a metal flange connection.

[0031] Optionally, based on the rate of change of the ice crystal distribution gradient, performing azimuthally weighted processing on the dynamic overload power margin parameter through the margin parameter compensation channel, adopting a pipeline axial stress distribution pattern to phase modulate the waveform envelope of the transient overload transmission power pulse, and generating a pulse parameter set that matches the check field of the ice crystal growth azimuth, including:

[0032] Decomposing the change rate of the ice crystal distribution gradient into the pipeline axial ice crystal growth rate and radial penetration rate components, and performing stress relaxation weight distribution on the two components in combination with the residual stress parameters at the pipeline weld to generate the ice crystal azimuthal anisotropy compensation coefficient;

[0033] The anisotropic thermal conductivity of the metal material of the pipeline is input through the margin parameter compensation channel, and the dynamic overload power margin parameter is corrected by heat flux vector in combination with the ice crystal azimuth anisotropy compensation coefficient to generate an azimuth weighted matrix aligned with the main direction of ice crystal growth;

[0034] Phase modulation is performed on the waveform envelope of the transient overload transmission power pulse based on the pipeline ellipticity deformation parameter, and a pulse amplitude attenuation gradient template is generated using the circumferential distribution characteristics of the bolt array at the metal flange connection;

[0035] The azimuth weighted matrix is ​​tensor-fused with the pulse amplitude attenuation gradient template, and combined with the pipeline circumferential temperature gradient data recorded in the check field of the ice crystal growth azimuth, a pulse parameter set matching the check field of the ice crystal growth azimuth is generated.

[0036] Optionally, the calculating of the rupture risk probability of a specific pipe section at a critical ice thickness based on a database of ice expansion coefficients of pipeline materials pre-stored in the remote control center and matching ice crystal dynamic parameters with material stress parameters in the composite freezing feature includes:

[0037] Decomposing the ice crystal dynamic parameters in the composite freezing feature into the ice crystal axial growth rate and the ice crystal radial penetration rate, and combining the pipe material low-temperature phase change characteristic parameters and the yield strength attenuation gradient in the pipeline material ice expansion coefficient database pre-stored in the remote control center, to generate an ice layer thickness-pipe material stress coupling incremental distribution map;

[0038] Based on the historical frost heave deformation records of the pipe section stored in the pipeline material ice expansion coefficient database, the maximum principal stress direction in the ice layer thickness-pipe stress coupling incremental distribution map is matched, and combined with the material stress parameters corresponding to the actual service life of the pipeline, a dynamic ice expansion stress concentration factor is generated;

[0039] Based on the proportional relationship between the critical ice thickness in the composite freezing feature and the dynamic ice expansion stress concentration factor, the anisotropic expansion coefficient of the pipeline material and the flange assembly prestress parameters are introduced to calculate the rupture risk probability of a specific pipe section under the proportional relationship.

[0040] Optionally, predicting pipeline antifreeze performance based on the rupture risk probability, and activating a pipeline heating strategy corresponding to the rupture risk probability according to the antifreeze performance prediction result, includes:

[0041] Based on the rupture risk probability, combined with the low-temperature brittle transition temperature of the pipeline material and the ice crystal growth stress field distribution, the pipeline section is divided into frost resistance grades, and a pipeline frost resistance prediction result including an frost resistance grade identifier is generated;

[0042] According to the antifreeze grade identifier, a pre-stored pipeline heating strategy library is matched to extract the heating rate and target temperature corresponding to the antifreeze grade of the pipe section;

[0043] Based on the spatial correlation between the pipeline axial stress distribution pattern and the ice crystal penetration path, the electric heating areas are activated in descending order according to the antifreeze level of the pipeline section, and a segmented progressive heating instruction set is generated;

[0044] The segmented progressive heating instruction set is executed, and the electric heating power output is dynamically adjusted in combination with the heating rate, the target temperature and the real-time frost thickness data on the pipeline surface to generate a pipeline heating control signal that matches the rupture risk probability.

[0045] In a second aspect, the present application provides a system for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system, comprising:

[0046] A capture module is used to collect ice crystal growth vibration signals caused by internal fluid phase changes on the pipe wall of the automatic sprinkler system through acoustic emission sensors, and synchronously correlate them with real-time freezing rate data collected by temperature sensors mounted on the pipe surface;

[0047] An output module is configured to input the ice crystal growth vibration signal and the real-time freezing rate data into a multi-scale long short-term memory network, capture the microsecond pulse waveform caused by the instantaneous rupture of ice crystals through short-time span memory units, and simultaneously track the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of pipeline materials through long-time span memory units, and output a composite freezing feature;

[0048] A transmission module is used to adaptively adjust the transmission power and modulation rate of the LoRaWAN node according to the spatial distribution of the pipeline network and the priority of the composite frozen feature, encode the high-priority composite frozen feature and the corresponding pipeline segment geographic location into an emergency data frame, and preferentially transmit it to the remote control center via multi-hop routing;

[0049] The activation module is used to match the ice crystal dynamic parameters and material stress parameters in the composite freezing characteristics based on the ice expansion coefficient database of the pipeline material pre-stored in the remote control center, calculate the rupture risk probability of a specific pipe section at a critical ice thickness, predict the pipeline antifreeze based on the rupture risk probability, and activate the pipeline heating strategy corresponding to the rupture risk probability according to the antifreeze prediction result.

[0050] In a third aspect, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a method for predicting the frost resistance of pipes of an automatic sprinkler fire extinguishing system as described in the first aspect above.

[0051] In an embodiment of the present application, an acoustic emission sensor is used to collect ice crystal growth vibration signals caused by internal fluid phase change on the pipe wall of an automatic sprinkler fire extinguishing system, and the signals are synchronously correlated with real-time freezing rate data collected by a temperature sensor mounted on the pipe surface. The ice crystal growth vibration signals and the real-time freezing rate data are input into a multi-scale long-short-term memory network. The short-time span memory unit captures the microsecond-level pulse waveform caused by the instantaneous rupture of ice crystals, while the long-time span memory unit tracks the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipe material, thereby outputting a composite freezing feature. The transmission power and modulation rate of the LoRaWAN node are adaptively adjusted based on the spatial distribution of the pipeline network and the priority of the composite freezing feature. The high-priority composite freezing feature and the geographic location of the corresponding pipe section are encoded into an emergency data frame and preferentially transmitted to a remote control center via multi-hop routing. Based on a database of ice expansion coefficients of pipeline materials pre-stored in the remote control center, the ice crystal dynamic parameters in the composite freezing feature are matched with the material stress parameters to calculate the rupture risk probability of a specific pipe section at a critical ice thickness. The pipeline antifreeze performance is predicted based on the rupture risk probability, and the pipeline heating strategy corresponding to the rupture risk probability is activated based on the antifreeze performance prediction result.

[0052] The technical solution of this application has the following beneficial effects:

[0053] By collecting vibration signals of ice crystal growth on the pipeline wall through acoustic emission sensors and combining them with real-time freezing rate data from temperature sensors, multi-dimensional dynamic monitoring of the phase change process inside the pipeline is achieved, improving the comprehensiveness and real-time nature of freezing risk identification. Short-time span memory units are used to capture the microsecond pulse waveforms of instantaneous ice crystal rupture (such as crack propagation signals), while long-time span memory units are used to track low-frequency oscillations of fatigue deformation of pipeline materials (such as stress relaxation trends), thereby enhancing the accuracy of analyzing composite freezing characteristics. Based on the spatial distribution of the pipeline network and the priority of freezing characteristics, the transmission power and modulation rate of LoRaWAN nodes are dynamically optimized, and high-priority data and geographic location are encoded into emergency data frames, prioritizing the efficient transmission of critical information and reducing response delays caused by network congestion. Based on a pre-stored database of pipeline material ice expansion coefficients, the dynamic parameters of ice crystals are matched with material stress parameters, the rupture risk probability under critical ice thickness is accurately calculated, and the electric heating gradient heating strategy is activated to avoid chain rupture of the pipeline due to local overload.

[0054] Furthermore, by eliminating the vibration components of the ice crystal growth vibration signal that are not subject to phase change interference, the vibration components coaxial with the axial stress direction of the pipeline are retained, and further combined with the high-frequency sub-signals and low-frequency sub-signals separated by the time-frequency decomposition layer, the phase mutation points between adjacent ice crystal rupture events are tracked and the implicit correlation between the change in pipe cross-section ellipticity and the stress relaxation rate is established. Finally, a composite freezing feature containing the ice crystal distribution gradient and the material yield strength attenuation trend is generated through dynamic weight distribution. This process significantly improves the ability to accurately identify the dynamics of ice crystal growth inside the pipeline and its impact on the material, enhances the system's prediction accuracy of the anti-freeze performance of automatic sprinkler fire extinguishing system pipelines in extreme cold environments, and effectively reduces the risk of pipeline rupture due to freezing.

[0055] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0057] Figure 1 A flow chart showing a method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system provided by the present application is shown;

[0058] Figure 2 A schematic diagram of the structure of a pipe frost resistance prediction system for an automatic sprinkler fire extinguishing system provided by the present application is shown;

[0059] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0060] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0061] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0062] The development strategy for this solution is to first deploy a network of acoustic emission and temperature sensors. Through multimodal sensor fusion, this system captures the physical characteristics (vibration signals) and thermodynamic parameters (freezing rate) of ice crystal growth, forming a real-time dynamic monitoring layer. Secondly, a multi-scale LSTM network architecture is designed, utilizing short-term memory units to analyze microsecond ice crack pulses and long-term memory units to track low-frequency oscillations associated with material fatigue, enabling cross-temporal feature extraction of freezing damage. Next, a priority assessment model is established based on the pipeline topology. Communication parameters are dynamically adjusted via LoRaWAN nodes, prioritizing the transmission of geocoded data from high-risk pipeline sections to ensure low-latency reporting of critical information. Finally, combined with a database of material ice expansion coefficients, a gradient heating strategy is triggered through coupled analysis of ice crystal dynamic parameters and pipeline stresses to achieve precise antifreeze intervention. This strategy can be modeled after the zoning control logic of electric heating in pre-action systems. The entire process forms a closed-loop control chain of "perception-analysis-transmission-decision-making," enhancing intelligent prediction and targeted protection capabilities based on existing antifreeze measures.

[0063] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0064] Figure 1 The present invention provides a flow chart of a method for predicting the antifreeze property of pipes in an automatic sprinkler fire extinguishing system. Figure 1 As shown, the method includes:

[0065] 101. Use acoustic emission sensors to collect ice crystal growth vibration signals caused by internal fluid phase change on the pipe wall of the automatic sprinkler fire extinguishing system, and synchronously correlate them with real-time freezing rate data collected by temperature sensors mounted on the pipe surface;

[0066] In this step, the acoustic emission sensor is a device used to capture elastic wave signals released during the expansion or phase change of tiny cracks inside the material. It is particularly suitable for monitoring the ice crystal growth vibration caused by the internal fluid phase change of the pipe wall of the automatic sprinkler fire extinguishing system. The temperature sensor is mounted on the surface of the pipe to collect and record the temperature change data of the pipe surface in real time to calculate the real-time freezing rate. The ice crystal growth vibration signal is collected by the acoustic emission sensor, which is a vibration signal generated during the ice crystal growth process caused by the internal fluid phase change. The real-time freezing rate data is collected by the temperature sensor, which reflects the temperature change of the pipe surface and is used to calculate the freezing rate. The synchronously correlated data set is the ice crystal growth vibration signal and the real-time freezing rate data that have been synchronously correlated using high-precision signal processing technology (such as wavelet transform and adaptive filtering).

