A cloud-edge collaboration based deep hole geological monitoring method and system

Abstract

The application relates to the technical field of geological monitoring, and discloses a deep-hole geological monitoring method and system based on cloud-edge cooperation, which aims to solve the problems of poor data processing efficiency and reliability of the existing method, and mainly comprises the following steps: collecting multi-source sensor signals in a deep hole; transmitting the multi-source sensor signals to a private cloud edge based on an anti-interference transmission layer; converting differential signals into standardized digital signal streams by the private cloud edge; transmitting the standardized digital signal streams to a GPU cloud processing center based on bandwidth scheduling after real-time preprocessing of the standardized digital signal streams; generating a space-time alignment tensor and a geological monitoring task by the GPU cloud processing center; executing the geological monitoring task in a virtual machine; and triggering an active restart protection process when running to a maximum allowed time if the success probability of executing the geological monitoring task by the virtual machine is not greater than a probability threshold. The application improves the data processing efficiency and reliability, and is suitable for geological disaster early warning or engineering safety evaluation.

Description

Technical Field

[0001] This invention relates to the field of geological monitoring technology, specifically to a deep borehole geological monitoring method and system based on cloud-edge collaboration. Background Technology

[0002] Deep-hole geological monitoring refers to a technical system that uses drilling equipment to implant sensors or detection devices deep underground (usually exceeding 300 meters or even thousands of meters) to conduct long-term or real-time observations of geological structure, resource distribution, and disaster risks. Deep-hole geological monitoring enables geological disaster early warning and engineering safety assessments.

[0003] In the field of deep borehole monitoring, the existing technology typically employs a solution where multi-source sensors directly transmit analog signals to surface equipment via ordinary shielded cables, and the raw data is directly uploaded to a remote cloud center. This approach has at least the following drawbacks: First, the existing solution relies solely on ordinary physical shielding for interference resistance, making it difficult to cope with common-mode interference caused by frequency converters within the borehole and the radio frequency absorption effect of rock strata. It is also susceptible to electromagnetic interference, temperature drift, and long-distance attenuation within the deep borehole, leading to severe loss of effective data or failure of real-time data transmission. Second, the standardization processing of heterogeneous data from multi-source sensors is limited by the CPU serial architecture. A single computing engine cannot simultaneously resolve the format barriers of various heterogeneous data, resulting in data processing time accounting for nearly half of the exploration cycle. Furthermore, the inefficient conversion capabilities of existing cloud CPU solutions create a bottleneck in the process. Third, the existing solution lacks intelligent fault tolerance mechanisms, relying solely on a passive recovery mode with manual intervention. This approach results in long fault recovery times, service interruption times far exceeding real-time thresholds, and poor reliability. Summary of the Invention

[0004] This invention aims to address the problems of poor data processing efficiency and reliability in existing deep-hole geological monitoring schemes, and proposes a deep-hole geological monitoring method and system based on cloud-edge collaboration.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a deep borehole geological monitoring method based on cloud-edge collaboration, the method comprising:

[0007] Acquire signals from multiple sensor sources inside deep holes;

[0008] The multi-source sensor signals are transmitted to the private cloud edge based on the anti-interference transmission layer. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize the multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal.

[0009] The private cloud edge converts the differential signal into a standardized digital signal stream, performs real-time preprocessing on the standardized digital signal stream, and then transmits it to the GPU cloud processing center based on bandwidth scheduling.

[0010] The GPU cloud processing center performs parallel heterogeneous operations on the preprocessed standardized digital signal stream and generates a spatiotemporal aligned tensor. It then generates a geological monitoring task corresponding to the spatiotemporal aligned tensor for each preset time period, and the geological monitoring task is submitted to the virtual machine for execution.

[0011] Calculate the success probability of the virtual machine executing the geological monitoring task, and determine whether the success probability is greater than the probability threshold. If it is, the virtual machine executes the corresponding geological monitoring task. If not, calculate the maximum allowable time when the success probability is equal to the probability threshold, and trigger the active restart protection process when the maximum allowable time is reached.

[0012] Furthermore, the method for calculating the success probability includes:

[0013] Calculate the probability of success based on the time required to execute the geological monitoring task, the failure rate of the virtual machine, and the failure rate of the server hardware:

[0014] ;

[0015] in, Indicates the probability of success. This indicates the failure rate of the virtual machine. This indicates the failure rate of the server hardware. This indicates the time required to perform a geological monitoring task. Represents the natural constant.

