Event-driven ultra-low power consumption sensing and awakening method and system for underground pipe network monitoring and storage medium

Abstract

The invention discloses an event-driven ultra-low power consumption sensing and awakening method and system for underground pipe network monitoring and a storage medium, and relates to the technical field of underground pipe network safety monitoring. The method comprises the following steps: acquiring a plurality of topological fingerprints representing multipath propagation characteristics of acoustic channels among monitoring nodes in a pipe network, and constructing a global echo topological graph based on the plurality of topological fingerprints; the initial monitoring node broadcasts a calibration probe signal and a data pulse sequence; the relay monitoring node identifies the identity of the initial monitoring node, and decodes the data pulse sequence to obtain a wake-up vector after the identity is confirmed; and based on the wake-up vector and the global echo topological graph, the relay monitoring node executes a local wake-up decision and determines whether to forward the updated wake-up vector to at least one next-hop forwarding monitoring node. The method solves the problems of high power consumption and short service life of a pipe network monitoring system in the prior art.

Description

Technical Field

[0001] This application relates to the field of underground pipeline safety monitoring technology, and in particular to an event-driven ultra-low power sensing and wake-up method, system and storage medium for underground pipeline monitoring. Background Technology

[0002] Underground pipe networks in urban water supply, gas, and petrochemical industries are critical infrastructure, and their safe and stable operation is of paramount importance. To promptly detect and locate pipeline leaks, abnormal pressure, blockages, and other faults, a large number of monitoring nodes are typically deployed on the network. However, the harsh environment of underground pipe networks makes it difficult to maintain or replace the batteries of these nodes after deployment. Therefore, the energy supply to these nodes becomes a core bottleneck restricting the long-term effective operation of the monitoring system.

[0003] Existing monitoring solutions mostly employ timed polling or periodic data reporting. This means that even when the network is in a normal state, all nodes need to periodically wake up their high-power sensing, processing, and communication modules, resulting in most of the energy being consumed in ineffective monitoring of the "normal state." To extend network lifespan, real-time monitoring must be sacrificed, for example, by setting the reporting cycle to several hours or even a day. However, this may lead to missing the optimal window for handling critical anomalies. Therefore, existing technologies face an irreconcilable contradiction: high real-time monitoring inevitably leads to high power consumption and short lifespan, while pursuing long lifespan requires sacrificing real-time performance.

[0004] Furthermore, the core technical challenge of using acoustic signals for communication lies in the dynamic changes in temperature, pressure, and flow rate of the medium within the pipeline network. These changes alter the speed of sound, rendering any authentication or communication mechanism reliant on the absolute propagation time of sound waves extremely unreliable and impractical in real-world industrial environments. A novel technical solution is urgently needed to overcome the impact of environmental changes on the stability of acoustic characteristics and to achieve a balance between ultra-long standby time and instantaneous response. Summary of the Invention

[0005] The purpose of this application is to provide an event-driven ultra-low power sensing and wake-up method, system and storage medium for underground pipeline network monitoring, which aims to solve the technical problems of excessive power consumption and short network life cycle caused by pipeline network monitoring nodes in order to ensure real-time performance in the prior art.

[0006] In a first aspect, this application provides an event-driven ultra-low-power sensing and wake-up method for underground pipeline network monitoring. The method includes: acquiring multiple normalized relative echo fingerprints characterizing the multipath propagation characteristics of acoustic channels between monitoring nodes in the pipeline network, and constructing a global echo topology map based on the multiple fingerprints; when an initial monitoring node senses a pipeline network anomaly event, generating a wake-up vector containing a wake-up command, encoding the wake-up vector into a data pulse sequence, and broadcasting a calibration probe signal and the data pulse sequence by the initial monitoring node; a relay monitoring node capturing a real-time echo sequence generated by the calibration probe signal, generating a real-time relative echo feature based on the real-time echo sequence, identifying the initial monitoring node by matching the real-time relative echo feature with multiple normalized relative echo fingerprints in the global echo topology map, and decoding the data pulse sequence to obtain the wake-up vector after confirming the identity; based on the wake-up vector and the global echo topology map, the relay monitoring node performs a local wake-up decision and determines whether to forward the updated wake-up vector to at least one next-hop forwarding monitoring node.

[0007] In one possible implementation of the first aspect, the normalized relative echo fingerprint is an ordered list containing multiple relative echo parameter tuples, and obtaining the normalized relative echo fingerprint includes: capturing an original echo sequence containing at least three echo pulses, the echo pulses including a first echo pulse, at least one intermediate echo pulse, and a final echo pulse; calculating the time interval between the at least one intermediate echo pulse and the first echo pulse, and normalizing it with the total time interval between the final echo pulse and the first echo pulse to obtain a time interval ratio; calculating the normalized amplitude of the at least one intermediate echo pulse relative to the first echo pulse to obtain an amplitude ratio; the relative echo parameter tuples include the time interval ratio and the amplitude ratio.

[0008] In one possible implementation of the first aspect, the wake-up vector is a data structure containing multiple fields, including: a source node ID field for identifying the initial monitoring node, an event type field for classifying the network abnormal events, a hop number field for recording the current transmission hop count, a lifetime field for limiting the maximum transmission hop count, a target range code field for defining the task execution range, and an execution instruction code field for specifying the local execution action.

[0009] In one possible implementation of the first aspect, encoding the wake-up vector into a data pulse sequence includes: serializing the wake-up vector into a bit stream; and modulating the bit stream into a data pulse sequence consisting of multiple sub-pulses using pulse position modulation or multi-frequency shift keying.

[0010] In one possible implementation of the first aspect, identifying the identity of the initial monitoring node by matching the real-time relative echo features with the plurality of normalized relative echo fingerprints includes: the relay monitoring node calculating a correlation coefficient value between the fingerprint and the real-time relative echo features for each normalized relative echo fingerprint in the global echo topology map with the relay monitoring node as the target monitoring node; and identifying the source monitoring node corresponding to the fingerprint whose correlation coefficient value exceeds a preset matching threshold as the identity of the initial monitoring node.