[0067] In an embodiment of the present application, acoustic emission sensors and temperature sensors are first deployed in the cold environment monitoring area. Specifically, a high-sensitivity acoustic emission sensor is directly installed on the wall of the automatic sprinkler fire extinguishing system pipe to capture the ice crystal growth vibration signal caused by the internal fluid phase change. These vibration signals contain the tiny elastic wave information released during the ice crystal formation process. At the same time, a high-precision temperature sensor is mounted on the surface of the pipe to collect and record the temperature change data on the pipe surface in real time to calculate the real-time freezing rate. In order to ensure the time synchronization of the data, the timestamp synchronization technology is used to associate the vibration signal with the temperature data. Then, an adaptive filtering algorithm is used to remove noise interference, and a pure vibration signal is extracted by independent component analysis (ICA). Furthermore, a wavelet transform is applied to perform multi-scale decomposition on the pure vibration signal to separate the features of different frequency bands. These processed data are converted into time series data through signal processing techniques (such as fast Fourier transform FFT and recursive graph analysis), and combined with the temperature data to finally output the ice crystal growth vibration signal and real-time freezing rate data.

[0068] Consider a scenario where a wet sprinkler system in a commercial building in a northern city experiences extreme cold, reaching -20°C in winter. Ice crystals form in the water flowing through the system pipes due to low-temperature phase transitions. Acoustic emission sensors (sensitivity 50kHz-1MHz, sampling rate 1MHz) are installed at key locations on the pipe's outer wall (such as elbows and valves). They capture the 20kHz-50kHz high-frequency elastic wave signals triggered by ice crystal growth, with a single pulse amplitude of 15mV and a duration of approximately 200μs. Simultaneously, a PT1000 platinum resistance temperature sensor (accuracy ±0.1°C) is attached to the pipe surface. The sensor data is aligned using a GPS synchronization module (time error <1μs). Wavelet threshold denoising (db4 wavelet basis, 5-level decomposition) is used to eliminate environmental vibration interference. Independent component analysis (ICA) is then used to separate the ice crystal growth vibration signal from the real-time ice formation rate data, improving the signal-to-noise ratio to 25dB.

[0069] 102. Input the ice crystal growth vibration signal and the real-time freezing rate data into a multi-scale long short-term memory network, capture the microsecond pulse waveform caused by the instantaneous rupture of ice crystals through the short time span memory unit, and simultaneously track the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipeline material through the long time span memory unit, and output the composite freezing characteristics;

[0070] In this step, a multi-scale long short-term memory network (LSTM)—a deep learning model capable of simultaneously processing information at different time scales—is used. Short-time span memory units capture microsecond pulse waveforms, while long-time span memory units track minute-long low-frequency oscillations. Composite freezing features, output by the LSTM network, combine the dynamic characteristics of ice crystals and the material's stress state for subsequent analysis and prediction.

[0071] In the embodiments of the present application, the obtained ice crystal growth vibration signal and real-time icing rate data are input into a multi-scale long short-term memory network (Multi-Scale LSTM). Specifically, a short-time span memory unit extracts microsecond-level pulse waveform features in a high-frequency sub-signal through a convolutional neural network (CNN), and these pulse waveforms reflect the dynamic characteristics of the instantaneous fracture of ice crystals. In order to enhance the feature extraction effect, local response normalization (LRN) and batch normalization (BN) techniques are used to standardize the feature maps. At the same time, a long-time span memory unit analyzes minute-level low-frequency oscillation features in a low-frequency sub-signal through a recurrent neural network (RNN) structure, and these oscillations reflect the situation of material fatigue deformation accumulation. In order to improve the robustness of the model, a gated recurrent unit (GRU) and an attention mechanism (Attention Mechanism) are introduced to better capture long-time span memory information. The two kinds of information are integrated through feature fusion techniques such as feature weighted average or feature splicing, and a composite freezing feature is output.

[0072] For example, continuing the above example, the pre-processed vibration signal (peak frequency 38 kHz, energy spectrum density 0.8 V 2 / Hz) and the icing rate time series data are input into a multi-scale LSTM network. The short-time memory unit (CNN layer, 3x1 convolution kernel) extracts microsecond-level pulse features: the pulse rising edge slope is 15 V / s, the decay time constant is 120 μs, and it corresponds to the ice crystal fracture transient process. The long-time memory unit (bidirectional GRU, time window 60 s) analyzes the 0.5 Hz low-frequency oscillation component, and its amplitude slowly increases from 5 mV to 12 mV, reflecting the fatigue accumulation of the HDPE pipeline material. In the feature fusion stage, the high-frequency features are given a weight of 0.7, and the low-frequency features are given a weight of 0.3, and a composite freezing index (CFI=0.82, threshold 0.75 triggers an early warning) is output.

[0073] 103. According to the spatial distribution of the pipeline network and the priority of the composite freezing feature, the transmission power and modulation rate of the LoRaWAN node are adaptively adjusted, the high-priority composite freezing feature and the corresponding pipe segment geographical location are encoded into an emergency data frame, and the emergency data frame is preferentially transmitted to the remote control center through multi-hop routing;

[0074] In this step, the LoRaWAN node is a wireless communication node based on LoRa technology, which is used for efficient and low-power data transmission in cold environments. The emergency data frame contains data packets of high-priority composite freezing features and corresponding pipe segment geographical location information, and is preferentially transmitted to the remote control center through multi-hop routing. Adaptive adjustment is a mechanism for dynamically adjusting the transmission power and modulation rate of the LoRaWAN node and optimizing data transmission efficiency according to the spatial distribution of the pipeline network and the priority of the composite freezing feature.

[0075] In an embodiment of the present application, a path loss model and a channel quality assessment technique are used to determine the optimal transmission power and modulation rate configuration based on the spatial distribution of the pipeline network and the priority of the composite frozen features. Specifically, an ant colony optimization algorithm (ACO) is used to optimize route selection and determine the most efficient transmission path. In order to encode the high-priority composite frozen features and the geographic location of the corresponding pipe section into an emergency data frame, the geographic location information is first converted into a standard format, such as GeoJSON, using geographic information system (GIS) technology. Then, the data is compressed using Huffman coding or LZ77 compression algorithm to reduce the transmission bandwidth requirement. Next, the reliability and efficiency of data transmission are ensured through forward error correction (FEC) technology and adaptive modulation and coding (AMC) strategy. Finally, a multi-hop routing protocol is used to prioritize the transmission of emergency data frames to the remote control center.

[0076] For example, continuing with the previous example, after identifying a CFI threshold exceeding the B2 layer pipe threshold, communication optimization is initiated. First, spatial parameters are obtained: the pipe segment is located in the northwest corner of the building (GPS 39.9042°N, 116.4074°E), 28 meters away from the relay node (path loss model attenuation is 42dB). Next, communication configuration optimization is performed: the LoRaWAN node switches to SF12 modulation (0.3kbps air rate), transmit power is increased to 20dBm, and data packets are encapsulated in GeoJSON format (65% compression). Finally, routing optimization is performed: an ant colony algorithm is used to select the optimal three-hop path (B2 → 5F relay → rooftop gateway). Transmission latency is reduced from the default 800ms to 320ms, with a packet loss rate of <0.1%. Emergency data frames, containing fields such as the pipe segment ID, CFI value, and GPS coordinates, are preferentially transmitted to the control center over QoS Level 1 channels.

[0077] 104. Based on the ice expansion coefficient database of pipeline materials pre-stored in the remote control center, match the ice crystal dynamic parameters and material stress parameters in the composite freezing characteristics, calculate the rupture risk probability of a specific pipe section under the critical ice thickness, predict the pipeline antifreeze based on the rupture risk probability, and activate the pipeline heating strategy corresponding to the rupture risk probability according to the antifreeze prediction result.

[0078] In this step, the rupture risk probability is calculated for a specific pipe section at a critical ice thickness based on a pre-existing pipeline material database, matching the ice crystal dynamic parameters and material stress parameters in the composite freezing feature. The pipeline electric heating gradient heating strategy activates the corresponding electric heating system based on the calculated rupture risk probability to prevent pipeline rupture due to freezing.

[0079] In an embodiment of the present application, based on the ice expansion coefficient database of the pipeline material pre-stored in the remote control center, the ice crystal dynamic parameters and material stress parameters in the composite freezing feature are first screened and standardized using a feature selection algorithm (such as recursive feature elimination (RFE)). Then, in the process of matching the ice crystal dynamic parameters and material stress parameters in the composite freezing feature, the current feature is input into a pre-trained classifier, and the Bayesian inference technique is used to estimate the probability of rupture risk. According to the calculated probability of rupture risk, the corresponding electric heating gradient heating strategy is automatically activated, and the heating power is adjusted in real time through fuzzy logic control or PID control algorithm to gradually increase the pipeline temperature, effectively inhibit the growth of ice crystals, and avoid pipeline rupture.

[0080] For example, continuing the above example, the control center retrieves the HDPE pipe ice expansion coefficient database (expansion coefficient 1.5×10 -4 / °C, tensile strength 19 MPa). Combined with the ice crystal dynamic parameters of a growth rate of 0.8 mm / min (exceeding the safety threshold of 0.5 mm / min) and the material stress parameter of a hoop stress of 14.7 MPa (safety threshold 16 MPa), the Bayesian risk model calculated a rupture probability of 82.3% (with a 95% confidence interval). The gradient heating strategy was immediately activated: the first stage was to increase the heating rate of the electric heating cable to -5°C at a rate of 2°C / min (PID control parameters Kp = 8, Ki = 0.5); the second stage was to maintain a constant temperature of 5°C for 30 minutes (energy consumption monitoring showed a power reduction from 500W to 300W). Post-intervention monitoring showed that the ice layer thickness retreated from 12 mm to 3 mm, and the CFI index dropped to 0.41, completing the activation of the pipeline electric heating gradient heating strategy.

[0081] This solution integrates real-time data from acoustic emission sensors and temperature sensors, combined with a multi-scale LSTM network to dynamically analyze ice crystal growth vibration signals and material deformation. This enables microsecond-level pulse capture and minute-level stress accumulation monitoring during pipeline freezing. Adaptive LoRaWAN communication prioritizes data transmission for high-risk pipeline sections, and based on a database of material ice expansion coefficients, it matches rupture probabilities, ultimately triggering an electric heating gradient heating strategy. This complete anti-freeze protection system, from monitoring and analysis to intervention, effectively prevents pipeline freeze-cracking risks.

