Cave ecological monitoring system and method based on multi-source sensor

Abstract

The invention relates to the technical field of environment monitoring, and discloses a cave ecological monitoring system and method based on a multi-source sensor, and the system comprises a plurality of static multi-mode sensing nodes disposed in a cave, and a mobile autonomous data aircraft carrier and calibration unit. The static nodes are responsible for periodically collecting and locally caching multi-source environment data, and the mobile unit autonomously cruises to the vicinity of each static node by adopting a hybrid navigation strategy, acquires cached data through near field communication, and selectively performs in-situ calibration and maintenance on the cached data. The system solves a series of problems that long-distance wireless data transmission is difficult in complex environments such as caves, a sensor is easy to pollute and drift, the manual maintenance cost is high, and a fixed sampling strategy cannot give consideration to energy consumption and emergency monitoring at the same time, and the real-time performance of the system is improved through autonomous cruise, near-field interaction, online analysis and adaptive task reconstruction of a mobile unit. And unattended, high-fidelity, low-disturbance and long-period monitoring of the cave ecological environment is realized.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring technology, specifically to a cave ecological monitoring system and method based on multi-source sensors. Background Technology

[0002] Caves, as unique underground ecosystems, are extremely sensitive to external changes in their internal hydrology, geology, climate, and biological communities, serving as natural laboratories for studying climate change and biodiversity. To effectively protect and conduct scientific research on cave ecosystems, it is necessary to deploy sensor networks to monitor key factors such as temperature, humidity, carbon dioxide concentration, air pressure, water parameters, and acoustic environment within the caves in a long-term, continuous, and high-resolution manner.

[0003] However, in enclosed environments such as caves with complex geological structures, winding spaces, and inaccessible GPS signals, the application of traditional wireless sensor networks faces severe challenges. The winding rock walls, high humidity, and potential water bodies inside caves significantly attenuate wireless radio frequency signals. This makes long-distance wireless communication for transmitting data from deep within caves to external base stations generally suffer from short data transmission distances, low communication bandwidth, and high packet loss rates, making it difficult to guarantee the integrity and timeliness of massive amounts of data under large-scale, long-term monitoring.

[0004] To overcome this problem, equipping each sensor node with a high-power, long-range communication module and a large-capacity battery capable of supporting its long-term operation would drastically increase the hardware cost and power consumption of a single node, making large-scale, high-density deployment in large caves requiring fine-grained monitoring impractical, thus creating an irreconcilable contradiction between system cost and monitoring range.

[0005] More importantly, the sensors operate in the high humidity and dusty environment of caves for extended periods. Their probes are inevitably affected by dust, water vapor condensation, or the adhesion of microbial films, leading to decreased measurement accuracy or even complete failure. Current technologies typically rely on manual entry into the depths of caves periodically to perform manual maintenance, data copying, and sensor calibration at each node. This approach is not only costly in terms of manpower and carries significant operational risks, but the frequent human activities themselves also cause significant interference with the lighting, air, and acoustic environment, resulting in irreversible disturbances to the fragile cave ecosystem. This fundamentally contradicts the original intention of ecological monitoring and protection.

[0006] Furthermore, once deployed, existing static monitoring nodes typically operate in a fixed mode, collecting data according to a pre-set, unchanging cycle. This static strategy cannot dynamically adjust its sensing behavior based on the actual state of the environment. When the cave environment remains stable for extended periods, the pre-set high-frequency sampling continues, resulting in a large amount of redundant data and a waste of valuable energy, shortening the node's survival time in the field. Conversely, when critical, short-lived, sudden ecological events occur, the pre-set, excessively low sampling frequency is highly likely to miss capturing the details of these events, leading to the permanent loss of valuable scientific data. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a cave ecological monitoring system and method based on multi-source sensors, which solves the problems of difficulty in long-distance data transmission, drift and contamination of measurement data from fixed sensors operating for a long time, and disturbance to the cave environment caused by monitoring activities in existing cave ecological monitoring schemes.

[0008] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of the present invention provides a cave ecological monitoring system based on multi-source sensors, the system comprising at least one static multimodal sensing node and a mobile autonomous data carrier and calibration unit.

[0009] The static multimodal sensing node is used to periodically collect multi-source environmental data and cache it locally.

[0010] The mobile autonomous data carrier and calibration unit is configured to work in conjunction with the static multimodal sensing node. The mobile autonomous data carrier and calibration unit includes:

[0011] The navigation module is used to drive the mobile autonomous data carrier and calibration unit to autonomously cruise to the position of the static multimodal sensing node;

[0012] The data acquisition module is used to acquire multi-source environmental data cached locally by the static multimodal sensing node;

[0013] The cruise analysis module is used to analyze the acquired multi-source environmental data to determine the environmental state of the location of the static multimodal sensing node.