[0011] In one possible implementation of the first aspect, the relay monitoring node determines whether to forward to at least one next-hop forwarding monitoring node by: parsing the task target range in the wake-up vector to obtain a routing direction intent; based on the hop number segment in the wake-up vector, preferentially selecting neighbor monitoring nodes that have not received the wake-up vector as candidate forwarding nodes to avoid generating routing loops in the pipeline loop; and selecting from the candidate forwarding nodes a neighbor monitoring node whose acoustic path length satisfies the routing direction intent and whose link status is active, and determining it as the at least one next-hop forwarding monitoring node.

[0012] In one possible implementation of the first aspect, the method further includes: after forwarding the updated wake-up vector to the next-hop forwarding monitoring node, the relay monitoring node listens for the secondary echo signal generated by the next-hop forwarding monitoring node's re-forwarding within a preset time window; if the secondary echo signal is not heard within the preset time window, the link state corresponding to the next-hop forwarding monitoring node in the relay monitoring node's local adjacency table is updated to an inactive state.

[0013] In one possible implementation of the first aspect, the method further includes: when the relay monitoring node performs the step of determining at least one next-hop forwarding monitoring node again, excluding neighbor monitoring nodes whose link state is inactive from the candidate next-hop forwarding monitoring nodes.

[0014] Secondly, this application provides an event-driven ultra-low power sensing and wake-up system for underground pipeline network monitoring, comprising multiple monitoring nodes, wherein the multiple monitoring nodes include:

[0015] An initial monitoring node is configured to generate a wake-up vector containing a wake-up command when a pipeline abnormality event is detected, encode the wake-up vector into a data pulse sequence, and broadcast a calibration probe signal and the data pulse sequence.

[0016] A relay monitoring node is configured to capture a real-time echo sequence generated by the calibration probe signal and generate a real-time relative echo feature. The identity of the initial monitoring node is identified by matching the real-time relative echo feature with a pre-stored global echo topology map containing multiple normalized relative echo fingerprints. After confirming the identity, the data pulse sequence is decoded to obtain the wake-up vector. Based on the wake-up vector and the global echo topology map, a local wake-up decision is performed and it is determined whether to forward the signal to the next hop monitoring node.

[0017] Thirdly, this application provides a computer-readable storage medium on which a program stored is executed to implement the aforementioned method.

[0018] This method employs an event-driven, chain-recall mechanism, ensuring a strict positive correlation between energy consumption and event frequency. During 99.9% of the network's normal operation, the overall network power consumption is extremely low; however, in the event of an incident, the wake-up signal can propagate precisely along a specified direction within the network at the speed of sound. This method resolves the power consumption-real-time performance contradiction in existing technologies, achieving the beneficial effect of extending network lifetime by several orders of magnitude without sacrificing event response speed. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of an event-driven ultra-low power sensing and wake-up method for monitoring underground pipeline networks according to an embodiment of this application.

[0021] Figure 2 This is a schematic diagram of the pipeline self-calibration and topology fingerprint learning stage according to an embodiment of this application.

[0022] Figure 3 This is a schematic diagram of the propagation of wake-up signals between monitoring nodes during the task execution phase according to an embodiment of this application.

[0023] Figure 4 This is a functional block diagram of an intelligent acoustic monitoring node according to an embodiment of this application.

[0024] Figure 5 This is a schematic diagram comparing the environmental robustness of normalized relative echo fingerprints according to an embodiment of this application.

[0025] Figure 6 This is a schematic diagram illustrating a routing strategy for avoiding loops in a ring network according to an embodiment of this application.

[0026] Figure 7 This is a schematic diagram of the monitoring node hierarchical power consumption and wake-up process according to an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0029] This application provides a system and method specifically designed for underground pipeline network monitoring, comprising multiple acoustic monitoring nodes. The method uses the pipe itself, made of metal or composite materials, as the medium and waveguide for acoustic signal propagation. By utilizing the highly stable and unique multipath effect formed by the reflection, scattering, and diffraction of sound waves at pipe structures (such as elbows, T-joints, valves, and diameter changes), a unique acoustic fingerprint is established for each physical path between nodes. Based on this, this application constructs an event-driven, ultra-low-power chained wake-up mechanism.

[0030] In one specific implementation, this method uses a one-time self-calibration during deployment to allow each node to learn and store the acoustic fingerprint map of the entire pipeline network. During long-term operation, nodes run with low power consumption, triggering a wake-up signal carrying the fingerprint to propagate through the network only when a node detects an abnormal event. Other nodes can then perform physical authentication of the signal source with extremely low power consumption and decide whether to fully wake themselves up or relay the wake-up signal. This method solves the technical problems of high energy consumption and short system lifespan caused by periodic polling used in existing pipeline monitoring technologies to ensure real-time monitoring, while maintaining fast event response capabilities and extending the service life of battery-powered nodes.

[0031] Example 1

[0032] Reference Figure 1 This embodiment provides an event-driven, ultra-low-power sensing and wake-up method for underground pipeline network monitoring. The method is executed in a distributed network consisting of multiple intelligent acoustic monitoring nodes, which are physically fixed (e.g., via magnetic or welded bases) to the outer wall of a long underground oil or gas pipeline. The method mainly includes two stages: a pipeline self-calibration stage (corresponding to step S100) and a runtime event-driven wake-up stage (corresponding to step S200).

[0033] S100: During the network self-calibration phase, topological fingerprints between monitoring nodes are obtained, and a global echo topology map is constructed.

[0034] Reference Figure 2 The diagram illustrates a scenario where, during the self-calibration phase, a source node (monitoring node A) broadcasts a calibration probe, while other nodes (B, C, D) act as receivers to listen and extract fingerprints.

[0035] This stage is performed after the initial deployment of the monitoring network or after significant modifications to the pipeline structure (such as the addition of valves). Its goal is to enable each monitoring node to learn the propagation patterns of sound waves within the current pipeline physical structure, providing global physical topology knowledge for subsequent event tracing and directional propagation of wake-up signals. This step can be further broken down into S110 to S140.