[0082] To address the issues of high vibration signal interference and difficulty in analyzing multi-physics field coupling in ice crystal growth monitoring, the research and development strategy of this solution revolves around multi-source interference suppression and cross-scale feature fusion: Axial stress constraint filtering is used to eliminate environmental noise interference and extract the intrinsic vibration of ice crystal growth; time-frequency decomposition is used to separate high-frequency rupture pulses from low-frequency material deformation signals and synchronize the freezing dynamic coefficients; short-term waveform tracking and long-term oscillation correlation models are combined to analyze the cross-scale correlation between ice crystal evolution and material damage; and finally, physical parameters and real-time data are fused through dynamic weight allocation to generate a composite feature of ice crystal distribution gradient and material performance attenuation, enabling accurate prediction of freezing damage risk. In some embodiments, the ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long-short-term memory network. The short-time span memory unit captures the microsecond pulse waveform caused by the instantaneous rupture of ice crystals, while the long-time span memory unit tracks the minute-level low-frequency oscillations accumulated by the fatigue deformation of the pipeline material, outputting a composite freezing feature, including:

[0083] 201. Before inputting the ice crystal growth vibration signal into the multi-scale long short-term memory network, perform a vibration component elimination process on the ice crystal growth vibration signal that is not subject to phase change interference, and retain the vibration component coaxial with the axial stress direction of the pipeline;

[0084] In step 201, the non-phase-change interference vibration component refers to the vibration signal caused by external mechanical vibration of the pipeline (such as pump and valve startup and shutdown, bracket resonance) and non-icing phase-change noise (such as bubble bursting and fluid turbulence). The axial stress coaxial vibration component is the vibration component aligned with the axial stress of the pipeline, reflecting the orthogonal stress effect of ice crystal growth on the pipe wall.

[0085] In this embodiment, an improved adaptive noise cancellation (ANC) technique is first used to eliminate non-phase-changing interference. A reference sensor mounted on the pipe support collects ambient vibration noise, and a least mean square (LMS) algorithm is used to generate an antiphase acoustic wave to cancel the interference component. Furthermore, combining pipeline structural parameters (such as bend radius and wall thickness gradient), wavelet packet decomposition is used to extract the vibration component coaxial with the pipeline's axial stress direction (main frequency range 20kHz-500kHz).

[0086] 202. Input the vibration component into the time-frequency decomposition layer to separate the high-frequency sub-signal representing the instantaneous breakage of ice crystals and the low-frequency sub-signal reflecting the plastic deformation of the material, and convert the real-time freezing rate data into freezing dynamic coefficients aligned with the timestamps of each sub-signal;

[0087] In step 202, the time-frequency decomposition layer is a multi-resolution analysis layer based on the local time-varying characteristics of the signal, used to separate the transient characteristics of ice crystal breakage from the slowly varying characteristics of material deformation. The ice dynamic coefficient is a dimensionless parameter that quantifies the correlation between the ice growth rate and the time-frequency energy of the vibration signal.

[0088] In this embodiment of the present application, the vibration components are input into the time-frequency decomposition layer, and empirical mode decomposition (EMD) is used to adaptively separate the high-frequency intrinsic mode functions (IMF1-IMF3) from the low-frequency residual component (Residue). The high-frequency sub-signal (IMF1) corresponds to the microsecond pulse of instantaneous ice crystal rupture, while the low-frequency sub-signal (Residue) reflects the minute-level oscillation of the material's plastic deformation. Simultaneously, the real-time freezing rate data is converted into freezing dynamic coefficients aligned with the timestamps of each sub-signal through cubic spline interpolation.

[0089] 203. Input the high-frequency sub-signal into a short-time span memory unit for microsecond waveform contour tracking to capture phase mutation points between adjacent ice crystal breakage events. Input the low-frequency sub-signal and the icing dynamic coefficient into a long-time span memory unit for oscillation modal correlation to establish an implicit correlation between the change in pipe cross-section ellipticity and the stress relaxation rate.

[0090] In step 203, microsecond waveform profile tracking is used to capture the temporal correlation between the rising edge slope, peak energy, and pulse interval of the ice crystal rupture pulse. Oscillation mode correlation is used to establish a nonlinear mapping relationship between the low-frequency oscillation of material deformation and the icing dynamic coefficient.

[0091] In this embodiment, the high-frequency sub-signal is fed into a short-time span memory unit, where a gated causal convolutional network (GCCN) is used to extract microsecond-level waveform features. Phase-sensitive detection (PSD) is then used to identify phase transitions between adjacent pulses (e.g., phase jumps exceeding π / 2). The low-frequency sub-signal and the icing dynamic coefficient are fed into a long-time span memory unit, where a bidirectional gated recurrent network (Bi-GRU) is used to correlate oscillation modes. Timoshenko beam theory is then used to establish an implicit equation for the change in pipe cross-section ellipticity and the stress relaxation rate.

[0092] 204. Dynamically assign weights to the phase mutation point and the implicit association, and generate a composite freezing feature including an ice crystal distribution gradient and a material yield strength attenuation trend by combining the Young's modulus parameter and thermal expansion coefficient parameter pre-stored in the pipeline.

[0093] In step 204, dynamic weight allocation is to adaptively allocate feature weights based on the correlation between phase mutation point density and stress relaxation rate. The composite frozen feature is a multi-dimensional feature vector that integrates the ice crystal distribution gradient (vector) and the material yield strength decay rate (scalar).

[0094] In this example, an entropy weighting method is used to calculate the weight coefficients of phase mutation points (high-frequency features) and stress relaxation rates (low-frequency features). This is combined with the pipeline's pre-stored Young's modulus (calibrated via nanoindentation testing) and thermal expansion coefficient (obtained via a thermomechanical analyzer (TMA)) to construct a coupled ice crystal growth and material degradation model. Tensor splicing is used to combine the ice crystal distribution gradient (in polar coordinates) with the material yield strength decay trend (in time series) into a composite frozen feature. The feature dimension is dynamically adjusted by the pipeline topology (e.g., the number of tees and elbows).

[0095] Here's a specific example:

[0096] The piezoelectric vibration sensor installed at the pipe elbow of a shopping mall’s wet automatic sprinkler fire extinguishing system monitors the original vibration signal with an amplitude of 0.8V (peak-to-peak value), and its spectrum shows multi-peak characteristics in the range of 15kHz-600kHz. The main frequency of the environmental mechanical vibration noise measured by the reference sensor at the bracket is 50Hz (amplitude 0.2V). The improved LMS algorithm is used to generate the anti-phase beamforming signal, suppressing the non-phase change interference component to below -45dB. Combined with the pipeline structural parameters (bending radius 3.5m, wall thickness gradient 0.2mm / m), the Daubechies 6 wavelet basis is selected for 5-layer decomposition, and the main frequency band of the axial coaxial vibration component is extracted to be 250kHz±5%, and its power spectrum density reaches 1.2×10^-3V 2 / Hz. After EMD processing of this component at the time-frequency decomposition layer, the high-frequency IMF1 subsignal appears in the time domain as a transient pulse train with a pulse width of 0.5μs and a peak voltage of 1.5V, with intervals between adjacent pulses randomly distributed between 10-150μs. The low-frequency residual component manifests as a low-frequency oscillation with an amplitude of 0.12V and a period of 2.8min. A synchronously connected icing rate sensor measures the current icing rate at 1.2mm / h. A timestamp-aligned icing dynamic coefficient α = 0.45±0.03 is generated using cubic spline interpolation. A short-time span memory unit performs gated convolution on the high-frequency pulses, detecting phase jumps of up to π / 3 radians between adjacent pulses, with a frequency of 8.5 per second. A long-time span memory unit uses a Bi-GRU network to correlate the low-frequency oscillation amplitude envelope with the α coefficient, deriving the pipe cross-sectional ellipticity change rate Δe = 0.017 / min. This parameter deviates within 5% from the value predicted by the implicit equation based on Timoshenko beam theory. Combining the pipeline material parameters (Young's modulus 210 GPa, thermal expansion coefficient 12.5 × 10^-6 / °C), the entropy weight method was used to assign a weight of 0.62 to high-frequency features and a weight of 0.38 to low-frequency features. Ultimately, a 128-dimensional composite frozen feature was generated, including the ice crystal radial distribution gradient vector (polar angle 45°±3°, modulus 1.8 mm / m) and the yield strength decay rate scalar (-2.3 MPa / h). This dimension was dynamically adjusted based on the three elbows and two tee nodes in the pipeline topology.

[0097] To meet the needs of pipeline ice crystal monitoring, steps 201-204 achieve directional noise reduction of axial stress vibration signals through adaptive noise cancellation and wavelet packet decomposition. Empirical mode decomposition and bidirectional gated recurrent network are combined to separate and correlate high-frequency rupture pulses and low-frequency deformation oscillations. The entropy weight method is used to dynamically integrate the ice crystal distribution gradient and the material yield strength attenuation trend to construct a highly interpretable composite freezing feature, realize microsecond-level rupture event detection and sub-hour-level deformation tracking, improve the axial signal fidelity, reduce the frost heave risk prediction error, and compress the end-to-end processing delay to meet the real-time monitoring requirements of extreme cold environments.

[0098] To improve the transmission reliability and energy efficiency of pipeline monitoring data in cold environments, based on the LoRaWAN network characteristics of pipeline monitoring scenarios, a coupling model of ice crystal gradient changes and node energy consumption is established by dynamically evaluating the relationship between pipeline rupture risk and signal attenuation. A link quality-driven adaptive power allocation strategy is adopted to achieve joint optimization of transmit power and modulation mode under residual energy constraints. The physical characteristics of the pipeline section are encoded into a topology-aware data frame structure, and the transmission reliability of ice crystal orientation information is enhanced through a checksum field. Finally, combining the attenuation characteristics of metal obstacles and a multipath propagation model, a dynamic relay selection algorithm based on path fading minimization is constructed to implement a multi-hop optimization mechanism for alternating transmission of ground-reflected waves and direct waves. In some embodiments, the transmit power and modulation rate of the LoRaWAN node are adaptively adjusted based on the spatial distribution of the pipeline network and the priority of the composite frozen feature. High-priority composite frozen features and the corresponding pipeline section's geographic location are encoded into emergency data frames, which are preferentially transmitted to the remote control center via multi-hop routing, including:

[0099] 301. Analyze the pipeline rupture risk probability and stress concentration factor in the composite freezing feature, and calculate the dynamic transmission path loss threshold of each LoRaWAN node by combining the distance parameters of adjacent nodes in the pipeline distribution topology and the historical signal interference parameters;

[0100] In step 301, the pipeline rupture risk probability quantifies the likelihood of a brittle fracture of a pipe section under ice expansion stress, with a value ranging from 0 to 1. The dynamic transmission path loss threshold is the critical value of wireless signal attenuation that is dynamically adjusted based on environmental interference and node spacing, and determines the stability of the communication link.

[0101] In the embodiment of the present application, the pipeline rupture risk probability (based on the Weibull survival analysis model) and the stress concentration factor (calibrated by finite element simulation) in the composite freezing feature are first analyzed. Combined with the Delaunay triangulation model of the pipeline distribution topology, the Euclidean distance between adjacent nodes and the historical signal interference parameters (such as the bit error rate fluctuation variance) are extracted. The improved Hata-COST231 hybrid propagation model is adopted, and the ice crystal dielectric constant (measured by time domain reflectometry) and the metal obstacle diffraction loss (UTD theoretical calculation) are introduced. The dynamic transmission path loss threshold of each LoRaWAN node is determined by Monte Carlo sampling to ensure that the signal reception strength within the 95% confidence interval is ≥-110dBm.