[0014] The task reconstruction module is used to generate and issue task reconstruction instructions to the static multimodal perception node based on the environmental state, so as to adjust its subsequent perception behavior.

[0015] In one possible implementation, the cruise analysis module quantifies the environmental state to identify the degree of environmental anomalies.

[0016] The quantitative assessment is based on at least one of the following: the current measurement of the static multimodal sensing node and its historical statistical characteristics, the measurement of its neighboring nodes, and its rate of change over time.

[0017] Specifically, the cruise analysis module achieves the quantitative assessment by calculating an environmental anomaly index. For nodes... A certain sensor at any time readings Its environmental anomaly index Calculated using the following formula:

[0018] ;

[0019] in:

[0020] The normalized amplitude deviation is calculated as follows:

[0021] Get the current measurement value With the sensor at the node long-term historical average The absolute value of the difference is then divided by the sensor's position at the node. long-term historical standard deviation ;

[0022] The normalized spatial gradient is calculated as follows:

[0023] Get the current measurement value At the same time node The arithmetic mean of measurements from all geographically neighboring nodes The absolute value of the difference, then divide that absolute value by the number of people at the same time. node The standard deviation of the measurements of its geographical neighbors and all its geographically nearest nodes ;

[0024] The normalized rate of change over time is calculated as follows:

[0025] First, calculate the rate of change of the current measured value over time. This rate of change is obtained by acquiring the current measurement value. The measurement value at the previous sampling time The difference is then divided by the time interval between the two sampling times. get;

[0026] Then take that rate of change over time. The absolute value of the sensor at the node is then divided by the absolute value of the sensor at the node. The absolute value of the largest historical rate of change recorded above ;

[0027] For nodes A certain sensor at any time The measured value;

[0028] For nodes The measurement value of the same sensor at the previous sampling time;

[0029] This is the preset sensor sampling time interval;

[0030] , , These are preset weighting coefficients corresponding to the normalized amplitude deviation, normalized spatial gradient, and normalized time rate of change, respectively.

[0031] Based on the above implementation, the task reconstruction module is further configured to: calculate the environmental anomaly index With the preset high threshold and low-bit threshold A comparison is made. When the degree of anomaly in the environment is higher than a preset high threshold, a task reconstruction instruction is generated to shorten the sensor sampling period of the static multimodal sensing node; when the degree of anomaly in the environment is lower than a preset low threshold, a task reconstruction instruction is generated to extend the sensor sampling period of the static multimodal sensing node.

[0032] In one possible implementation, the mobile autonomous data carrier and calibration unit further includes a standard reference sensor and a mechanism configured to drive the reference sensor and the sensor of the static multimodal sensing node to perform parallel measurements to obtain in-situ calibration data.

[0033] Accordingly, the system may also include a back-end server, which is configured to use the in-situ calibration data to correct the multi-source environmental data collected by the static multimodal sensing node in order to compensate for the measurement drift of the sensor.

[0034] In one possible implementation, the mobile autonomous data carrier and calibration unit further includes a maintenance module configured to perform in-situ cleaning and maintenance of the sensor probes of the static multimodal sensing node.

[0035] In one possible implementation, the static multimodal sensing node includes sensors for acquiring physical parameters such as temperature, humidity, or air pressure.

[0036] Sensors used to collect chemical parameters such as carbon dioxide concentration;

[0037] Sensors used to collect data on volatile organic compounds in organisms;

[0038] At least one of the sensors used to acquire acoustic features.

[0039] Furthermore, the static multimodal sensing node also includes an edge computing unit configured to process the acquired raw acoustic signals to extract acoustic feature vectors of dripping frequency or biological calls as acoustic event features.

[0040] In one possible implementation, the navigation module is configured to employ a hybrid navigation mechanism that combines line-following navigation on a preset path with real-time localization and map-building navigation when approaching the static multimodal sensing node.

[0041] A second aspect of the present invention provides a cave ecological monitoring method based on multi-source sensors, the method comprising:

[0042] The static multimodal sensing node periodically collects multi-source environmental data and caches it locally;

[0043] The mobile autonomous data carrier and calibration unit perform the following steps:

[0044] Autonomous navigation to the location of the static multimodal sensing node;

[0045] Acquire the multi-source environmental data locally cached by the static multimodal sensing node;

[0046] The acquired multi-source environmental data is analyzed during the cruise to determine the environmental state at the location of the static multimodal sensing node.