[0036] S110: Triggers the entire network to enter calibration mode.

[0037] In one specific implementation, deployment engineers use handheld configuration devices to broadcast a "Enter Calibration Mode" command to the entire network via near-field communication (NFC or Bluetooth), either individually or through a designated initial node. Upon receiving this command, each smart acoustic monitoring node in the network performs a series of initialization operations. These operations include clearing any existing old Global Echo Topology Map (GETM) and Local Adjacency Table (LAT) from its internal non-volatile memory. Simultaneously, the node's coprocessor enters a standby state to prepare for topology fingerprint learning. Understandably, this step ensures that all monitoring nodes begin the learning process in a uniform and clean state, avoiding interference from old data in the construction of the new topology.

[0038] S120: Monitoring nodes take turns broadcasting calibration probes.

[0039] To systematically map the acoustic topology of the entire pipeline network, monitoring nodes need to be used as signal sources one by one. A handheld device or initial node sequentially designates each node in the network as the current broadcast source node (SourceNode) according to a predetermined order (e.g., sorted by the node's unique ID from smallest to largest). The designated SourceNode injects a standardized calibration probe signal (Pprobe) into the pipe wall through its signal transmission module. This acoustic signal propagates at high speed along the metal pipe wall in the form of an elastic wave. To be detectable from complex background noise and multipath echoes and for high-precision time synchronization, the calibration probe signal is designed to have specific physical properties. In one embodiment, the autocorrelation function of the calibration probe signal has a very sharp peak, while having extremely low values ​​at other time offsets.

[0040] For example, suppose a monitoring node, numbered Node0x01 to Node0x0A, is installed every 100 meters along a 1-kilometer straight pipe. The controller first designates Node0x01 as the SourceNode. Node0x01 then generates and broadcasts a calibration probe signal. This calibration probe signal can be a carrier signal with a center frequency of 50 kHz modulated by a 1023-bit M-sequence (Maximal Length Sequence). M-sequences are widely used for channel sounding and synchronization due to their excellent pseudo-randomness and Dirac-like autocorrelation properties. The duration of this signal can be set to 20.46 milliseconds (1023 chips, each lasting 20 microseconds, dimensionless). The transmission power of this signal is set to a standard value to ensure reliable reception within the typical communication distance of the network. After Node0x01 completes its broadcast, after a preset silence period (e.g., 1 second to ensure all echoes have sufficiently attenuated), the controller designates the next node, Node0x02, as the new SourceNode and repeats the broadcast process. This process repeats until all N nodes in the network have completed a broadcast as Source_Nodes. This polling mechanism ensures that the unidirectional acoustic channel characteristics between any two nodes in the network can be measured.

[0041] S130: Full network monitoring and extraction of normalized relative echo fingerprints.

[0042] When a SourceNode (e.g., Node0x01) is broadcasting a calibration probe signal, all other monitoring nodes in the network (Node0x02 to Node_0x0A) act as receivers, activating their listening front-ends to continuously monitor the vibration signal from the pipe wall. Once the listening front-end detects that the signal energy exceeds a preset wake-up threshold, it immediately wakes up the node's coprocessor. The coprocessor takes over the subsequent signal processing tasks, specifically executing the following sub-steps to generate a topology fingerprint.

[0043] S131: Direct Wave Detection and Raw Echo Sequence Acquisition. The coprocessor feeds the raw acoustic signal stream acquired by the monitoring front end into a digital matched filter. The impulse response of this matched filter is designed to be the time-conjugate of the calibration probe signal. According to matched filtering theory, when the received signal is the direct wave of the calibration probe signal (i.e., the signal propagating in a straight line along the pipe wall), the filter output will produce a peak with the highest energy. The coprocessor uses a peak search algorithm to accurately locate the time point of this peak occurrence and records this time point as the local zero-time reference, i.e. .this The establishment of this mechanism enables high-precision time synchronization between the receiving node and the Source_Node broadcast events, which forms the basis for all subsequent echo timestamp calculations.

[0044] exist Once the timing is established, the coprocessor does not immediately go to sleep. Instead, it continuously samples the acoustic channel at a high sampling rate (e.g., 1 MHz in Hertz) for a preset time window (e.g., 500 milliseconds). The length of this time window needs to be determined based on the complexity of the pipeline network and the echo attenuation characteristics. Its purpose is to capture all significant echoes that, after being reflected and scattered by various discontinuous structures (such as flanges, welds, elbows, and valves) as the calibration probe signal propagates through the pipeline network, reach the receiving node along different paths. The coprocessor performs energy detection on the signal sequence acquired within these 500 milliseconds, identifying all subsequent pulses whose energy exceeds a specific threshold and is temporally separated from the direct wave. For each identified echo pulse, the coprocessor calculates two core parameters: the first is the arrival time of the peak energy of the echo pulse relative to... absolute timestamp The unit is microseconds (μs); the second is the peak amplitude of the echo pulse relative to the direct wave (in Peak amplitude of the pulse detected at any time .

[0045] The coprocessor will extract all ( , ) tuple, according to The echoes are arranged in ascending order. This ordered list is [(tsabs1, ampabs1), (tsabs2, ampabs2), ..., (tsabsN,ampabsN)]. This original echo sequence forms the basis for subsequent calculations, but it is itself sensitive to environmental changes.

[0046] For example, when Node0x01 broadcasts the calibration probe signal, the coprocessor of Node_0x02 is awakened. It determines the arrival time of the direct wave through matched filtering. Subsequently, Within the next 500 milliseconds, it detected three significant echoes. Assume there is a flange 50 meters between Node0x01 and Node0x02, and a 90-degree bend 80 meters away. The first echo could be a reflection from the flange. It arrives in μs, and its amplitude is that of a direct wave. The second and third echoes may originate from complex scattering at the bend, respectively in μs and It arrives in μs, and its amplitude is that of a direct wave. and Therefore, Node0x02 generates a topological fingerprint for Node0x01: Fingerprint0102 = [(1520, 0.81), (3150, 0.65), (3280, 0.68)]. Meanwhile, Node0x03 (200 meters from Node0x01) receives a completely different echo sequence due to its different physical location, and may generate a fingerprint such as Fingerprint0103 = [(980, 0.92), (4500, 0.55)]. Each monitoring node in the network generates a unique fingerprint for the current SourceNode (i.e., Node0x01).