[0016] Furthermore, the formula for calculating the maximum permissible time is as follows:

[0017] ;

[0018] in, Indicates the maximum allowed time. This represents the probability threshold.

[0019] Furthermore, the proactive restart protection process includes:

[0020] Freeze the process and capture a runtime snapshot, generate snapshot data, compress and encrypt the snapshot data, and then store it in a cross-node disaster recovery storage pool.

[0021] The corresponding geological monitoring task is scheduled to the healthy node. The healthy node obtains snapshot data from the cross-node disaster recovery storage pool, decrypts and decompresses the snapshot data, loads it into the new environment, and restores the process state.

[0022] Furthermore, the anti-interference cable comprises, from the inside out, a shielding layer, a silver-plated copper wire braided mesh, and a nano-magnetic absorbing layer.

[0023] Furthermore, the enhanced differential driver includes a signal input interface, a preprocessing module, a level converter, a drive enhancement circuit, a differential signal generation module, and a signal output interface;

[0024] The signal input interface is used to receive signals from multiple sensor sources;

[0025] The preprocessing module is used to perform pre-emphasis adaptive processing and edge-jump compensation processing on multi-source sensor signals to generate preprocessed signals.

[0026] The level converter is used to boost the preprocessed signal to generate a high-voltage signal;

[0027] The drive enhancement circuit is used to amplify the power of the high-voltage signal to generate an enhanced high-voltage signal.

[0028] The differential signal generation module is used to sample the enhanced high voltage signal in real time using a fully differential operational amplifier, extract the common-mode voltage, and generate a differential signal with a high common-mode rejection ratio.

[0029] The signal output module is used to output differential signals.

[0030] Furthermore, the multi-source sensor signals include sensor signals corresponding to seismic wave data, gamma spectrum data, and infrared video stream data.

[0031] Furthermore, the GPU cloud processing center performs parallel heterogeneous operations on the preprocessed standardized digital signal stream, including:

[0032] For the preprocessed standardized digital signal stream corresponding to the seismic wave data, the GPU cloud processing center starts the CUDA kernel function to perform db4 wavelet basis noise reduction.

[0033] For the preprocessed standardized digital signal stream corresponding to the γ energy spectrum data, the GPU cloud processing center initiates parallel Savitzky-Golay peak correction.

[0034] For the preprocessed standardized digital signal stream corresponding to the infrared video stream data, the GPU cloud processing center activates the NVDEC hardware decoder to extract the heat matrix.

[0035] Furthermore, the spatiotemporal alignment tensor is generated, including:

[0036] The results of parallel heterogeneous computation are spatiotemporally aligned based on the DTW algorithm to generate a spatiotemporally aligned tensor. During the computation of the DTW algorithm, the computation of time-warped paths is decomposed into GPU thread blocks, and the shared memory access mode is optimized.

[0037] Secondly, the present invention provides a deep borehole geological monitoring system based on cloud-edge collaboration, used to implement the deep borehole geological monitoring method based on cloud-edge collaboration described in the first aspect, the system comprising:

[0038] Multi-source sensor, used to collect multi-source sensor signals inside deep holes;

[0039] An anti-interference transmission layer is used to transmit the multi-source sensor signals to the private cloud edge. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize the multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal.

[0040] The private cloud edge is used to convert the differential signal into a standardized digital signal stream. After real-time preprocessing of the standardized digital signal stream, it is transmitted to the GPU cloud processing center based on bandwidth scheduling.

[0041] The GPU cloud processing center is used to perform parallel heterogeneous operations on the preprocessed standardized digital signal stream and generate a spatiotemporal aligned tensor. It generates a geological monitoring task corresponding to the spatiotemporal aligned tensor in each preset time period, and the geological monitoring task is submitted to the virtual machine for execution.

[0042] The fault handling model is used to calculate the success probability of the virtual machine executing a geological monitoring task, determine whether the success probability is greater than a probability threshold, if so, the virtual machine executes the corresponding geological monitoring task; if not, it calculates the maximum allowable time when the success probability is equal to the probability threshold, and triggers an active restart protection process when the maximum allowable time is reached.