[0047] Based on the environmental state, a task reconstruction instruction is generated for the static multimodal sensing node;

[0048] The task reconstruction instruction is issued to the static multimodal sensing node to adjust its subsequent sensing behavior.

[0049] This invention provides a cave ecological monitoring system and method based on multi-source sensors. It has the following beneficial effects:

[0050] 1. This invention uses the cruise analysis module of the mobile autonomous data carrier and calibration unit to perform real-time analysis of the acquired multi-source environmental data, and generates and issues task reconstruction instructions based on the analysis results. It dynamically adjusts the sensing behavior of static multimodal sensing nodes. When the cave environment changes significantly, it can automatically perform refined monitoring, while automatically reducing power consumption when the environmental state is stable. This achieves adaptive allocation of monitoring resources and improves the targeting of monitoring.

[0051] 2. This invention uses a mobile autonomous data carrier and calibration unit to carry standard reference sensors. During the cruise, it performs in-situ parallel measurements on the sensors of static multimodal sensing nodes to obtain in-situ calibration data. It can also optionally perform in-situ cleaning and maintenance. Through back-end algorithm correction, it can actively compensate for measurement drift and data contamination caused by the sensors working in the high humidity and high corrosion environment of caves for a long time, thereby improving the fidelity and reliability of long-term monitoring data.

[0052] 3. This invention adopts a mode of physical cruising and close-range data acquisition using a mobile autonomous data carrier and calibration unit, replacing the long-distance wireless transmission method with severe signal attenuation in rock formations. This solves the technical problem of data recovery in deep caves. At the same time, the autonomous cruising and multi-task execution capabilities significantly reduce the reliance on and frequency of manual entry into caves for data recovery or equipment maintenance. Attached Figure Description

[0053] Figure 1 This is a system architecture diagram of the present invention;

[0054] Figure 2 This is a structural block diagram of the static multimodal sensing node of the present invention;

[0055] Figure 3 This is a structural block diagram of the mobile autonomous data carrier and calibration unit of the present invention;

[0056] Figure 4 This is a flowchart of the method of the present invention.

[0057] The system comprises: 10. Static multimodal sensing node; 11. Core processing unit; 12. Multimodal sensor array; 13. Data storage unit; 14. Near-field communication unit; 15. Power supply unit; 20. Mobile autonomous data carrier and calibration unit; 22. Navigation module; 221. Tracking navigation unit; 222. Real-time positioning and map building navigation unit; 223. Navigation control unit; 23. Data acquisition module; 24. Cruise analysis module; and 25. Mission reconstruction module. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Example:

[0060] Please see the appendix Figure 1 - Appendix Figure 3 This invention provides a cave ecological monitoring system based on multi-source sensors, comprising:

[0061] At least one static multimodal sensing node 10 is used to periodically collect multi-source environmental data and cache it locally;

[0062] The static multimodal sensing node 10 is encapsulated in a dustproof and moisture-proof housing, which includes a core processing unit 11, a multimodal sensor array 12, a data storage unit 13, a near-field communication unit 14, and a power supply unit 15.

[0063] The core processing unit 11, such as a microcontroller with ultra-low power consumption, is electrically connected to other units in the system via an internal bus. This core processing unit 11 is responsible for executing the node's firmware program, including: controlling the power-on and power-off timing of each peripheral device by the power supply unit 15; scheduling the multimodal sensor array 12 to acquire data according to a preset sampling period or received task reconstruction instructions; performing edge computing processing on the raw data from specific sensors; and controlling read and write operations on the data storage unit 13.

[0064] The multimodal sensor array 12 consists of a group of sensors for monitoring environmental parameters in different dimensions. In one specific embodiment, the array includes:

[0065] A sensor array used to collect physical parameters includes a temperature sensor, a humidity sensor, and a barometric pressure sensor.

[0066] The sensor array used to acquire chemical parameters includes a nondispersive infrared sensor for measuring carbon dioxide concentration, and an electronic nose sensor composed of multiple metal oxide semiconductor sensors for acquiring bio-volatile organic compound gas fingerprints.

[0067] A sensor array used to acquire acoustic event characteristics, comprising one or more microelectromechanical system (MEMS) microphones.

[0068] For the majority of the time, the core processing unit 11 controls the static multimodal sensing node 10 to enter a deep sleep mode. At this time, except for the real-time clock that maintains the time count, most of the other circuits are in a power-off state. When the RTC reaches the preset sampling time point, the core processing unit 11 is woken up and powers on each sensor in the multimodal sensor array 12 in sequence according to the preset timing. After the sensors stabilize, data acquisition is performed.