[0047] S132: Calculation and normalization of relative echo characteristics.

[0048] However, the velocity of sound in the fluid medium in underground pipe networks is extremely sensitive to environmental factors such as temperature and pressure, resulting in absolute timestamps in the original echo sequence. Unstable and impractical as a reliable fingerprint. To address this issue, this application processes the original echo sequence to extract relative features unaffected by changes in sound speed. The coprocessor performs the following operations:

[0049] Selecting the reference echo: From the original echo sequence, select the first significant echo ( () is used as the starting reference, and its absolute time is , amplitude Select the last clearly identifiable significant echo ( The absolute time is used as the termination reference. The total time span is defined as .

[0050] Calculate the relative parameter tuple: for any intermediate echo pulse in the sequence (in ), calculate its relative parameter tuple :

[0051] Time interval ratio ( This parameter characterizes the relative position of the intermediate echo within the entire echo sequence's time span. It is calculated as: This is a dimensionless value between 0 and 1. As the speed of sound changes, all... It will scale proportionally, but the scale value remains unchanged due to the synchronous scaling of the numerator and denominator.

[0052] Amplitude ratio ( This parameter characterizes the relative intensity of the intermediate echo to the initial echo. It is calculated as: This is a dimensionless value, mainly determined by the geometry and material of the reflector, and is not sensitive to changes in the speed of sound.

[0053] S133: Normalized relative echo fingerprint generation.

[0054] The coprocessor will process all the calculated relative parameter tuples Arrange them in order to form the final, environment-robust normalized relative echo fingerprint. For example: Fingerprint = [(ratiots2, ratioamp2), (ratiots3, ratioamp3), ...].

[0055] For example, suppose in At that time, node B received the original echo sequence from node A as: [(1520μs,0.81), (3150μs, 0.65), (4500μs, 0.55)].

[0056] Starting reference:

[0057] Termination Criteria:

[0058] Total time span: μs

[0059] Calculate intermediate echo ( The relative parameters of )

[0060]

[0061]

[0062] The generated fingerprint is: [(0.547, 0.802)].

[0063] Now, assuming that the increase in pipeline temperature causes a change in the speed of sound, the new original echo sequence becomes: [(1532.16μs,0.81), (3175.2μs, 0.65), (4536μs, 0.55)].

[0064] New total time span: μs

[0065] Recalculate the relative parameters of the intermediate echoes:

[0066]

[0067] The amplitude ratio remains unchanged. .

[0068] The newly generated fingerprint is still [(0.547, 0.802)], which proves the robustness of the fingerprint to environmental changes.

[0069] Reference Figure 5 This figure visually compares the impact of environmental changes on acoustic characteristics. Figure 5 A shows that the absolute timestamps of the original echo sequence experienced significant time drift with temperature changes, while Figure 5 B clearly shows that the normalized relative echo fingerprint proposed in this application maintains high stability in this process, thus proving the robustness of the fingerprint to environmental changes.

[0070] S140: Fingerprint reporting and global topology graph construction.

[0071] After Node0x01's broadcast cycle ends, all other monitoring nodes in the network (Node0x02 to Node0x0A) have generated their respective topology fingerprints. Next, these nodes need to aggregate the generated fingerprint data. In one embodiment, each node uses its onboard backup low-power wide-area network module (e.g., a LoRa module) to report the data. Each node packages its generated fingerprint with the current SourceNode's ID (0x01) and its own ID into a data packet, and sends it asynchronously at a low data rate to a handheld device or a specific node designated as the aggregation node. This low-rate asynchronous reporting strategy is used to avoid channel conflicts during data reporting and further reduce power consumption.

[0072] After receiving fingerprint data about SourceNode0x01 from all nodes, the handheld device stores them. For example, it records: the fingerprint from Node0x01 to Node0x02 is Fingerprint0102, the fingerprint from Node0x01 to Node0x03 is Fingerprint0103, and so on.

[0073] After the handheld device polls all nodes in the network (from Node0x01 to Node0x0A) as SourceNodes, it collects the topological fingerprints of all possible unidirectional acoustic channels between any two nodes in the network. At this point, the handheld device can construct a complete Global Echo Topology Map (GETM). In a specific implementation, GETM is a nested hash table or dictionary structure, in the form GETM[SourceID][Destination_ID] -> Fingerprint.

[0074] For example, a completed GETM might contain the following entries:

[0075] GETM[0x01][0x02] -> [(1520, 0.81), (3150, 0.65), (3280, 0.68)]

[0076] GETM[0x01][0x03] -> [(980, 0.92), (4500, 0.55)]

[0077] GETM[0x02][0x01] -> [(1490, 0.83), (3100, 0.69), (3250, 0.71)] (Note that the channel is not necessarily reciprocal)

[0078] ...and so on, containing the fingerprints of all N*(N-1) directed links in the network.

[0079] The handheld device distributes this completed GETM broadcast to every monitoring node in the network. Each node receives this global topology map and stores it in its own non-volatile memory. At this point, the network self-calibration phase is complete. Each monitoring node now possesses a global map of the entire network's acoustic connectivity, laying the foundation for event-driven wake-up at runtime. The network then exits calibration mode and enters ultra-low-power sentinel listening mode.

[0080] S200: During the runtime event-driven wake-up phase, it performs perception and wake-up based on global echo topology map and acoustic fingerprint matching.

[0081] Reference Figure 3 The diagram illustrates how, during runtime, a wake-up signal sent by an initial node (node ​​A) is received by a relay node (node ​​B) and forwarded to downstream nodes (C and D) according to intelligent routing decisions, while the upstream node (E) ignores the signal.