[0102] 302. Dynamically allocate transmit power levels and orthogonal modulation indices based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, and simultaneously constrain the maximum adjustable power range in conjunction with the node residual energy parameter.

[0103] In step 302, the orthogonal modulation index is a communication efficiency parameter that defines the combination of the LoRa signal spreading factor and bandwidth. The maximum adjustable power range is a safe transmit power range determined based on the node's remaining energy and the circuit's heat dissipation capacity.

[0104] In an embodiment of the present application, a coupling relationship model is established between the rate of change of the ice crystal distribution gradient (calculated by the optical flow method) and the path loss threshold. The power allocation strategy is designed using non-cooperative game theory: the transmit power levels are discretized into a Nash equilibrium solution set, and the orthogonal modulation index is solved by convex optimization to obtain the Pareto optimal solution. The node residual energy parameter (obtained by the Coulomb counting method) and the PCB thermal resistance parameter (JEDEC JESD51 standard) are simultaneously introduced to construct a thermal-electrical joint constraint condition and limit the maximum adjustable power range (e.g., 14-20dBm).

[0105] 303. Classify and encode the high-priority composite freezing features and the corresponding pipe segment geographic locations according to pipe diameter specifications, construct a frame header containing the pipe segment identifier and a check field for the ice crystal growth azimuth, and generate an emergency data frame with a topology-aware flag.

[0106] In step 303, the pipe segment identifier is a unique code that combines the pipe diameter, material, and installation date, encoded using a Base32 variant. The ice crystal growth azimuth check field records the polar angle information of the main growth direction of the ice crystals for spatial correlation verification.

[0107] In the embodiment of the present application, the high-priority composite freezing feature is divided into DN50-DN300 categories by the pipeline specification classifier (base SVM), and the improved Geohash algorithm is used to encode the geographical location of the pipe section into a 12-character string. The ice crystal growth azimuth is extracted by Hough transform to extract the main direction angle (0-360°), and an 8-bit check code is generated in combination with the cyclic redundancy check (CRC-16-CCITT). The frame header structure is designed as follows: 4-byte identifier (pipe diameter + material code) + 2-byte azimuth check field + 1-byte emergency level identifier, and the frame payload is encrypted in AES-256-CTR mode to generate an emergency data frame with a topology-aware tag.

[0108] 304. Based on the pre-stored metal obstacle attenuation map in the pipeline distribution topology, a relay node sequence with minimum path fading is selected, and multi-hop relay transmission of the emergency data frame is achieved by alternately using ground reflection wave and direct wave propagation modes.

[0109] In step 304, the metal obstacle attenuation map is pre-stored in the electromagnetic wave attenuation characteristic database of metal components such as flanges and valves in the pipeline system. Multi-hop relay transmission realizes the alternating propagation mode of ground reflected waves and direct waves through relay nodes.

[0110] In an embodiment of the present application, a metal obstacle attenuation map is constructed based on a ray tracing algorithm (such as the SBR / Image method) to quantify the Fresnel zone loss at different incident angles. An improved AODVjr routing protocol is used, combining the attenuation map with real-time RSSI (received signal strength indication) data to dynamically select the relay sequence with the minimum path attenuation. When the ground reflection wave mode is enabled, BPSK modulation and diversity reception technology are used to compensate for the multipath effect; the direct wave mode uses LoRa Chirp spread spectrum and MRC (maximum ratio combining) technology to improve the signal-to-noise ratio. By adaptively switching the propagation mode, reliable multi-hop relay transmission of emergency data frames is achieved.

[0111] Here's a specific example:

[0112] Suppose this solution is applied to monitoring the pipelines of a wet automatic sprinkler fire extinguishing system in a cold-region logistics cold storage. First, the composite freezing characteristics analyzed in step 301 yield a rupture risk probability of 0.93 (DN150 carbon steel pipe) and a stress concentration factor Kt of 2.3. Combining the 28m spacing between adjacent nodes and a historical bit error rate variance of 0.12, the dynamic path loss threshold calculated using the Hata-COST231 model is -107dBm. Secondly, in step 302, the ice crystal distribution gradient change rate of 1.8 mm / s triggers a high priority. Game theory is used to allocate a transmit power of 18 dBm (SF10 / BW125 kHz). The node's remaining energy of 3.2 Ah constrains the maximum power to 19 dBm. Thirdly, in step 303, the pipe segment identifier is encoded as "DN15-C-2023" and the azimuth angle is 225°. The check code 0x7B is generated and an emergency data frame (6-byte header + 128-byte payload) is constructed. Finally, based on step 304, a three-hop path (node ​​A → valve V7 → bracket S12 → control center) is selected according to the attenuation spectrum, alternating between ground reflected waves (bit error rate < 10 -6 ) and direct wave (RSSI = -102dBm) to complete multi-hop relay transmission.

[0113] To address the problem of data transmission reliability, steps 301-304 dynamically calculate the path loss threshold based on Weibull survival analysis and the Hata-COST231 hybrid model, use non-cooperative game theory to optimize the power-modulation parameter combination, design a topology-aware emergency data frame structure and integrate multimodal propagation control, and realize alternating transmission of ground reflected waves / direct waves through ray tracing and adaptive routing, so that the path loss prediction accuracy reaches ±1.5dB, the node energy consumption balance is optimized by 32%, the emergency data packet delivery success rate exceeds 99.99%, and the multipath interference suppression capability is improved by 28dB, effectively ensuring low-latency and high-reliability multi-hop relay transmission in complex metal pipeline environments.

[0114] To optimize the energy efficiency and anti-interference capabilities of cold-region pipeline monitoring systems, a thermodynamic coupling model is used to dynamically correlate the ice crystal evolution rate with node energy consumption. Phase-sensitive orthogonal modulation calibration compensation is triggered based on a path loss threshold, and a limited transient power pulse is activated in conjunction with a frost thickness threshold. Finally, a Doppler-shift-resistant adaptive communication parameter set is generated through time-slot interleaving coding, achieving cross-domain collaborative optimization of ice crystal dynamics, environmental attenuation, and wireless transmission characteristics. In some embodiments, the dynamic allocation of transmit power levels and orthogonal modulation indices based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, while simultaneously constraining the maximum adjustable power range in conjunction with the node residual energy parameter, includes:

[0115] 401. Establish a transmission power allocation model, convert the change rate of the ice crystal distribution gradient in the composite freezing feature and the node residual energy parameter through thermodynamic expansion margin to generate a power allocation reference value;

[0116] In step 401, the thermodynamic expansion margin is converted to quantify the energy loss caused by the volume expansion of ice crystals during phase transitions on the pipe wall, which is used as a proportional conversion factor for power allocation. The power allocation baseline value is a basic power configuration parameter that combines the thermodynamic effects of ice crystal growth and the energy status of the node.

[0117] In the embodiment of the present application, a coupling model is established between the rate of change of the ice crystal distribution gradient (based on the ice-tube contact stress gradient calculated by Hertz contact theory) and the node residual energy parameter (corrected by Coulomb counting and Peukert equation). Using the non-equilibrium thermodynamic method, the ratio of the latent heat of ice crystal phase change (334kJ / kg) to the elastic strain energy of the pipe (Hooke's law) is calculated as the thermodynamic expansion margin coefficient. The energy constraint optimization problem is solved by the Lagrangian multiplier method to generate a power distribution benchmark value, in which the distribution weight of the high gradient pipe section (>1.5mm / s) is increased by 30%-50%.

[0118] 402. Perform phase sensitivity calibration on the orthogonal modulation index according to the dynamic transmission path loss threshold. When the dynamic transmission path loss threshold exceeds the theoretical attenuation value of the pipeline insulation layer, trigger a compensation mechanism for shifting the orthogonal modulation index in the direction of anti-multipath interference, and output a time-varying compensation coefficient.

[0119] In step 402, phase sensitivity calibration dynamically adjusts the phase response characteristics of the modulation index to address signal phase distortion caused by multipath interference. The time-varying compensation coefficient is a dynamic correction factor for the quantized modulation parameter offset and has a nonlinear relationship with the path loss.

[0120] In an embodiment of the present application, a phase-locked loop (PLL) technique is used to calibrate the phase sensitivity of the orthogonal modulation index. When the dynamic transmission path loss threshold exceeds the theoretical attenuation value of the pipeline insulation layer (calibrated by COMSOL electromagnetic simulation), a compensation mechanism based on Lyapunov stability theory is triggered: a phase error dynamic equation is constructed, and the modulation index is driven to shift in the direction of anti-multipath interference through adaptive sliding mode control. The Kalman filter is iteratively updated in real time, outputting a time-varying compensation coefficient to suppress the phase jitter caused by channel time variation, and the phase tracking accuracy reaches ±0.1 radians.

[0121] 403. Inject the time-varying compensation coefficient into the transmit power allocation model. When the frost layer thickness exceeds the critical value of the sealing gap at the metal flange connection, trigger a transient overload transmit power pulse based on the power allocation reference value, and constrain the transient overload transmit power pulse to be within the maximum adjustable power range allowed by the node residual energy parameter.

[0122] In step 403, the transient overload transmission power pulse is a short-term high-power transmission mode triggered when the frost layer reaches a critical thickness, and is used to break through channel congestion.

[0123] The maximum adjustable power range is a safe power limit based on the node thermal design power (TDP) and the battery discharge curve.

[0124] In the embodiment of the present application, the time-varying compensation coefficient is injected into the power distribution model, and the pulse width modulation (PWM) technology is used to generate a transient overload transmission power pulse. When the frost layer thickness (measured by a microwave resonant sensor) exceeds the critical value of the flange sealing gap (ASTM F37 standard), the pulse trigger logic based on Bang-Bang control is activated. The pulse amplitude is dynamically adjusted according to the power distribution reference value and the node residual energy parameter (estimated by the RC circuit equivalent model), and is limited to the maximum adjustable power range allowed by the node residual energy parameter calculated by the Arrhenius equation (PCB maximum temperature rise threshold (ΔT<15°C)), ensuring that the pulse energy density does not exceed 2.8J / cm 3 .

[0125] 404. Perform time slot interleaving encoding on the orthogonal modulation rope after phase sensitivity calibration and the power level of the transient overload transmission power pulse to generate an adaptive communication parameter configuration table that adapts to the Doppler frequency shift characteristics at the pipeline elbow.

[0126] In step 404, time slot interleaving interleaves different modulation parameters by time slot to combat inter-symbol interference caused by Doppler shift. The adaptive communication parameter configuration table is a dynamic lookup table that integrates power, modulation, and time slot parameters, supporting complex channel adaptation at bends.

[0127] In the embodiment of the present application, the fractional Fourier transform (FRFT) is used to analyze the Doppler frequency shift characteristics at the pipe elbow (typical value ±2.4kHz). The calibrated orthogonal modulation index and the transient overload power level are time-slot interleaved and encoded: the time slot allocation template is generated by the Gold sequence, and combined with the Turbo code error correction mechanism, an adaptive communication parameter configuration table is constructed to adapt to the Doppler frequency shift characteristics at the pipe elbow. The configuration table contains 128 groups of parameter combinations (16 power levels × 8 modulation levels × 4 time slot modes) and supports dynamic switching based on the channel coherence time (switching period <10ms).