[0069] To reduce data storage requirements when acquiring acoustic event features, the core processing unit 11 does not directly store the original waveform data after acquiring the original acoustic signal. Instead, it executes an edge computing process: first, it performs bandpass filtering on the original acoustic signal to remove ambient background noise;

[0070] Then, a short-time Fourier transform is performed on the filtered signal to obtain its time-frequency spectrum.

[0071] Finally, predefined acoustic feature vectors are extracted from the time-spectrum graph, and only these low-dimensional acoustic feature vectors are stored in the data storage unit 13 along with the timestamp.

[0072] Data storage unit 13, such as a high-capacity NAND Flash chip, is used to locally cache all acquired multi-source environmental data and extracted feature vectors in the form of data packets. Each data packet contains a unique node ID, a high-precision timestamp, and corresponding measurement data.

[0073] The near-field communication unit 14, such as a Bluetooth Low Energy module, is used to receive wake-up and connection requests from the mobile autonomous data carrier and calibration unit 20 when they approach each other. After establishing a connection, it transmits all data packets cached in the data storage unit 13 to the mobile autonomous data carrier and calibration unit 20 at high speed. Simultaneously, the near-field communication unit 14 is also responsible for receiving and parsing task reconfiguration instructions issued by the mobile autonomous data carrier and calibration unit 20, and transmitting the instruction content to the core processing unit 11 to update its operating parameters.

[0074] The power supply unit 15, for example, is composed of a disposable lithium thionyl chloride battery pack or a rechargeable lithium-ion battery pack. Its output is regulated and distributed by a power management chip to provide power to all electronic components within the node.

[0075] The mobile autonomous data carrier and calibration unit 20 is configured to work in conjunction with the static multimodal sensing node. The mobile autonomous data carrier and calibration unit 20 includes:

[0076] Navigation module 22 is used to drive the mobile autonomous data carrier and calibration unit to autonomously cruise to the position of the static multimodal perception node;

[0077] The navigation module 22 of the mobile autonomous data carrier and calibration unit 20 is used to drive its autonomous movement inside the cave. The navigation module 22 adopts a hybrid navigation mechanism, which includes: a line-following navigation unit 221, a real-time positioning and mapping navigation unit 222, and a navigation control unit 223 for coordinating and switching the operation of the two units.

[0078] The tracking navigation unit 221 is used to move along the main path pre-laid with navigation markers inside the cave. In one specific embodiment, this unit is installed in front of the chassis of the mobile autonomous data carrier and calibration unit 20, on a linear grayscale sensor array. This array is arranged in a straight line perpendicular to the direction of travel and is used to continuously detect the relative position of the navigation markers on the ground below. When the navigation control unit 223 activates the tracking navigation mode, this unit transmits the real-time readings of the grayscale sensor array to the navigation control unit 223.

[0079] The real-time positioning and mapping navigation unit 222 is used to perform movement in complex areas without navigation markers, or to avoid temporary obstacles on main paths. In a specific embodiment, this unit includes: a two-dimensional lidar for acquiring two-dimensional planar point cloud data of the surrounding environment;

[0080] An inertial measurement unit is used to acquire the angular velocity and linear acceleration of the mobile autonomous data carrier and calibration unit 20; and a wheel encoder connected to the drive wheels is used to provide odometer information.

[0081] The navigation control unit 223, such as a dedicated microprocessor, is electrically connected to the drive system of the tracking navigation unit 221, the real-time positioning and mapping navigation unit 222, and the mobile autonomous data carrier and calibration unit 20. The navigation control unit 223 is responsible for executing specific navigation algorithms and switching modes.

[0082] In line-following navigation mode, the navigation control unit 223 receives a grayscale array from the line-following navigation unit 221 and calculates the lateral offset error of the navigation marker line relative to the center of the sensor array based on the values. Subsequently, the offset error is input to a proportional-integral-derivative controller, which calculates the differential speed value of the left and right wheels required to correct the deviation and generates a corresponding pulse width modulation signal to send to the drive system to drive the mobile autonomous data carrier and calibration unit 20 to move stably along the navigation marker line.

[0083] The switching logic for the hybrid navigation mechanism is as follows:

[0084] On the main path, the system defaults to using the low-power line-following navigation mode. When the line-following navigation unit 221 detects a special marker preset on the navigation marker line, the marker indicates that a fork in the road has been reached at a static multimodal sensing node 10. At this time, the navigation control unit 223 performs a mode switch, disables the line-following navigation logic, and activates the real-time positioning and map building navigation unit 222.