[0082] This is the normal operating state the network enters after calibration. During this stage, any monitoring node may become the initiator of a wake-up chain if its local sensors detect an abnormal event in the network. The core of this stage lies in utilizing GETM learned in S100 to achieve physical layer authentication of the event source and efficient targeted wake-up. The following process will be explained in detail using monitoring node A (ID 0x01) as the initial initiating node and monitoring node B (ID 0x02) as its downstream relay node.

[0083] S210: The initial node (node ​​A) senses the event, generates and encodes the wake-up vector.

[0084] In a specific application scenario, suppose node A is a pressure monitoring node deployed on an oil pipeline. Its local pressure sensor (which itself employs an ultra-low power design, such as event interrupt-based operation) detects a reading that drops by 30% within 100 milliseconds. This is judged by its internal logic as a Level 2 pressure drop event, which is highly likely to indicate a pipeline leak. The interrupt signal of this event triggers node A's coprocessor.

[0085] The coprocessor then creates a wake-up vector ( The wake-up vector is a standardized data packet. This vector is the core carrier of wake-up and task information. In one embodiment, the wake-up vector is a C-style structure containing multiple predefined fields. The coprocessor populates this structure as follows:

[0086] Source_ID: Enter the ID of node A itself, i.e., 0x01. This field is used to trace the original initiator of an event in multi-hop forwarding.

[0087] EventType: Enter a predefined event classification code. For example, PressureDrop_L2 is encoded as 0x02.

[0088] Hop_Count: Initialized to 0, indicating that this is the first hop in the wake-up chain.

[0089] TTL (Time-To-Live): Sets a maximum number of hops to live, for example, 10. HopCount is incremented by 1 after each hop. When HopCount reaches TTL, the packet is dropped to prevent infinite looping within the network.

[0090] TargetScope: This is a routing intent field. It defines the target range for the wake-up signal. For example, DOWNSTREAM10HOPS is encoded as 0x100A, indicating that this wake-up signal needs to propagate "downstream" (i.e., in the direction of medium flow in the pipe) away from the source ID for up to 10 hops.

[0091] ActionCommand: Specifies the local action that the node receiving this vector needs to perform. For example, WAKEANDREPORT is encoded as 0x01, instructing the receiving node to wake up its main MCU and perform detailed data acquisition and reporting tasks. Another example might be WAKEANDCLOSEVALVE, encoded as 0x05.

[0092] After all fields are filled, the coprocessor serializes the entire wake-up vector structure into a byte stream. Next, it calculates a cyclic redundancy check (CRC-32) on this byte stream and appends the calculated CRC value to the Vector_CRC field of the vector to ensure data integrity during transmission.

[0093] Finally, the coprocessor needs to encode this complete byte stream, including the CRC, into an acoustic signal that can be transmitted at the physical layer. To achieve ultra-low power transmission, traditional continuous carrier modulation is not used; instead, a pulse-based modulation scheme is employed. In one embodiment, pulse position modulation (PPM) is used. For example, the byte stream is converted into a bit stream. A short acoustic pulse delayed by 50 microseconds relative to a reference time represents bit '0', and a short acoustic pulse delayed by 100 microseconds represents bit '1'. In this way, the entire wake-up vector is encoded into a sequence of short pulses, called the data pulse sequence (Pdata). Compared to a single pulse, Pdata carries richer information, but its total energy consumption remains extremely low.

[0094] S220: Initial node (node ​​A) broadcasts wake-up sequence.

[0095] The signal transmitting module of node A performs a broadcast operation, which consists of two parts.

[0096] First, Node A broadcasts the exact same standard calibration probe signal Pprobe defined in S120. This signal serves to identify itself; it does not carry data, but the multipath echoes it generates as it propagates through the pipe will act as Node A's physical layer identity.

[0097] Immediately after the Pprobe broadcast ends, at a small, fixed time interval (e.g., 1 millisecond in seconds), node A immediately broadcasts the Pdata pulse sequence encoded in S210.

[0098] Pprobe and Pdata together constitute a complete wake-up sequence.

[0099] S230: Relay node (Node B) wake-up, matching, decoding and verification.

[0100] Node B's monitoring front-end remains in an ultra-low power monitoring state. When the direct wave of the calibration probe signal from node A propagates along the pipe wall and reaches node B, its energy triggers the monitoring front-end, waking up node B's coprocessor. Node B's coprocessor immediately begins executing a series of operations:

[0101] S231: Topological fingerprint matching and source localization. Similar to the calibration phase, the coprocessor immediately begins capturing and recording the echo sequence arriving immediately after the direct wave, forming a real-time raw echo sequence. It then performs the exact same computational process as S131 and S132, generating a real-time relative echo feature. This process is real-time and may only last for tens of milliseconds. The key difference is that the coprocessor now accesses its internally stored Global Echo Topology Map (GETM) and performs correlation calculations between this real-time feature and all normalized relative echo fingerprints stored in the GETM that target this node. Because the comparison is based on environmentally invariant scale values, the matching process is no longer affected by factors such as temperature and pressure, thus achieving reliable source authentication in real industrial environments.

[0102] For each retrieved stored fingerprint, the coprocessor calculates its correlation with the real-time relative echo feature just captured. In one embodiment, this can be achieved by treating both the fingerprint and the feature as vectors in a multidimensional space and calculating the cosine similarity or Pearson correlation coefficient between them. The result of this calculation is a correlation coefficient value between -1 and 1, with the value closer to 1 indicating a stronger match.

[0103] For example, the coprocessor of node B calculates:

[0104] The correlation coefficient between the real-time relative echo sequence and the stored GETM[0x01][0x02] (i.e., the fingerprint from node A) is (dimensionless).

[0105] The correlation coefficient between the real-time relative echo sequence and the stored GETM[0x03][0x02] (fingerprint from node C) is (dimensionless).

[0106] The correlation coefficient between the real-time relative echo sequence and the stored GETM[0x04][0x02] (fingerprint from node D) is (dimensionless).