[0128] Here's a specific example:

[0129] Assume that this solution is applied to the pipe network monitoring of a wet automatic sprinkler fire extinguishing system in a subway tunnel in a cold region. First, through step 401, the ice crystal distribution gradient is obtained to be 1.8mm / s (X70 steel grade pipe), the node residual energy is 3.2Ah, the thermodynamic expansion margin conversion coefficient is 0.67, and the generated power reference value is 18.3dBm. According to step 402, the path loss threshold of -105dBm exceeds the insulation layer attenuation value of -98dBm, triggering the sliding mode compensation mechanism, and the modulation index is changed from SF9 is offset to SF11, and a time-varying compensation coefficient of 0.82 is output. According to step 403, the frost thickness at the flange is 2.3 mm (critical value 2.0 mm), and a transient overload pulse (amplitude 21 dBm, pulse width 50 ms) is activated, subject to the node maximum power of 22 dBm. Finally, based on step 404, a 1.8 kHz Doppler shift is detected at the elbow, and the configuration table selects time slot mode 3 (interleaving depth 8) and a turbo code rate of 3 / 4 to achieve a bit error rate of <10. -7 .

[0130] Steps 401-404 achieve a 40% increase in ice crystal phase change energy utilization through thermodynamics-communication cross-domain modeling, sliding mode compensation control, transient pulse dynamic constraints and Doppler adaptive coding, with a phase tracking accuracy of ±0.1 radians, a 46% increase in transient pulse peak power compared to the steady state and controllable temperature rise, an extension of the Doppler frequency shift tolerance to ±3kHz, a reduction of 2 orders of magnitude in the bit error rate in complex curved environments, and a 37% optimization of the node's comprehensive energy efficiency ratio, meeting the needs of reliable all-weather monitoring of extremely cold pipelines.

[0131] In order to improve the pulse emission efficiency and spatial energy adaptability of the cold region pipeline monitoring system, the phase change latent heat parameters are combined with the axial stress distribution of the metal pipe by real-time monitoring of the frost layer thickness and ice crystal distribution gradient, and a time-varying compensation model is constructed; based on the creep characteristics of the sealing material and the attenuation law of the bolt preload force, energy constraints are established, and the pulse parameter set is optimized through azimuth weighting and phase modulation technology, ultimately achieving the adaptive generation of transient overload emission instructions. Its core lies in the integration of thermodynamic phase change, material mechanics and stress wave modulation technology to form a closed-loop feedback control system to ensure that the system stability is maintained through dynamic power compensation when the frost layer thickness exceeds the limit. In some embodiments, the time-varying compensation coefficient is injected into the transmission power allocation model. When the frost layer thickness exceeds the critical value of the sealing gap at the metal flange connection, a transient overload transmission power pulse based on the power allocation reference value is triggered, and the transient overload transmission power pulse is constrained to be within the maximum adjustable power range allowed by the residual energy parameter of the node, including:

[0132] 501. Input the time-varying compensation coefficient into the transmission power allocation model, combine the real-time growth rate of the frost layer thickness with the phase change latent heat parameter of the pipeline metal material, and generate a dynamic overload power margin parameter. The node corresponding to the pipe section where the frost layer thickness exceeds the critical value of the sealing gap automatically activates the margin parameter compensation channel.

[0133] In step 501, the dynamic overload power margin parameter quantifies the dynamic balance between frost phase change energy absorption and node power supply capacity during transient pulse transmission. The margin parameter compensation channel is an energy redistribution logic channel activated when frost layer thickness exceeds the limit, ensuring power supply to key nodes.

[0134] In the embodiment of the present application, the intrinsic orthogonal decomposition (POD) method is used to establish the spatiotemporal correlation matrix between the time-varying compensation coefficient and the frost layer growth rate (measured by laser diffraction). Combined with the latent heat parameters of the phase change of the pipeline metal material (calibrated by DSC differential scanning calorimetry), a thermal-electric coupling transfer function is constructed, and the characteristic frequency components (0.1-10Hz) are extracted by the Prony algorithm. When the frost layer thickness (monitored by microwave resonant sensor) exceeds the critical value of the sealing gap (ASTM F37 standard), the compensation channel based on the fractional integrator is activated to generate a dynamic overload power margin parameter (per unit value 0-1), and the compensation amount increases exponentially with the axial penetration rate of the ice crystal.

[0135] 502. Based on the rate of change of the ice crystal distribution gradient, perform azimuth-weighted processing on the dynamic overload power margin parameter through the margin parameter compensation channel, phase-modulate the waveform envelope of the transient overload transmit power pulse using the pipeline axial stress distribution pattern, and generate a pulse parameter set that matches the check field of the ice crystal growth azimuth.

[0136] In step 502, azimuth weighting is a polarization modulation technique that spatially weights the power margin based on the main growth direction of the ice crystals. The pulse parameter set includes a set of pulse waveform characteristics such as amplitude, pulse width, and phase, which match the spatial distribution of ice crystals.

[0137] In the embodiment of the present application, the principal components (PC1-PC3) of the rate of change of the ice crystal distribution gradient are extracted by tensor decomposition, and combined with the pipeline axial stress distribution pattern (based on Timoshenko beam theory finite element simulation), an azimuth weighting function (0°-360° Gaussian distribution) is constructed. IQ modulation technology is used to phase modulate the transient overload pulse waveform envelope: the carrier frequency is dynamically offset (±5% of the center frequency point) according to the ice crystal azimuth check field (CRC-8 check code), and a pulse parameter set is generated that matches the check field of the ice crystal growth azimuth angle, including 16 groups of pulse parameters, supporting Barker code spread spectrum and Chirp linear frequency modulation composite modulation.

[0138] 503. Based on the creep parameters of the sealing material at the metal flange connection and the bolt preload attenuation coefficient, establish a time domain energy distribution constraint condition so that the pulse parameter set forms a non-uniform energy density distribution within the maximum adjustable power range allowed by the node residual energy parameter;

[0139] In step 503, the creep parameter of the sealing material is used to quantify the stress relaxation characteristics of the flange sealing gasket at low temperatures, characterized by creep compliance (J(t)). The non-uniform energy density distribution is a pulse energy envelope that decays exponentially in the time domain, matching the material stress relaxation curve.

[0140] In the embodiment of the present application, the creep parameters of the sealing material are fitted based on the generalized Maxwell model (time constant τ = 120min), combined with the bolt preload attenuation coefficient (obtained by an ultrasonic bolt stress detector), and the time domain energy distribution constraint equation is established. A dynamic programming algorithm is used to optimize the energy distribution of the pulse parameter set: within the maximum adjustable power range (limited by the node supercapacitor discharge curve), the first peak energy is designed to account for 60%, and the subsequent pulses are attenuated according to the e^(-t / τ) law, forming a non-uniform energy density distribution, ensuring that the energy density of the pulse group does not exceed 3.2J / cm 3 (IEC 62368-1 safety standard) maximum adjustable power range.

[0141] 504. Time-interleave the non-uniform energy density distribution and the dynamic overload power margin parameter to generate a transient overload emission instruction set carrying a stress concentration factor correction parameter at a metal flange connection.

[0142] In step 504, time slot interleaving interleaves pulses of different energy densities according to preset time slices to enhance the ability to resist multipath interference. The stress concentration factor correction parameter is a dynamic correction coefficient that reflects the local stress amplification effect at the flange connection.

[0143] In the embodiment of the present application, a Gold sequence is used to generate a time slot allocation template (period 31 bits), and the non-uniform energy density distribution and the dynamic overload power margin parameter are two-dimensionally interleaved and encoded. Combined with the flange stress concentration factor (calibrated by photoelastic experiments), a 32-bit correction parameter field (IEEE754 floating point format) is embedded in the instruction set, and Turbo code (code rate 1 / 3) is used for forward error correction. The resulting transient overload transmission instruction set contains 256 groups of instructions and supports a dynamic selection strategy based on the curvature radius of the pipe elbow (R / D ≥ 1.5).

[0144] Here's a specific example:

[0145] Assume that this solution is applied to the monitoring node of a wet automatic sprinkler fire extinguishing system in a data center. First, through step 501, the frost layer thickness of 1.8mm (critical value 1.5mm) is obtained to trigger the compensation channel, the phase change latent heat parameter is 334kJ / kg, and the dynamic overload power margin parameter is generated as 0.78. Secondly, according to step 502, the ice crystal azimuth angle is 135° (check code 0x5E) and the axial stress peak is 85MPa. The pulse parameter set [amplitude 20dBm, pulse width 40ms, frequency modulation slope 1] is generated. 2kHz / ms]; again through step 503, obtain the flange sealing gasket creep compliance J (60min) = 0.032GPa-1, the bolt preload is attenuated to 85% of the initial value, and a non-uniform energy distribution is designed (the first peak is 18dBm, and the subsequent pulses are attenuated according to e^(-t / 80)); finally, based on step 504, according to the Gold sequence time slot template (chip offset 15) and Turbo code error correction, an instruction set with a stress concentration factor correction coefficient of 1.8 is generated, and the bit error rate is <10 -8 .

[0146] Steps 501-505 achieve a 55% increase in transient pulse emission efficiency, an azimuth matching accuracy of ±3°, and a 92% consistency between energy density distribution and material creep characteristics through phase change thermodynamic feature extraction, tensor decomposition azimuth weighting, Maxwell creep modeling, and Gold sequence time slot interleaving. This improves the reliability of instruction set transmission, reduces flange stress concentration correction errors, and reduces node comprehensive energy consumption, meeting the needs of high-reliability monitoring of polar pipelines.

[0147] In order to improve the spatial energy adaptability and material deformation coordination control capability of the cold region pipeline monitoring system, the axial / radial components of the ice crystal distribution gradient are decomposed and the weld residual stress parameters are integrated to generate anisotropic compensation coefficients; the heat flux density vector is corrected in combination with the thermal conductivity of the material to generate an azimuth weighted matrix aligned with the main direction of the ice crystals; the pulse envelope phase is modulated using the ellipticity deformation parameter, and an amplitude attenuation template is formed based on the bolt array distribution; finally, the azimuth weight, attenuation template and temperature gradient data are integrated through tensor fusion technology to generate a pulse parameter set that strictly matches the ice crystal azimuth check field, thereby achieving precise adaptation of the dynamic growth characteristics of ice crystals to the electromagnetic wave energy distribution. In some embodiments, based on the rate of change of the ice crystal distribution gradient, the dynamic overload power margin parameter is azimuthally weighted through the margin parameter compensation channel, and the waveform envelope of the transient overload transmission power pulse is phase-modulated using the pipeline axial stress distribution pattern to generate a pulse parameter set that matches the check field of the ice crystal growth azimuth, including:

[0148] 601. Decompose the change rate of the ice crystal distribution gradient into the pipeline axial ice crystal growth rate and radial penetration rate components, perform stress relaxation weight allocation on the two components in combination with the residual stress parameters at the pipeline weld, and generate an ice crystal azimuthal anisotropy compensation coefficient;

[0149] In step 601, the axial ice crystal growth rate is the rate of ice layer expansion along the length of the pipeline, reflecting the volume expansion effect of the fluid phase change. The radial penetration rate component is the ice crystal intrusion velocity perpendicular to the pipe wall, indicating the adhesion strength of the ice-pipe interface. The ice crystal azimuthal anisotropy compensation coefficient is a dynamic correction factor that quantifies the coupling effect between ice crystal growth direction and the pipeline residual stress field.