[0043] The beneficial effects of this invention are as follows: The cloud-edge collaborative deep-hole geological monitoring method and system provided by this invention not only physically shields the multi-source sensor signals through anti-interference cables during transmission, but also converts non-standard signals into standard differential signals through enhanced differential drivers, and eliminates strong electromagnetic interference in deep holes through common-mode rejection ratio, thus avoiding signal distortion problems in extreme deep-hole environments; it utilizes a private cloud edge to complete signal conversion and preprocessing, reducing the amount of data uploaded, and dynamically allocates transmission channels based on task priority to ensure low-latency uploading of critical data, reducing latency and saving bandwidth; through GPU parallel heterogeneous computing, the geological monitoring tasks corresponding to the spatiotemporal aligned tensors of each time period are distributed to virtual machines, avoiding the problem of analysis lag caused by the inability of CPU serial processing to parse multi-source heterogeneous data in real time, thus improving data processing efficiency; and through predictive proactive protection, it shortens fault recovery time and improves reliability. Attached Figure Description

[0044] Figure 1A schematic flowchart of a deep borehole geological monitoring method based on cloud-edge collaboration provided for an embodiment;

[0045] Figure 2 A schematic diagram of the anti-interference cable provided in the embodiment;

[0046] Figure 3 A schematic diagram of the structure of a deep borehole geological monitoring system based on cloud-edge collaboration provided in this embodiment;

[0047] Explanation of reference numerals in the attached diagram: 1-Shielding layer, 2-Silver-plated copper wire braided mesh, 3-Nano magnetic absorbing layer. Detailed Implementation

[0048] The technical solution of the present invention is applicable to application scenarios that require deep-hole geological exploration, such as geological disaster early warning or engineering safety assessment.

[0049] Current deep-hole geological exploration methods typically transmit analog signals from multi-source sensors to surface equipment via ordinary shielded cables, and then directly upload the raw data to a remote cloud center. The inventors discovered that this method suffers from at least the following problems: poor anti-interference capability, large data transmission delay, low data processing efficiency, and long fault recovery time.

[0050] Based on this, the technical solution of this invention is proposed. In this invention, after acquiring multi-source sensor signals from deep holes, the multi-source sensor signals are transmitted to the private cloud edge based on an anti-interference transmission layer. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver, by standardizing, suppressing common-mode, and enhancing the drive of the multi-source sensor signals, can eliminate strong electromagnetic interference in deep holes and increase signal power, thereby significantly improving signal quality and avoiding signal distortion. In this invention, the private cloud edge converts the differential signals into a standardized digital signal stream. After real-time preprocessing of the standardized digital signal stream, it is transmitted to the GPU cloud processing center based on bandwidth scheduling. This invention compresses data scale through edge preprocessing and optimizes bandwidth utilization by combining scheduling strategies, reducing the burden on the cloud while ensuring timely processing of critical data. The GPU cloud processing center performs parallel heterogeneous operations on the preprocessed standardized digital signal stream and generates a spatiotemporally aligned tensor. It then generates geological monitoring tasks corresponding to the spatiotemporally aligned tensor within each preset time period, and these geological monitoring tasks are submitted to a virtual machine for execution. Spatiotemporal alignment tensors require the fusion of multi-source sensor data, but existing CPU serial processing cannot parse multi-source heterogeneous data in real time, resulting in slow processing speed and delayed analysis. This invention, through GPU heterogeneous acceleration and tensor modeling, can quickly generate spatiotemporal alignment tensors, achieving second-level detection of geological anomalies. This invention also includes a predictive proactive protection process: calculating the success probability of a virtual machine executing a geological monitoring task, determining whether the success probability is greater than a probability threshold; if so, the virtual machine executes the corresponding geological monitoring task; if not, calculating the maximum allowable time when the success probability is at the probability threshold, and triggering an active restart protection process when the maximum allowable time is reached. By calculating the success probability of a geological monitoring task being successfully executed, and when the probability is lower than the threshold, the corresponding geological monitoring task is proactively scheduled to a healthy node at the critical time point, thereby achieving accurate fault determination and autonomous state transfer, avoiding fault recovery time caused by virtual machine or hardware failures, and improving system reliability.

[0051] The technical solutions in this embodiment 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.