[0107] The coprocessor compares these correlation coefficient values ​​with a preset matching threshold (e.g., 0.90, dimensionless). Since only the correlation coefficients from GETM[0x01][0x02] exceed this threshold, the coprocessor can make a high-confidence judgment: the verification passes, and the source of the current wake-up sequence is confirmed as the upstream monitoring node A (ID 0x01). This process completes the authentication of the event source at the physical layer. Any node that cannot generate the correct pipeline echo cipher or any external interference source cannot impersonate node A, thus avoiding false wake-ups.

[0108] S232: Data Decoding and Verification. After confirming that the source is node A, since the calibration probe signal is immediately followed by a Pdata sequence encoded using a specific PPM scheme, the coprocessor immediately initiates the corresponding demodulation program to decode the captured P_data pulse sequence. By measuring the precise position of each sub-pulse relative to its time window reference, it restores it to a bitstream, then converts the bitstream back to a byte stream, and finally reconstructs the wake-up vector. The data structure is described. The data decoding process is a common technique used by those skilled in the art and will not be elaborated upon here.

[0109] S233: CRC Check. The coprocessor calculates a CRC-32 checksum for all bytes of the reconstructed wake-up vector (except for the VectorCRC field itself). Then, it compares the result with the value of the VectorCRC field inherent in the vector. If they match perfectly, it means no errors occurred during data transmission, and data integrity is guaranteed.

[0110] Only when source matching, data decoding, and CRC verification are all successfully completed can node B consider that it has received a legitimate and valid wake-up command.

[0111] S240: The relay node (Node B) executes local decisions and actions.

[0112] After the verification is successful, the coprocessor of node B begins to parse the wake-up vector. The content is processed, and local actions are executed according to preset logic. This decision-making process can be viewed as a simple decision tree:

[0113] First, the lifecycle is checked. The coprocessor reads Hop_Count (value 0) and TTL (value 10). Since 0 < 10, the vector has not expired and processing can continue.

[0114] Next, the local command is executed. The coprocessor reads the ActionCommand field, which has a value of 0x01, corresponding to WAKEAND_REPORT. Based on this instruction, the coprocessor wakes up the main MCU, which is in deep sleep, via an internal interrupt signal. After being woken up, the main MCU receives task parameters from the coprocessor (such as event type and source), and then executes a predefined, more complex task, such as starting a high-power precision vibration sensor to perform detailed leakage acoustic characteristic analysis and sending a detailed report to the monitoring center via its remote communication module (such as satellite or cellular network).

[0115] S250: The relay node (Node B) performs intelligent routing calculations.

[0116] After triggering the local action, node B also needs to decide whether and to whom to continue passing on the wake-up signal. This is the core of the targeted wake-up solution, implemented by the intelligent routing decision function on the coprocessor.

[0117] S251: Read routing intent. The coprocessor parses the vectorW.TargetScope field, which has a value of 0x100A, corresponding to DOWNSTREAM10HOPS. This indicates that the wake-up signal needs to propagate downstream along the flow of the pipeline medium.

[0118] S252: Query Topology and Determine Direction. The coprocessor queries the locally stored GETM to understand the network topology. To determine the downstream direction, it needs to know the positions of neighboring nodes relative to the initial source A. In one embodiment, this can be achieved by comparing the lengths of the acoustic paths, which can be approximated by the timestamp of the first echo (typically the direct wave) of the fingerprint in the GETM. Assume that node B's neighbors are C (0x03), D (0x04), and E (0x05). The coprocessor will look up the following information:

[0119] The path length from A to B (approximately the first timestamp in GEM[0x01][0x02]):

[0120] Path length from A to C:

[0121] Path length from A to D:

[0122] Path length from A to E:

[0123] If a neighbor node X satisfies Therefore, node X is considered a "downstream" node relative to node B. This aligns with an intuitive physical fact: on a linear pipe, the farther a node is from the event source, the longer it takes for the signal to propagate along the pipe wall.

[0124] In some embodiments, to avoid loops, the coprocessor can read the wake-up vector. It retrieves the SourceID and EventType (or a unique event ID) from the vector and checks a local "processed event cache". Simultaneously, it reads the Hop_Count field from the vector. It only considers neighboring nodes that have never processed this event and whose link status is active as the candidate set for this forwarding. This step leverages the natural growth of message hop count, a mechanism to prevent messages from repeatedly propagating in loops.

[0125] Reference Figure 6 The diagram illustrates that in a network containing a BCDB loop, when a wake-up signal is sent from A, propagates through B, C, and D, and arrives at B again, node B will refuse to receive it because it detects that the event has already been processed (or the hop count does not meet expectations), thus effectively avoiding the routing loop.

[0126] S253: Filtering Forwarding Targets. The coprocessor iterates through all its neighboring nodes and identifies all downstream nodes based on the criteria in S252. Simultaneously, it queries its local adjacency table (LAT), which records the link status with each neighbor. Only nodes considered downstream and whose Status field in the LAT is 'Active' are selected as the final forwarding targets.

[0127] For example, after calculation, the coprocessor finds:

[0128] And the state of C in LAT is 'Active'.

[0129] And the state of D in LAT is 'Active'.

[0130] (E may be a node between A and B, and is the "upstream" node).

[0131] Therefore, the result returned by the intelligent routing decision function is forwarding_targets = [C, D].

[0132] S260: Relay node (Node B) encodes and forwards wake-up signals.

[0133] Node B forwards the wake-up signal to downstream nodes C and D.

[0134] S261: Update wake-up vector. The coprocessor copies the received wake-up vector. This generates a new vector `vectorWnew`. It increments the value of `vectorWnew.HopCount` by 1 (making it 1), recalculates the CRC checksum, and updates the `VectorCRC` field. All other fields remain unchanged.

[0135] S262: Encode the new Pdata. The coprocessor encodes the updated vectorWnew into a new data pulse sequence Pdata_new.