[0150] In the embodiment of the present application, principal component analysis (PCA) is used to decompose the gradient change rate of ice crystal distribution, and the first principal component (PC1) is extracted as the axial growth rate, and the second principal component (PC2) is extracted as the radial penetration rate. The residual stress distribution at the pipeline weld is measured by digital image correlation (DIC), and the stress relaxation weight ratio of the axial and radial components is calculated by combining the Zener-Wert-Avrami phase transition kinetic equation. The Hankel matrix singular value decomposition (SVD) is used to generate the ice crystal azimuthal anisotropy compensation coefficient, in which the weight of the high residual stress area (>200MPa) is increased by 20%-40%.

[0151] 602. Input the anisotropic thermal conductivity of the pipeline metal material through the margin parameter compensation channel, perform heat flux density vector correction on the dynamic overload power margin parameter in combination with the ice crystal azimuth anisotropy compensation coefficient, and generate an azimuth weighting matrix aligned with the main direction of ice crystal growth;

[0152] In step 602, the anisotropic thermal conductivity is the difference in thermal conductivity of a metal material along different crystal directions and is calibrated by electron backscatter diffraction (EBSD). The azimuth weighting matrix is ​​a polarization distribution matrix that spatially modulates the power margin according to the main growth direction of the ice crystal.

[0153] In the embodiment of the present application, a thermal-electric coupling finite element model (COMSOL Multiphysics) is constructed, and the anisotropic thermal conductivity (such as 16.2 W / m·K in the axial direction and 14.5 W / m·K in the radial direction of 304 stainless steel) and the compensation coefficient matrix are input. The heat flux density vector field is solved by the discrete dipole approximation (DDA) algorithm, and the Gram-Schmidt orthogonalization method is used to correct the spatial distribution of the dynamic overload power margin parameter. Finally, a 32×32 azimuth weighted matrix aligned with the main direction of ice crystal growth is generated. The matrix elements are proportional to the square of the cosine of the main direction of ice crystal growth (detected by Hough transform), and the azimuth resolution reaches ±1.5°.

[0154] 603. Phase modulate the waveform envelope of the transient overload transmission power pulse based on the pipeline ellipticity deformation parameter, and generate a pulse amplitude attenuation gradient template using the circumferential distribution characteristics of the bolt array at the metal flange connection;

[0155] In step 603, the pipe ovality deformation parameter is a dimensionless parameter that quantifies the change in the roundness of the pipe cross section and is measured using a laser profilometer. The pulse amplitude attenuation gradient template is a reference pattern for designing the pulse energy attenuation law based on the circumferential distribution of flange bolts.

[0156] In the embodiments of the present application, a correlation model between ellipticity deformation and pulse phase modulation is established based on the Mindlin-Reissner plate theory. Digital holographic interferometry (DHI) technology is used to monitor the pipe cross-section deformation in real time, and the deformation parameters (such as ellipticity 1.05→1.12) are fitted using Zernike polynomials. The circumferential distribution characteristics of the flange bolt array (such as 8 bolts evenly distributed) are expanded using Fourier series to generate a pulse amplitude attenuation gradient template: the main lobe amplitude is distributed according to the cosine distribution of the bolt position, and the side lobe attenuation rate is inversely proportional to the preload attenuation coefficient (measured by the ultrasonic time difference method).

[0157] 604. Perform tensor fusion of the azimuth weighting matrix and the pulse amplitude attenuation gradient template, and combine the pipeline circumferential temperature gradient data recorded in the check field of the ice crystal growth azimuth to generate a pulse parameter set that matches the check field of the ice crystal growth azimuth.

[0158] In step 604, tensor fusion is to perform high-order coupling operations on the multi-dimensional spatial distribution matrix according to a given summation rule. The circumferential temperature gradient data is the thermal stress driving factor formed by the temperature difference in the circumferential direction of the pipeline.

[0159] In the embodiment of the present application, Tucker decomposition is used to fuse the azimuth weighting matrix (3rd order tensor) and the attenuation gradient template (2nd order tensor) into a 4th order composite tensor. Combined with the circumferential temperature gradient data in the ice crystal azimuth check field (measured by an infrared thermal imager, ΔT_max = 8°C), the tensor kernel function is optimized by the Levenberg-Marquardt algorithm. Finally, a pulse parameter set that matches the check field of the ice crystal growth azimuth is generated, which includes 128 groups of parameter combinations (16 levels of amplitude × 8 levels of pulse width × 4 types of frequency modulation modes), and supports an adaptive selection strategy based on the pipe curvature radius (R / D ≥ 2).

[0160] Here's a specific example:

[0161] Assume that this solution is adopted in the monitoring task of the wet automatic sprinkler fire extinguishing system network of a data center. First, through step 601, the ice crystal axial velocity is 1.5mm / s, the radial velocity is 0.08μm / s, the weld residual stress is 225MPa, and the anisotropy compensation coefficient is generated as 0.18. Through step 602, the thermal conductivity is obtained as 15.8W / m·K in the axial direction and 14.1W / m·K in the radial direction. After correction, the main lobe azimuth angle of the azimuth weighting matrix is ​​147°, and the half-power beamwidth is ±12°. ; Through step 603, the ellipticity deformation is obtained to be 1.08, the preload force of flange 8 bolts is attenuated to 78%, and a hexagonal attenuation template is generated (main lobe amplitude -1.2dB / bolt, side lobe attenuation slope -3.5dB / 10°); finally, based on step 604, the circumferential temperature gradient ΔT = 6°C, the Tucker decomposition kernel function rank (3, 2, 2, 1), and the pulse parameter set [amplitude 19dBm, pulse width 35ms, frequency modulation slope 15kHz / ms] are generated, and the bit error rate is <10 -9 .

[0162] Steps 601-604 achieve an ice crystal main growth direction tracking accuracy of ±1.8° through principal component decomposition of ice crystal anisotropy, thermo-electric coupling vector correction, ellipticity deformation phase modulation and high-order tensor fusion, improve the pulse energy spatial matching, reduce the ellipticity deformation compensation error, achieve circumferential temperature gradient adaptability of ΔT=10°C, increase the pulse parameter set channel capacity, and achieve flange bolt stress distribution control error of <4%, meeting the needs of multi-dimensional physical field collaborative monitoring of extremely cold pipelines.

[0163] To solve the problem of insufficient precision of ice-pipe multi-physical field coupling modeling in the risk assessment of pipe frost heave rupture in cold regions, the axial / radial components of ice crystal dynamic parameters are decomposed, and an ice layer thickness-pipe stress coupling increment atlas is generated by combining an ice expansion coefficient database; based on historical frost heave deformation data, the maximum principal stress direction is matched, and a dynamic stress concentration factor is generated by fusing material aging characteristics; the material anisotropic expansion coefficient and flange pre-stress parameters are introduced to establish a proportional model of critical ice thickness and stress concentration factor to calculate the rupture risk probability; when the risk value exceeds the material critical yield threshold, according to the geometric mapping relationship between the ice crystal penetration rate and the pipe segment azimuth angle, an axial progressive electric heat tracing warming strategy is dynamically triggered to realize the hierarchical and accurate inhibition of frost damage risk. In some embodiments, based on the pipe material ice expansion coefficient database pre-stored in the remote control center, the ice crystal dynamic parameters and material stress parameters in the composite freezing feature are matched to calculate the rupture risk probability of a specific pipe segment under the critical ice thickness, including:

[0164] 701. decomposing the ice crystal dynamic parameters in the composite freezing feature into ice crystal axial growth rate and ice crystal radial penetration rate, combining the pipe material low-temperature phase change characteristic parameters and yield strength attenuation gradient in the pipe material ice expansion coefficient database pre-stored in the remote control center, and generating an ice layer thickness-pipe stress coupling increment distribution atlas;

[0165] In step 701, the ice crystal axial growth rate is the ice layer extension length per unit time along the pipe axis, reflecting the fluid phase change volume expansion driving force. The ice layer thickness-pipe stress coupling increment distribution atlas is a space-time correlation matrix quantifying the ice layer thickening and pipe stress evolution, including a principal stress vector field and an equivalent strain cloud map.

[0166] In the embodiments of the present application, the crystal plastic finite element model (CPFEM) is used to decompose the ice crystal dynamic parameters, and the axial growth rate (based on the Norton power law equation) and the radial penetration rate (based on the Darcy-Stokes coupling equation) are calculated by the slip system activation theory. Combined with the pipe material low-temperature phase change characteristic parameters (calibrated by differential scanning calorimetry) and the yield strength attenuation gradient (obtained by in-situ nanoindentation testing) in the ice expansion coefficient database, a multi-scale damage evolution equation is constructed. The extended finite element method (XFEM) is used to simulate the ice-pipe interface crack propagation to generate an ice layer thickness-pipe stress coupling increment distribution atlas containing 10^4 grid nodes, with a spatial resolution of 0.1 mm / pixel.

[0167] 702. According to the pipe segment historical frost heave deformation record stored in the pipe material ice expansion coefficient database, the maximum principal stress direction in the ice layer thickness-pipe stress coupling increment distribution atlas is matched, and a dynamic ice expansion stress concentration factor is generated by combining the material stress parameters corresponding to the actual service life of the pipe;

[0168] In step 702, the historical frost heave deformation record is a time series database of deformation, temperature gradient, and stress concentration coefficient recorded in the pipeline's annual frost heave event. The dynamic ice heaving stress concentration factor is a three-dimensional stress amplification factor that combines real-time ice layer stress and historical damage cumulative effect.

[0169] In the embodiments of the present application, the dominant mode (the contribution rate of the first 5 modes is > 85%) is extracted from the historical frost heave deformation record by proper orthogonal decomposition (POD), and the maximum principal stress direction in the incremental distribution map is matched by Kriging interpolation (determined by Mohr circle analysis). Combined with the material fatigue parameters (Coffin-Manson equation fitting) corresponding to the service life of the pipeline, the Chaboche nonlinear follow-up hardening model is introduced to generate the dynamic ice heaving stress concentration factor, and the factor value of the weld area is increased by 30%-50%.

[0170] 703、Based on the proportional relationship between the critical ice thickness in the composite freezing feature and the dynamic ice heaving stress concentration factor, the anisotropic expansion coefficient of the pipeline material and the flange assembly prestress parameter are introduced to calculate the rupture risk probability of a specific pipe section under the proportional relationship. In step 703, the anisotropic expansion coefficient is the thermal expansion difference parameter of the pipe along the axial, radial, and circumferential directions, which is measured by laser speckle interferometry. The flange assembly prestress parameter is the contact stress distribution parameter of the flange sealing surface caused by the initial pretightening force of the bolt.

[0171] In the embodiments of the present application, a nonlinear proportional relationship model (R=P·Kt^m, m=2.3) between the critical ice thickness and the dynamic stress concentration factor is established. The anisotropic expansion coefficient (such as axial 17.5×10^-6 / ℃, radial 15.2×10^-6 / ℃) and the flange prestress (measured by ultrasonic bolt stress meter) are introduced by multi-axial fatigue criterion (Crossland equation), and the rupture risk probability of a specific pipe section under the proportional relationship is calculated by Monte Carlo simulation (10^5 times sampling). The probability value is mapped to the interval of 0-1 by Weibull distribution function, and the threshold sensitivity is ±0.5%.