[0085] In the real-time localization and mapping (RTD) navigation mode, the navigation control unit 223 integrates data from the 2D LiDAR, inertial measurement unit, and wheel encoder, and runs an RTD algorithm, such as filter-based Extended Kalman Filter (SLAM) or graph-optimized Cartographer algorithm, to accurately locate the mobile autonomous data carrier and calibration unit 20 in a pre-stored local map in real time. Subsequently, a path planning algorithm plans a collision-free path based on the pre-stored coordinates of the target static multimodal sensing node 10. The navigation control unit 223 then uses this path, combined with a local path planner such as a dynamic window method to avoid unforeseen small obstacles, to continuously generate drive commands until the predetermined docking position of the static multimodal sensing node 10 is reached.

[0086] After completing the collaborative operation with the static multimodal perception node 10, the navigation control unit 223 drives the mobile autonomous data carrier and calibration unit 20 to use real-time positioning and map construction to return to the main path. When the line-following navigation unit 221 re-stablely detects the navigation marker line, the navigation control unit 223 switches the navigation mode back to the line-following navigation mode and continues to cruise along the main path.

[0087] Data acquisition module 23 is used to acquire multi-source environmental data cached locally by static multimodal sensing nodes;

[0088] The mobile autonomous data carrier and calibration unit 20 includes a data acquisition module 23. This data acquisition module 23 is used to establish a communication link with the static multimodal sensing node 10 and acquire multi-source environmental data cached within the node after the mobile autonomous data carrier and calibration unit 20 arrives at the predetermined berthing position of the static multimodal sensing node 10.

[0089] In one specific embodiment, the data acquisition module 23 includes a communication transceiver that matches the near-field communication unit 14 in the static multimodal sensing node 10, such as a Bluetooth Low Energy master module, which is controlled by the main control unit 21 of the mobile autonomous data carrier and calibration unit 20.

[0090] The detailed data acquisition process is as follows:

[0091] Once the navigation module 22 confirms that the mobile autonomous data carrier and calibration unit 20 have reached the communication range of the target static multimodal perception node 10 and are stably docked, the main control unit 21 activates the data acquisition module 23 and initiates the data acquisition protocol.

[0092] First, the data acquisition module 23 enters the scanning state, listens for and searches for broadcast signals emitted by the target static multimodal sensing node 10. The broadcast signal contains the unique device identification code of the node. After receiving a matching broadcast signal, the data acquisition module 23 initiates a connection request to the static multimodal sensing node 10.

[0093] After the static multimodal sensing node 10 accepts the connection request, an encrypted pairing and authentication process is performed between the two to establish a secure communication link. This process employs a pre-shared key or asymmetric encryption mechanism to ensure the confidentiality and integrity of data exchange.

[0094] After the communication link is established, the mobile autonomous data carrier and calibration unit 20, acting as the main device, sends a control command to the static multimodal sensing node 10 via the link to begin data transmission. Upon receiving and parsing the command, the core processing unit 11 of the static multimodal sensing node 10 reads all data packets cached since the last access from its data storage unit 13.

[0095] Data transmission is performed in packet form. The static multimodal sensing node 10 encapsulates each data packet into a data frame and appends a cyclic redundancy check (CRC) code to the end of the frame. The data acquisition module 23 performs a CRC check on each received data frame. If the check passes, it returns an acknowledgment signal to the static multimodal sensing node 10.

[0096] If the verification fails, a non-acknowledgment signal is returned. After receiving the NACK signal, the static multimodal sensing node 10 will retransmit the data frame. This process is repeated until all buffered data packets are successfully transmitted.

[0097] After all data transmission is complete, the static multimodal sensing node 10 sends a transmission completion flag signal to the data acquisition module 23. Upon confirming receipt of all data, the data acquisition module 23 sends a disconnect command to the static multimodal sensing node 10. The communication link is then terminated, and the core processing unit 11 of the static multimodal sensing node 10 controls it to return to deep sleep mode.

[0098] All acquired multi-source environmental data is temporarily stored in the memory of the mobile autonomous data carrier and calibration unit 20, and immediately transferred by the main control unit 21 to the cruise analysis module 24 for subsequent environmental status analysis.

[0099] Cruise analysis module 24 is used to analyze the acquired multi-source environmental data to determine the environmental state of the location of the static multimodal sensing node;

[0100] The mobile autonomous data carrier and calibration unit 20 includes a cruise analysis module 24. This module's function is to analyze the acquired multi-source environmental data immediately after the data acquisition module 23 completes data acquisition from the static multimodal sensing node 10, in order to quantitatively assess the current environmental state of the node's location.