[0136] S263: Forwarding. Node B's coprocessor control signal transmission module broadcasts a new wake-up sequence (calibration probe signal + Pdata_new). Since the signal propagates along the pipeline, it will naturally propagate in both upstream and downstream directions. However, when upstream nodes (such as A) perform fingerprint matching, they will find that the signal source is their downstream node B, which does not match the TargetScope (DOWNSTREAM) intent, or they simply cannot match any legitimate fingerprints originating from B, and therefore will not respond. Downstream nodes C and D, on the other hand, can correctly match and decode, thus achieving directional chained wake-up.

[0137] Through the loop from S210 to S260, the wake-up vector reliably propagates hop-by-hop in the pipeline along the path triggered by the event and determined by the task intent and the pipeline topology, until it reaches the boundary of its TTL or target range.

[0138] Example 2

[0139] Suppose that at some point, monitoring node C fails due to physical damage.

[0140] After the failure occurs, its upstream node B needs to forward a wake-up signal to its downstream nodes, and its calculated forwarding target list is still [C, D]. Node B broadcasts the wake-up sequence (S260).

[0141] In some embodiments, an implicit link health acknowledgment mechanism exists between nodes. For example, after forwarding, node B enters a brief listening window. It expects to hear the secondary echo generated by node C or D successfully receiving and forwarding the wake-up signal. Since C has failed, it cannot forward the signal again. If the network is designed for hop-by-hop acknowledgment, B may only hear the secondary echo originating from D within a preset timeout period (e.g., 2 seconds), without hearing the expected secondary echo signal originating from C at all.

[0142] This timeout event triggered B's coprocessor. It updated its Local Adjacency List (LAT), changing the Status field of the entry corresponding to node C from 'Active' to 'Inactive', and recording the current timestamp Last_Heard.

[0143] Subsequently, the wake-up vector of another (or the same) event. The route has reached node B. Node B then performs the intelligent route calculation in S250 again. During the S253 phase of filtering forwarding targets, when it evaluates neighbor C, although C still satisfies the downstream geometry conditions ( However, when the coprocessor queries the LAT, it finds that node C's state is 'Inactive'. Therefore, node C is excluded from the target list for this forwarding. At this point, the wake-up chain will only continue to propagate along the path B->D.

[0144] By employing a mechanism where each node performs real-time route recalculation based on local link state awareness (updating LAT) and global topology knowledge (querying GEM), the entire monitoring network can automatically bypass failed monitoring nodes and dynamically find new available paths to deliver wake-up signals without any intervention from a central controller. This distributed, real-time fault bypass capability significantly improves the robustness of the entire monitoring system and the success rate of task delivery.

[0145] Example 3

[0146] This application also provides an event-driven, ultra-low-power sensing and wake-up system for underground pipeline network monitoring, designed to implement the aforementioned method. The system consists of multiple physically separate but functionally coordinated intelligent acoustic monitoring nodes. These nodes are deployed at different locations within the underground pipeline network, collectively forming a distributed monitoring network.

[0147] In this system, any node has the ability to play different roles depending on the actual situation. When a node first detects an abnormal event in the pipeline network, it becomes the initial monitoring node; while other nodes in the network that receive its signal become relay monitoring nodes. Figure 4 The internal functional block diagram of a single monitoring node is shown, which is the physical entity unit that implements the above methods.

[0148] The intelligent acoustic monitoring node 400 includes:

[0149] A listening front-end 401, which can be an analog circuit, comprises a high-sensitivity piezoelectric accelerometer (for sensing tube wall vibration) and an amplifier and comparator circuit with ultra-low quiescent current. Its function is to continuously monitor tube wall vibration with nanowatt-level power consumption for the vast majority of the time. When the received acoustic signal energy exceeds a configurable hardware threshold, it generates a level-shifting signal as a wake-up signal output.

[0150] A timer 402 can be a hardware counter driven by a highly stable temperature-compensated crystal oscillator (TCXO). It provides a uniform reference for all time-related measurements.

[0151] A signal transmitting module 403 may include a digital signal generator, a D / A converter, and a power amplifier for driving a transmitting transducer (e.g., a piezoelectric ceramic plate) to convert electrical signals into mechanical vibrations injected into the tube wall. It can generate and broadcast complex acoustic waveforms such as calibration probe signals and Pdata according to instructions.

[0152] A coprocessor 404 can be a microcontroller with low power consumption in active mode but sufficient computing power (e.g., an ARM Cortex-M4 core) to perform tasks such as matched filtering, normalized relative echo fingerprint calculation and matching, CRC checksum, and routing algorithms. Most of the time, it is in a deep sleep state, only activated by the wake-up signal from the listening front-end 401. It is responsible for executing all the complex decision logic in the methods of this application without waking up the more power-consuming main processor.

[0153] A main microcontroller unit (MCU) 405, which can be a more powerful but power-hungry processor, is responsible for executing the node's core business logic, such as controlling a high-precision pressure / flow sensor (not shown) for data acquisition, running complex leak diagnosis algorithms, or transmitting data back via a remote communication module (not shown). It is typically in a powered-off or deep sleep state, only starting operation when explicitly awakened by the coprocessor 404 based on the decoded Action_Command.

[0154] A non-volatile memory 406, which may be a Flash or FRAM memory chip, is used to persistently store critical data structures, retaining them even after the node is powered off and restarted. The memory 406 stores the program instructions required for the coprocessor 404 to run, as well as the global echo topology graph (GETM) learned during the self-calibration phase and the local adjacency list (LAT) dynamically maintained during operation.

[0155] In the coordinated operation of the entire system: when a node acts as the initial monitoring node, its local sensor triggers the coprocessor 404, which generates and encodes a wake-up vector and broadcasts the wake-up sequence through the signal transmission module 403. When other nodes in the system act as relay monitoring nodes, their listening front-end 401 is triggered, waking up the coprocessor 404. The coprocessor 404 performs real-time generation of relative echo characteristics and matches them with GETM in the memory 406 to identify the signal source. Subsequently, it decodes the data and determines whether to forward it through the signal transmission module 403 based on intelligent routing decisions. This division of labor and coordination enables the entire system to achieve reliable sensing and chained wake-up of pipeline events with extremely low power consumption.