[0172] The following is a specific example:

[0173] Assuming that this scheme is adopted in the wet automatic sprinkler system pipeline of a high-rise residential building in a cold area, first based on step 701, the ice crystal axial velocity is 2.1 mm / s, the radial velocity is 0.12 μm / s, the ice expansion coefficient database calls the X80 steel parameters (phase change expansion rate 0.018% / ℃, yield strength attenuation -0.4 MPa / ℃), and the incremental distribution map is generated to show that the maximum equivalent strain is 0.35%; through step 702, match the historical frost heaving deformation mode 3 (contribution rate 28%), calculate the dynamic stress concentration factor Kt=2.8 (the weld reaches 3.5); and then through step 703, the critical ice thickness is 7.2 mm, the anisotropic expansion coefficient is 18.1×10^-6 / ℃ in the axial direction, the flange prestress is 82 MPa, and the Monte Carlo simulation obtains a fracture risk probability of 94%.

[0174] In order to solve the problem of insufficient matching accuracy of the warming strategy and the risk probability after the pipeline frost heaving and cracking in the cold region, the application proposes a gradient temperature control method based on the spatial correlation of material brittle transition characteristics and stress field. In some embodiments, the pipeline frost resistance is predicted based on the fracture risk probability, and the pipeline warming strategy corresponding to the fracture risk probability is activated according to the frost resistance prediction result, including:

[0175] 801, based on the fracture risk probability, combining the low-temperature brittle transition temperature of the pipeline material and the ice crystal growth stress field distribution, dividing the pipeline segment frost resistance level, and generating a pipeline frost resistance prediction result containing a frost resistance level identifier;

[0176] In step 801, the low-temperature brittle transition temperature is the critical temperature at which the material changes from ductile fracture to brittle fracture, which is calibrated by Charpy impact test. The ice crystal growth stress field distribution is a three-dimensional stress tensor field generated by the ice layer expansion on the pipe wall, which includes the principal stress direction and the equivalent stress amplitude.

[0177] In the embodiments of the application, the crack driving force of different pipeline segments under the action of the ice crystal growth stress field is calculated by the J integral method in fracture mechanics, and a two-parameter criterion of ductility and brittleness is constructed by combining the low-temperature brittle transition temperature of the pipeline material (such as -45℃ for X70 steel). When the local stress exceeds the material ductility reserve, it is divided into high, medium and low three frost resistance levels according to the risk probability threshold (0.3 / 0.6 / 0.9), and a pipeline frost resistance prediction result containing a level identifier (H / M / L) is generated, and the identifier is embedded in the pipeline segment ID and stress concentration coordinates.

[0178] 802, according to the frost resistance level identifier, match the pre-stored pipeline warming strategy library, extract the warming rate and target temperature corresponding to the pipeline segment frost resistance level;

[0179] In step 802, the pipeline warming strategy library is a pre-stored set of temperature control parameters corresponding to different frost resistance levels, including warming rate and target temperature.

[0180] In this application example, a case-based reasoning (CBR) system was used to design a pipeline heating strategy library. Each case contains the environmental parameters (temperature gradient, ice thickness) of historical frost heave events and successful temperature control records. Using OLAP multidimensional indexing technology, matching cases are quickly retrieved by antifreeze grade identifier, extracting the heating rate (e.g., 5°C / min for H grade and 3°C / min for M grade) and the target temperature (20°C above the brittle transition temperature). The target temperature is corrected by introducing the material thermal hysteresis coefficient, and a time constraint is imposed on the heating parameter package.

[0181] 803. Based on the spatial correlation between the pipeline axial stress distribution pattern and the ice crystal penetration path, activate the electric heating areas in descending order of the antifreeze level of the pipeline section, and generate a segmented progressive heating instruction set;

[0182] In step 803, the pipeline axial stress distribution pattern characterizes stress propagation along the length of the pipeline, reflecting the energy accumulation trend along the ice crystal penetration path. The segmented progressive heating instruction set is a control instruction sequence that activates electric heating in stages based on risk level priority.

[0183] In the embodiment of the present application, a correlation model between the pipeline axial stress distribution pattern and the ice crystal penetration path is established based on the stress wave propagation theory, and the DBSCAN spatial clustering algorithm is used to identify high stress concentration areas. The pipe sections are sorted from high to low according to their antifreeze levels, and the pipe sections are divided into a core heating zone, a transition zone, and a monitoring zone. The generation of the segmented progressive heating instruction set adopts Petri net modeling technology to define the state migration rules of the temperature climbing stage (preheating, rapid rise, steady state), ensuring that there is a 10-15 minute time overlap buffer period in the heating process of adjacent sections to avoid secondary stress concentration caused by thermal shock.

[0184] 804. Execute the segmented progressive heating instruction set, combine the heating rate, the target temperature, and the real-time frost thickness data on the pipeline surface, dynamically adjust the electric heating power output, and generate a pipeline heating control signal that matches the rupture risk probability.

[0185] In step 804, the real-time frost thickness data on the pipeline surface is the frost thickness value continuously measured by the microwave resonant sensor. The pipeline temperature increase control signal is a closed-loop control signal that integrates power output instructions and risk feedback.

[0186] In the embodiment of the present application, a fuzzy PID controller is constructed to perform multi-objective optimization on the heating rate, target temperature parameters and real-time data of frost thickness (sampling period 500ms). When the frost thickness exceeds the set threshold (such as 3mm), the anti-integral saturation mechanism is triggered to dynamically increase the slope of the electric heating power output. The pipeline heating control signal is transmitted in Manchester encoding format, embedded with CRC-16 checksum and timestamp information, and supports RS485 bus and power carrier dual-mode communication to ensure the transmission integrity of the segmented progressive heating instruction set at a low temperature of -50°C.

[0187] Here's a specific example:

[0188] Assume that an ice crystal infiltration event occurs in a DN150 pipe section of a wet automatic sprinkler fire extinguishing system in a large storage center. According to step 801, the brittle transition temperature of X80 steel is measured to be -40°C, the maximum principal stress of the ice crystal stress field is 185 MPa, and the J integral value is 280 kJ / m 2 , the antifreeze level is determined to be H, and a prediction result is generated [pipe segment ID: GP-203, level: H, coordinates: (x = 235m, y = 0)]. Step 802 matches the H-level case from the strategy library, extracts a heating rate of 5°C / min and a target temperature of 20°C, and superimposes a thermal hysteresis coefficient of 1.2 to correct it to 24°C. Step 803 identifies the axial stress core area (235-240m), generates a segmented progressive heating instruction set [preheating area 230-235m (2°C / min), rapid heating area 235-245m (5°C / min), monitoring area 245-250m (1°C / min)], and sets a 12-minute buffer overlap period. Step 804 obtains a real-time frost layer thickness of 2.8mm, a fuzzy PID output power slope of 0.8kW / min, and transmits the Manchester coded signal to the heating cable via RS485. Within 30 minutes, the pipe segment temperature rises from -38°C to 20°C.

[0189] Figure 2 The present invention provides a schematic diagram of a pipe frost resistance prediction system for an automatic sprinkler fire extinguishing system. Figure 2 As shown, the device includes:

[0190] Capture module 21, used to collect ice crystal growth vibration signals caused by internal fluid phase change on the pipe wall of the automatic sprinkler system through acoustic emission sensors, and synchronously correlate them with real-time freezing rate data collected by temperature sensors mounted on the pipe surface;

[0191] Output module 22 is configured to input the ice crystal growth vibration signal and the real-time freezing rate data into a multi-scale long short-term memory network, capture the microsecond pulse waveform caused by the instantaneous rupture of ice crystals through short-time span memory units, and simultaneously track the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of pipeline materials through long-time span memory units, and output a composite freezing feature;

[0192] The transmission module 23 is configured to adaptively adjust the transmission power and modulation rate of the LoRaWAN node according to the spatial distribution of the pipeline network and the priority of the composite frozen feature, encode the high-priority composite frozen feature and the corresponding pipeline segment geographic location into an emergency data frame, and preferentially transmit it to the remote control center via multi-hop routing;

[0193] The activation module 24 is used to match the ice crystal dynamic parameters and material stress parameters in the composite freezing characteristics based on the ice expansion coefficient database of the pipeline material pre-stored in the remote control center, calculate the rupture risk probability of a specific pipe section under the critical ice thickness, predict the pipeline antifreeze based on the rupture risk probability, and activate the pipeline heating strategy corresponding to the rupture risk probability according to the antifreeze prediction result.

[0194] Figure 2 The device for predicting the antifreeze property of pipes in an automatic sprinkler fire extinguishing system can be used to Figure 1 The implementation principles and technical effects of the method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system described in the illustrated embodiment will not be elaborated upon. The specific manner in which the various modules and units perform their operations in the aforementioned embodiment of the device for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system have been described in detail in the related embodiments and will not be further elaborated upon here.

[0195] In one possible design, Figure 2 The apparatus for predicting the antifreeze property of pipes in an automatic sprinkler fire extinguishing system of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0196] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0197] The processing component 32 is used for the above Figure 1 The embodiment provides a method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system.

[0198] The processing component 32 may include one or more processors to execute computer instructions to perform all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0199] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0200] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0201] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0202] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0203] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0204] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The illustrated embodiment provides a method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system.

[0205] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0206] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0207] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system, characterized in that: include: In the cold environment monitoring area, acoustic emission sensors are used to directly capture ice crystal growth vibration signals caused by internal fluid phase changes on the walls of sprinkler fire extinguishing pipes, and synchronously correlated with real-time freezing rate data collected by temperature sensors mounted on the pipe surface; The ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long short-term memory network. The short-time span memory unit captures the microsecond pulse waveform caused by the instantaneous rupture of ice crystals. The long-time span memory unit tracks the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipeline material, and outputs a composite freezing feature. Adaptively adjust the transmit power and modulation rate of the LoRaWAN node based on the spatial distribution of the pipeline network and the priority of the composite frozen feature, encode the high-priority composite frozen feature and the corresponding pipeline segment geographic location into an emergency data frame, and preferentially transmit it to the remote control center via multi-hop routing; Based on the ice expansion coefficient database of pipeline materials pre-stored in the remote control center, the ice crystal dynamic parameters and material stress parameters in the composite freezing characteristics are matched, the rupture risk probability of a specific pipe section at a critical ice thickness is calculated, and the pipeline electric heating gradient heating strategy corresponding to the rupture risk probability is activated; The ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long short-term memory network, and the microsecond pulse waveform caused by the instantaneous rupture of ice crystals is captured through the short-time span memory unit. At the same time, the minute-level low-frequency oscillation caused by the fatigue deformation accumulation of the pipeline material is tracked through the long-time span memory unit, and the composite freezing characteristics are output, including: Before the ice crystal growth vibration signal is input into the multi-scale long short-term memory network, the ice crystal growth vibration signal is subjected to vibration component elimination processing of non-phase change interference, and the vibration component coaxial with the axial stress direction of the pipeline is retained; Inputting the vibration component into the time-frequency decomposition layer to separate the high-frequency sub-signal representing the instantaneous breakage of ice crystals and the low-frequency sub-signal reflecting the plastic deformation of the material, and converting the real-time freezing rate data into freezing dynamic coefficients aligned with the timestamps of each sub-signal; The high-frequency sub-signal is input into a short-time span memory unit for microsecond waveform contour tracking to capture the phase mutation points between adjacent ice crystal breakage events. The low-frequency sub-signal and the icing dynamic coefficient are input into a long-time span memory unit for oscillation mode correlation to establish an implicit correlation between the change in pipe cross-section ellipticity and the stress relaxation rate. Dynamic weight allocation is performed on the phase mutation point and the implicit association, and combined with the pre-stored Young's modulus parameters and thermal expansion coefficient parameters of the pipeline to generate a composite freezing feature including the ice crystal distribution gradient and the material yield strength attenuation trend.