[0101] In one specific embodiment, the cruise analysis module 24 is implemented by a dedicated algorithm program running on the main control unit 21 inside the mobile autonomous data carrier and calibration unit 20. Its specific workflow is as follows:

[0102] The cruise analysis module 24 receives the complete dataset from the data acquisition module 23. This dataset contains a series of timestamped data packets. The module first parses the dataset, classifying the data according to sensor type and arranging it in chronological order.

[0103] Subsequently, for each sensor type requiring condition assessment, the cruise analysis module 24 selects the latest or most recent data points and calculates its environmental anomaly index. This index is a comprehensive quantitative indicator, and its calculation formula is as follows:

[0104] ;

[0105] The definition of each symbol in the above formula is as follows:

[0106] : indicates that the number is The static multimodal sensing node at time An environmental anomaly index for a specific sensor.

[0107] , , : These are pre-set weighting coefficients used to adjust the sensitivity of different analysis dimensions, and the sum of the three is 1.

[0108] Normalized amplitude deviation is used to measure the degree of deviation of the current measurement value from its own long-term statistical regularity.

[0109] The calculation method is as follows: ;

[0110] in:

[0111] : indicates that the number is At a specific time on a node The measured value.

[0112] : Indicates that the sensor is at the node The long-term historical average value is stored in the memory of the mobile autonomous data carrier and calibration unit 20 and is updated after each cruise.

[0113] : Indicates that the sensor is at the node The long-term historical standard deviation is also stored and updated periodically.

[0114] : This is the normalized spatial gradient, used to measure the difference between the current measurement and its geographical neighbors, in order to identify local anomalies in space.

[0115] The calculation method is as follows:

[0116] in:

[0117] , indicating at time ,node set of geographically nearest nodes The arithmetic mean of all node measurements.

[0118] gather During the system deployment phase, the physical distance between nodes is predefined and stored.

[0119] For set The number of nodes in the middle.

[0120] For set any neighboring node At any moment The measured value.

[0121] : Indicates the time. , by node and its neighboring node set The standard deviation of the sample set consisting of the measurements of all nodes.

[0122] This is the normalized rate of change over time, used to measure the drastic change in measured values ​​to capture sudden environmental events. Its calculation method is as follows:

[0123] ;

[0124] in:

[0125] , indicating at time The instantaneous rate of change of time.

[0126] This is the measurement value of the sensor at the previous sampling time. This represents the sampling time interval of the sensor.

[0127] : Indicates that the sensor is at the node The absolute value of the historical maximum rate of change has been recorded and is stored in the memory of the mobile autonomous data carrier and calibration unit 20, and is updated when a new, larger rate of change is discovered.

[0128] Cruise analysis module 24 calculates the corresponding environmental anomaly index for each key sensor on the current static multimodal sensing node 10. Then, these index values ​​are collected as a set and output to the task reconstruction module 25 for subsequent decision-making and instruction generation.

[0129] The task reconstruction module 25 is used to generate and issue task reconstruction instructions to static multimodal perception nodes based on the environmental state, so as to adjust their subsequent perception behavior.

[0130] The mobile autonomous data carrier and calibration unit 20 includes a mission reconfiguration module 25. This module receives an environmental anomaly index from the cruise analysis module 24 and makes decisions based on this index to generate mission reconfiguration instructions for adjusting the subsequent perception behavior of the static multimodal perception node 10.

[0131] In one specific embodiment, the mission reconfiguration module 25 is implemented by a decision logic program running on the main control unit 21 inside the mobile autonomous data carrier and calibration unit 20. It receives a set of environmental anomaly indices output by the cruise analysis module 24, targeting the various key sensors on the current static multimodal sensing node 10. Then, the following decision-making process is executed:

[0132] Task refactoring module 25 will process each received environmental anomaly index It is compared with two preset thresholds: a high-order threshold. and a low-bit threshold .

[0133] These two thresholds are stored in the configuration parameters of the mobile autonomous data carrier and calibration unit 20, where .

[0134] The decision-making logic is as follows: If the environmental anomaly index of any sensor... Above the high threshold ,Right now The task reconstruction module 25 determines that the current environmental state has undergone significant abnormalities or drastic changes, requiring high-frequency and high-precision monitoring.

[0135] If the environmental anomaly index of all sensors All are below the low threshold That is, it satisfies the following for all sensors. The task reconstruction module 25 determines that the current environmental state is very stable, and can reduce the monitoring frequency to save energy consumption of the static multimodal sensing node 10.