[0156] Reference Figure 7 This diagram schematically illustrates the hierarchical power consumption model and tiered wake-up process within a single monitoring node. Most of the time, only the nanowatt-level monitoring front-end 401, operating in "sentinel mode," is active. When an event triggers an acoustic signal, the monitoring front-end wakes up the low-power coprocessor 404, which is in decision-making mode, via a hardware interrupt. This coprocessor performs low-power tasks such as fingerprint matching and routing decisions. Only when explicitly indicated by the Action_Command in the wake-up vector will the coprocessor further wake up the highest-power milliwatt-level main MCU 405 and its peripherals, which are in execution mode. This hierarchical power management and task allocation architecture is the core mechanism for achieving ultra-low power consumption in this system.

[0157] In addition, this application embodiment also provides a computer-readable storage medium, such as a non-volatile memory 406, on which a computer program stored is executed by a coprocessor 404 and / or a main MCU 405, and can completely implement all the steps in the above method embodiments.

[0158] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An event-driven, ultra-low-power sensing and wake-up method for underground pipeline network monitoring, characterized in that, include: Multiple normalized relative echo fingerprints characterizing the multipath propagation characteristics of acoustic channels between monitoring nodes in the pipeline network are obtained, and a global echo topology map is constructed based on the multiple fingerprints. When an initial monitoring node detects a pipeline anomaly, it generates a wake-up vector containing a wake-up command, encodes the wake-up vector into a data pulse sequence, and broadcasts a calibration probe signal and the data pulse sequence by the initial monitoring node. A relay monitoring node captures a real-time echo sequence generated by the calibration probe signal and generates a real-time relative echo feature based on the real-time echo sequence. The identity of the initial monitoring node is identified by matching the real-time relative echo feature with multiple normalized relative echo fingerprints in the global echo topology map. After confirming the identity, the data pulse sequence is decoded to obtain the wake-up vector. Based on the wake-up vector and the global echo topology, the relay monitoring node performs a local wake-up decision and determines whether to forward the updated wake-up vector to at least one next-hop forwarding monitoring node.

2. The method according to claim 1, characterized in that, The normalized relative echo fingerprint is an ordered list containing multiple tuples of relative echo parameters. Obtaining the normalized relative echo fingerprint includes: Capture a raw echo sequence containing at least three echo pulses, the echo pulses including a first echo pulse, at least one intermediate echo pulse, and a final echo pulse; Calculate the time interval between the at least one intermediate echo pulse and the first echo pulse, and normalize it with the total time interval between the final echo pulse and the first echo pulse to obtain a time interval ratio; Calculate the normalized amplitude of the at least one intermediate echo pulse relative to the first echo pulse to obtain an amplitude ratio; The relative echo parameter tuple includes the time interval ratio and the amplitude ratio.

3. The method according to claim 1, characterized in that, The wake-up vector is a data structure containing multiple fields, including: a source node ID field for identifying the initial monitoring node, an event type field for classifying the network abnormal events, a hop number field for recording the current transmission hop count, a lifetime field for limiting the maximum transmission hop count, a target range code field for defining the task execution scope, and an execution instruction code field for specifying the local execution action.

4. The method according to claim 1, characterized in that, Encoding the wake-up vector into a data pulse sequence includes: The wake-up vector is serialized into a bit stream; The bit stream is modulated into a data pulse sequence consisting of multiple sub-pulses using pulse position modulation or multi-frequency shift keying.

5. The method according to claim 1, characterized in that, Identifying the identity of the initial monitoring node by matching the real-time relative echo features with the plurality of normalized relative echo fingerprints includes: For each normalized relative echo fingerprint in the global echo topology map with the relay monitoring node as the target monitoring node, the relay monitoring node calculates a correlation coefficient value between the fingerprint and the real-time relative echo feature. The source monitoring node corresponding to the fingerprint whose correlation coefficient value exceeds a preset matching threshold is identified as the identity of the initial monitoring node.

6. The method according to claim 3, characterized in that, The relay monitoring node determines whether to forward the message to at least one next-hop forwarding monitoring node, including: Analyze the task target range in the wake-up vector to obtain a route direction intention; Based on the hop number segment in the wake-up vector, neighbor monitoring nodes that have not received the wake-up vector are preferentially selected as candidate forwarding nodes to avoid routing loops in the pipeline network. From the candidate forwarding nodes, select a neighbor monitoring node whose acoustic path length satisfies the routing direction intent and whose link status is active, and determine it as the at least one next-hop forwarding monitoring node.

7. The method according to claim 6, characterized in that, The method further includes: After forwarding the updated wake-up vector to the next-hop forwarding monitoring node, the relay monitoring node listens for the secondary echo signal generated by the next-hop forwarding monitoring node's re-forwarding within a preset time window. If the secondary echo signal is not detected within the preset time window, the link status corresponding to the next-hop forwarding monitoring node in the local adjacency table of the relay monitoring node is updated to inactive.

8. The method according to claim 7, characterized in that, The method further includes: When the relay monitoring node performs the step of determining at least one next-hop forwarding monitoring node again, it excludes neighbor monitoring nodes whose link status is inactive from the candidate next-hop forwarding monitoring nodes.

9. An event-driven, ultra-low-power sensing and wake-up system for underground pipeline network monitoring, characterized in that, It includes multiple monitoring nodes, among which: An initial monitoring node is configured to generate a wake-up vector containing a wake-up command when a pipeline abnormality event is detected, encode the wake-up vector into a data pulse sequence, and broadcast a calibration probe signal and the data pulse sequence. A relay monitoring node is configured to capture a real-time echo sequence generated by the calibration probe signal and generate a real-time relative echo feature. The identity of the initial monitoring node is identified by matching the real-time relative echo feature with a pre-stored global echo topology map containing multiple normalized relative echo fingerprints. After confirming the identity, the data pulse sequence is decoded to obtain the wake-up vector. Based on the wake-up vector and the global echo topology map, a local wake-up decision is performed and it is determined whether to forward the signal to the next hop monitoring node.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-8.

Images