2. The method according to claim 1, characterized in that Based on the spatial distribution of the pipeline network and the priority of the composite frozen feature, the transmission power and modulation rate of the LoRaWAN node are adaptively adjusted, and the high-priority composite frozen feature and the corresponding pipeline segment geographic location are encoded into an emergency data frame, which is preferentially transmitted to the remote control center via multi-hop routing, including: The pipeline rupture risk probability and stress concentration factor in the composite freezing feature are analyzed, and the dynamic transmission path loss threshold of each LoRaWAN node is calculated by combining the distance parameters of adjacent nodes in the pipeline distribution topology and the historical signal interference parameters. Dynamically allocate transmit power levels and orthogonal modulation indexes based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, and synchronously constrain the maximum adjustable power range in conjunction with the node residual energy parameter; The high-priority composite freezing features and the corresponding pipe segment geographic locations are categorized and coded according to pipe diameter specifications. A frame header containing the pipe segment identifier and a checksum field for the ice crystal growth azimuth angle is constructed to generate an emergency data frame with a topology-aware tag. Based on the pre-stored metal obstacle attenuation map in the pipeline distribution topology, a relay node sequence with minimum path fading is selected, and multi-hop relay transmission of the emergency data frame is achieved by alternately using ground reflection wave and direct wave propagation modes.

3. The method according to claim 2, characterized in that The method of dynamically allocating transmit power levels and orthogonal modulation indexes based on the dynamic transmission path loss threshold and the rate of change of the ice crystal distribution gradient in the composite freezing feature, and synchronously constraining the maximum adjustable power range in conjunction with the node residual energy parameter, includes: Establishing a transmission power allocation model, converting the change rate of the ice crystal distribution gradient in the composite freezing feature and the node residual energy parameter through thermodynamic expansion margin to generate a power allocation reference value; performing phase sensitivity calibration on the orthogonal modulation index according to the dynamic transmission path loss threshold, and triggering a compensation mechanism for shifting the orthogonal modulation index in the direction of anti-multipath interference when the dynamic transmission path loss threshold exceeds the theoretical attenuation value of the pipeline insulation layer, and outputting a time-varying compensation coefficient; Injecting the time-varying compensation coefficient into the transmit power allocation model, triggering a transient overload transmit power pulse based on the power allocation reference value when the frost layer thickness exceeds a critical value of the sealing gap at the metal flange connection, and constraining the transient overload transmit power pulse to be within a maximum adjustable power range allowed by the node's residual energy parameter; The orthogonal modulation rope after phase sensitivity calibration and the power level of the transient overload transmission power pulse are time-slot interleaved and coded to generate an adaptive communication parameter configuration table adapted to the Doppler frequency shift characteristics at the pipeline elbow.

4. The method according to claim 3, characterized in that The step of injecting the time-varying compensation coefficient into the transmit power allocation model, triggering a transient overload transmit power pulse based on the power allocation reference value when the frost layer thickness exceeds a critical value of the sealing gap at the metal flange connection, and constraining the transient overload transmit power pulse to be within a maximum adjustable power range allowed by the node residual energy parameter, includes: The time-varying compensation coefficient is input into the transmission power allocation model, and the dynamic overload power margin parameter is generated by combining the real-time growth rate of the frost layer thickness and the phase change latent heat parameter of the pipeline metal material. The node corresponding to the pipe section where the frost layer thickness exceeds the critical value of the sealing gap automatically activates the margin parameter compensation channel; Based on the rate of change of the ice crystal distribution gradient, the dynamic overload power margin parameter is subjected to azimuth weighting processing through the margin parameter compensation channel, and the waveform envelope of the transient overload transmission power pulse is phase modulated using the pipeline axial stress distribution mode to generate a pulse parameter set that matches the check field of the ice crystal growth azimuth angle; Based on the creep parameters of the sealing material at the metal flange connection and the bolt preload attenuation coefficient, a time-domain energy distribution constraint condition is established so that the pulse parameter set forms a non-uniform energy density distribution within the maximum adjustable power range allowed by the node residual energy parameter; The non-uniform energy density distribution and the dynamic overload power margin parameter are time-interleaved to generate a transient overload emission instruction set carrying a stress concentration factor correction parameter at a metal flange connection.

5. The method according to claim 4, characterized in that The method includes performing azimuth weighted processing on the dynamic overload power margin parameter based on the rate of change of the ice crystal distribution gradient through the margin parameter compensation channel, phase modulating the waveform envelope of the transient overload transmission power pulse using the pipeline axial stress distribution mode, and generating a pulse parameter set that matches the check field of the ice crystal growth azimuth, including: Decomposing the change rate of the ice crystal distribution gradient into the pipeline axial ice crystal growth rate and radial penetration rate components, and performing stress relaxation weight distribution on the two components in combination with the residual stress parameters at the pipeline weld to generate the ice crystal azimuthal anisotropy compensation coefficient; The anisotropic thermal conductivity of the metal material of the pipeline is input through the margin parameter compensation channel, and the dynamic overload power margin parameter is corrected by heat flux vector in combination with the ice crystal azimuth anisotropy compensation coefficient to generate an azimuth weighted matrix aligned with the main direction of ice crystal growth; Phase modulation is performed on the waveform envelope of the transient overload transmission power pulse based on the pipeline ellipticity deformation parameter, and a pulse amplitude attenuation gradient template is generated using the circumferential distribution characteristics of the bolt array at the metal flange connection; The azimuth weighted matrix is ​​tensor-fused with the pulse amplitude attenuation gradient template, and combined with the pipeline circumferential temperature gradient data recorded in the check field of the ice crystal growth azimuth, a pulse parameter set matching the check field of the ice crystal growth azimuth is generated.

6. The method according to claim 1, characterized in that The method includes matching ice crystal dynamic parameters and material stress parameters in the composite freezing feature based on a database of pipeline material ice expansion coefficients pre-stored in the remote control center, calculating the rupture risk probability of a specific pipe section at a critical ice thickness, and activating a pipeline electric heating gradient heating strategy corresponding to the rupture risk probability, including: Decomposing the ice crystal dynamic parameters in the composite freezing feature into the ice crystal axial growth rate and the ice crystal radial penetration rate, and combining the pipe material low-temperature phase change characteristic parameters and the yield strength attenuation gradient in the pipeline material ice expansion coefficient database pre-stored in the remote control center, to generate an ice layer thickness-pipe material stress coupling incremental distribution map; Based on the historical frost heave deformation records of the pipe section stored in the pipeline material ice expansion coefficient database, the maximum principal stress direction in the ice layer thickness-pipe stress coupling incremental distribution map is matched, and combined with the material stress parameters corresponding to the actual service life of the pipeline, a dynamic ice expansion stress concentration factor is generated; Based on the proportional relationship between the critical ice thickness in the composite freezing feature and the dynamic ice expansion stress concentration factor, the anisotropic expansion coefficient of the pipeline material and the flange assembly prestress parameter are introduced to calculate the rupture risk probability of a specific pipe section under the proportional relationship; When the superposition effect value exceeds the critical yield stress threshold of the corresponding pipe in the ice expansion coefficient database of the pipeline material, the pipeline electric heating gradient heating strategy corresponding to the rupture risk probability is progressively activated in the direction of the pipeline axial stress based on the correspondence between the ice layer penetration rate component and the spatial azimuth angle of the pipe section.

7. A system for predicting the frost resistance of pipes in an automatic sprinkler fire extinguishing system, characterized in that: include: A capture module is used to directly capture ice crystal growth vibration signals caused by internal fluid phase changes on the walls of sprinkler pipes within the cold environment monitoring area using acoustic emission sensors, and synchronously correlates these signals with real-time freezing rate data collected by temperature sensors mounted on the pipe surfaces. An output module is configured to input the ice crystal growth vibration signal and the real-time freezing rate data into a multi-scale long short-term memory network, capture the microsecond pulse waveform caused by the instantaneous rupture of ice crystals through short-time span memory units, and simultaneously track the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of pipeline materials through long-time span memory units, and output a composite freezing feature; A transmission module is used to adaptively adjust the transmission power and modulation rate of the LoRaWAN node according to the spatial distribution of the pipeline network and the priority of the composite frozen feature, encode the high-priority composite frozen feature and the corresponding pipeline segment geographic location into an emergency data frame, and preferentially transmit it to the remote control center via multi-hop routing; An activation module is configured to match the ice crystal dynamic parameters and material stress parameters in the composite freezing feature based on a database of ice expansion coefficients of pipeline materials pre-stored in the remote control center, calculate the rupture risk probability of a specific pipe section at a critical ice thickness, and activate a pipeline electric heating gradient heating strategy corresponding to the rupture risk probability; The ice crystal growth vibration signal and the real-time freezing rate data are input into a multi-scale long short-term memory network, and the microsecond pulse waveform caused by the instantaneous rupture of ice crystals is captured through the short-time span memory unit. At the same time, the minute-level low-frequency oscillation caused by the accumulated fatigue deformation of the pipeline material is tracked through the long-time span memory unit, and the composite freezing characteristics are output, including: Before the ice crystal growth vibration signal is input into the multi-scale long short-term memory network, the ice crystal growth vibration signal is subjected to vibration component elimination processing of non-phase change interference, and the vibration component coaxial with the axial stress direction of the pipeline is retained; Inputting the vibration component into the time-frequency decomposition layer to separate the high-frequency sub-signal representing the instantaneous breakage of ice crystals and the low-frequency sub-signal reflecting the plastic deformation of the material, and converting the real-time freezing rate data into freezing dynamic coefficients aligned with the timestamps of each sub-signal; The high-frequency sub-signal is input into a short-time span memory unit for microsecond waveform contour tracking to capture the phase mutation points between adjacent ice crystal breakage events. The low-frequency sub-signal and the icing dynamic coefficient are input into a long-time span memory unit for oscillation mode correlation to establish an implicit correlation between the change in pipe cross-section ellipticity and the stress relaxation rate. Dynamic weight allocation is performed on the phase mutation point and the implicit association, and combined with the pre-stored Young's modulus parameters and thermal expansion coefficient parameters of the pipeline to generate a composite freezing feature including the ice crystal distribution gradient and the material yield strength attenuation trend.

8. A computing device, characterized in that The method comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method for predicting the frost resistance of pipes of an automatic sprinkler fire extinguishing system as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for predicting the frost resistance of pipes of an automatic sprinkler fire extinguishing system according to any one of claims 1 to 6 is implemented.

Images