[0136] If none of the environmental anomaly indices from all sensors meet the above two conditions, meaning at least one sensor's index is within the range... and Between, and none of the indices are higher than The task refactoring module 25 determines that the current environment is fluctuating normally and there is no need to adjust the existing working parameters.

[0137] Based on the above decision results, the task refactoring module 25 executes the corresponding instruction generation operation:

[0138] When it is determined that high-frequency monitoring is required, the module generates a task reconfiguration instruction to shorten the sensor sampling cycle.

[0139] For example, if the current sampling period of the static multimodal sensing node 10 is The module will be based on the preset acceleration factor. (For example Calculate a new, shorter sampling period. .

[0140] When it is determined that the monitoring frequency can be reduced, the module generates a task reconfiguration instruction to extend the sensor sampling period. For example, the module will adjust the reconfiguration instruction based on a preset deceleration factor. (For example Calculate a new, longer sampling period .

[0141] When it is determined that no adjustment is needed, the module does not generate a task refactoring instruction, or generates an empty instruction that keeps the current parameters unchanged.

[0142] The generated task refactoring instructions are encapsulated into a data packet of a specific format. This data packet includes an instruction header, target node ID, instruction code, and the specific parameters of the instruction.

[0143] Finally, the task reconstruction instruction data packet is sent to the target static multimodal sensing node 10 through the same communication link used by the data acquisition module 23. After receiving and parsing the instruction, the core processing unit 11 of the static multimodal sensing node 10 updates its internally stored operating parameters and starts operating according to the new sampling period from the next sampling period. Perform perception tasks.

[0144] See attached document Figure 4 Another embodiment of the present invention provides a cave ecological monitoring method based on multi-source sensors, comprising the following steps:

[0145] S501, the static multimodal sensing node 10, at its deployment site, performs periodic data acquisition and storage according to its internally set current sampling period.

[0146] The steps specifically include: being awakened from deep sleep mode, controlling its multimodal sensor array 12 to perform a synchronous or time-division data acquisition, performing edge computing on the acquired specific raw data to extract feature vectors, then encapsulating all measurement data and feature vectors with the current timestamp into a data packet, storing it in its local data storage unit 13, and then entering deep sleep mode again.

[0147] S502, the mobile autonomous data carrier and calibration unit 20, according to the preset task schedule, start from their base station to carry out a cruise data collection task.

[0148] S503, the mobile autonomous data carrier and calibration unit 20 uses its navigation module 22 to autonomously cruise to the position of the first or next target static multimodal perception node 10. On the main path, it uses the tracking navigation unit 221 to move.

[0149] When approaching the fork in the road near the target node, the system switches to the real-time positioning and map building navigation unit 222 based on the preset markings on the path to perform precise positioning and navigation of the end area until the predetermined parking position is reached.

[0150] S504. After arriving at the berthing position, the data acquisition module 23 of the mobile autonomous data carrier and calibration unit 20 establishes an encrypted communication link with the near-field communication unit 14 of the target static multimodal perception node 10, and acquires all multi-source environmental data packets cached in its data storage unit 13 since the last access.

[0151] S505. After data acquisition is completed, the cruise analysis module 24 of the mobile autonomous data carrier and calibration unit 20 immediately analyzes the newly acquired data and calculates the environmental anomaly index for each key sensor of the node. The index combines the magnitude deviation of the measured value relative to its historical statistical characteristics, the spatial gradient relative to its geographical neighbors, and its own rate of change over time.

[0152] S506, Task Reconstruction Module 25 receives the environmental anomaly index and compares it with a preset high threshold. and low-bit threshold If any index is higher than the previous one, then... If all indices are below a certain level, then it is determined that the monitoring frequency needs to be increased; if all indices are below a certain level... If the current monitoring frequency is maintained, it is determined that the monitoring frequency can be reduced; otherwise, it is determined that the current monitoring frequency should be maintained.

[0153] S507. Based on the determination result of step S506, the task reconstruction module 25 generates a corresponding task reconstruction instruction, which includes specific parameter values ​​for updating the sampling period of the static multimodal sensing node 10. Subsequently, the instruction is sent to the static multimodal sensing node 10 through the established communication link.

[0154] S508. While performing data exchange and mission reconfiguration, or afterward, the mobile autonomous data carrier and calibration unit 20 may optionally perform in-situ calibration and maintenance tasks.

[0155] This step includes: driving its standard reference sensor and the sensor of the static multimodal sensing node 10 to perform parallel measurements to obtain calibration data, and its maintenance module cleaning the sensor probe of the static multimodal sensing node 10.

[0156] S509. After completing all operations for a node, the mobile autonomous data carrier disconnects the communication link with the calibration unit 20 and continues to cruise to the next static multimodal sensing node according to the cruise mission list stored in its internal storage, and repeats steps S503 to S508. If all nodes have been visited, then step S510 is executed.

[0157] The S510, mobile autonomous data carrier and calibration unit 20 autonomously return to the base station using its navigation module 22. After returning to the base station, it uploads all the multi-source environmental data and in-situ calibration data collected during this cruise to the back-end server through the network interface of the base station. The back-end server uses the in-situ calibration data to correct the multi-source environmental data and generate the final cave ecological environment dataset.

[0158] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-source sensor based cave ecological monitoring system, characterized in that, Comprising: at least one static multi-modal sensing node for periodically collecting multi-source environmental data and locally caching; a mobile autonomous data mother ship and calibration unit configured to work in cooperation with the static multi-modal sensing node, the mobile autonomous data mother ship and calibration unit comprising: a navigation module for driving the mobile autonomous data mother ship and calibration unit to autonomously cruise to the location of the static multi-modal sensing node; a data acquisition module for acquiring the multi-source environmental data locally cached by the static multi-modal sensing node; a cruise analysis module for analyzing the acquired multi-source environmental data to determine the environmental state of the location where the static multi-modal sensing node is located; a task reconstruction module for generating and issuing task reconstruction instructions to the static multi-modal sensing node based on the environmental state to adjust its subsequent sensing behavior.

2. The cave ecological monitoring system based on multi-source sensors according to claim 1, characterized in that, The cruise analysis module comprises: quantitative assessment of the environmental state based on at least one of the current measurement value of the static multi-modal sensing node, its historical statistical characteristics, the measurement value of its neighboring nodes, and its time variation rate, to identify the degree of abnormality of the environment.

3. The cave ecological monitoring system based on multi-source sensors according to claim 1, characterized in that, The task reconstruction module comprises: when the degree of abnormality of the environment is higher than a preset high threshold, generating a task reconstruction instruction for shortening the sensor sampling period of the static multi-modal sensing node; when the degree of abnormality of the environment is lower than a preset low threshold, generating a task reconstruction instruction for lengthening the sensor sampling period of the static multi-modal sensing node.

4. The multi-source sensor based cave ecological monitoring system according to claim 1, wherein, The mobile autonomous data mother ship and calibration unit further comprises: a standard reference sensor; a mechanism for driving the standard reference sensor to perform parallel measurement with the sensor of the static multi-modal sensing node to acquire in-situ calibration data.

5. The multi-source sensor based cave ecological monitoring system according to claim 1, wherein, Further comprising a backend server for: using the in-situ calibration data to correct the multi-source environmental data collected by the static multi-modal sensing node to compensate for the measurement drift of the sensor.

6. The multi-source sensor based cave ecological monitoring system according to claim 1, wherein, The mobile autonomous data mother ship and calibration unit further comprises: a maintenance module for in-situ cleaning and maintenance of the sensor probe of the static multi-modal sensing node.

7. The multi-source sensor based cave ecological monitoring system according to claim 1, wherein, The static multi-modal sensing node comprises at least one of the following sensors: a sensor for collecting physical parameters such as temperature, humidity, or air pressure; a sensor for collecting chemical parameters such as carbon dioxide concentration; a sensor for collecting biological volatile organic compound data; a sensor for collecting acoustic event features.

8. The multi-source sensor based cave ecological monitoring system according to claim 7, wherein, The static multi-modal sensing node further comprises an edge computing unit for: processing the collected raw acoustic signals to extract acoustic feature vectors of dripping frequency or biological calling as the acoustic event features.

9. The multi-source sensor based cave ecological monitoring system according to claim 1, wherein, The navigation module is configured to: adopt a tracking navigation on a preset path, and a hybrid navigation mechanism combining real-time positioning and map construction navigation when approaching the static multi-modal sensing node.

10. A multi-source sensor-based cave ecological monitoring method, according to the multi-source sensor-based cave ecological monitoring system of any one of claims 1-9, characterized in that, Comprising the following steps: periodically collecting multi-source environmental data and locally caching by the static multi-modal sensing node; by the mobile autonomous data mother ship and calibration unit, the following steps are performed: autonomously cruise to the location of the static multi-modal sensing node; acquiring multi-source environment data cached locally by the static multi-modal perception node; performing in-cruise analysis on the acquired multi-source environment data to determine an environment state of a location where the static multi-modal perception node is located; generating task reconstruction instructions for the static multi-modal perception node based on the environment state; issuing the task reconstruction instructions to the static multi-modal perception node to adjust its subsequent perception behavior.

Images