[0052] Figure 1 A schematic diagram of a massage device is shown below. Please refer to [link / reference]. Figure 1 The method includes the following steps:

[0053] Step 1: Collect signals from multiple sensor sources inside the deep hole.

[0054] In this embodiment, the multi-source sensor signals include sensor signals corresponding to seismic wave data, gamma spectrum data, and infrared video stream data.

[0055] Among these, the sensor for acquiring seismic wave data can be a seismograph, which can monitor microseismic activity in rock strata and fault activity. The sensor for acquiring gamma-ray spectrum data can be a gamma-ray spectrum probe, which can identify radioactive anomalies and mineral veins. The sensor for acquiring infrared thermal imager data can be an infrared thermal imager, which can detect seepage zones and geothermal anomalies through infrared video stream data.

[0056] Step 2: Transmit the multi-source sensor signals to the private cloud edge based on the anti-interference transmission layer.

[0057] The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal.

[0058] Please see Figure 2 In this embodiment, the anti-interference cable includes, from the inside out, a shielding layer 1, a silver-plated copper wire braided mesh 2, and a nano-magnetic absorbing layer 3.

[0059] Among them, shielding layer 1 (such as copper foil / aluminum foil shielding layer) can suppress electric field coupling noise, isolate cross-interference between multi-source sensors, and prevent signal crosstalk; silver-plated copper wire braided mesh 2 can resist high-frequency radiation noise and improve the tensile strength of the cable; nano-magnetic absorbing layer 3 can absorb low-frequency magnetic field interference and has strong resistance to mechanical damage. The above-mentioned anti-interference cable can effectively avoid the combined interference of electric field, magnetic field and radiation field coexisting in deep hole monitoring through the synergistic mechanism of electric field shielding, high-frequency reflection and low-frequency magnetic absorption, and improve the tensile strength and wear resistance of the cable.

[0060] In this embodiment, the enhanced differential driver includes a signal input interface, a preprocessing module, a level converter, a drive enhancement circuit, a differential signal generation module, and a signal output interface. The signal input interface receives signals from multiple sensor sources. The preprocessing module performs pre-emphasis adaptive processing and edge-jump compensation on the multi-source sensor signals to generate a preprocessed signal. The level converter boosts the preprocessed signal to generate a high-voltage signal. The drive enhancement circuit amplifies the high-voltage signal to generate an enhanced high-voltage signal. The differential signal generation module performs real-time sampling of the enhanced high-voltage signal using a fully differential operational amplifier to extract the common-mode voltage and generate a differential signal with a high common-mode rejection ratio. The signal output module outputs the differential signal.

[0061] In practical applications, the signal input interface, as the entry point for receiving sensor signals, primarily serves to provide a physically and electrically compatible connection point and provides basic protection or buffering. The preprocessing module, through pre-emphasis adaptation and edge-jump compensation, can compensate for high-frequency attenuation and suppress edge degradation caused by cable capacitive loads. The level converter employs a differential current source structure, with the input signal controlling the current flow along two paths, generating a high-voltage swing signal across the load resistor and boosting the core logic voltage from 1.2V to the driver stage's high-voltage domain of 3.3V, ensuring the output stage transistors are fully turned on. In static conditions, it will... The MOSFET is biased in a weak conduction state, eliminating the switching delay of traditional level converters. Voltage boosting enhances signal attenuation resistance. The drive enhancement circuit provides a high current drive capability of 5mA~20mA, achieving signal power amplification and short-circuit protection. The differential signal generation module uses a fully differential operational amplifier to sample the common-mode voltage in real time, generating a differential signal with a high common-mode rejection ratio (CMRR>60dB) to suppress common-mode noise. The signal output interface serves as the physical endpoint connecting the driver to external transmission lines, its main function being to achieve low-loss connection between the chip and the transmission line, integrating ESD protection and impedance matching. The enhanced differential driver, by sequentially preprocessing, level-converting, driving enhancement, and differential generation of multi-source sensor signals, solves the problems of signal attenuation over long distances, signal overwhelming by common-mode interference, and high-frequency pulse waveform distortion, achieving lossless transmission of multi-source sensor signals from the sensor to the cloud.

[0062] Step 3: The private cloud edge converts the differential signal into a standardized digital signal stream, performs real-time preprocessing on the standardized digital signal stream, and then transmits it to the GPU cloud processing center based on bandwidth scheduling.

[0063] Specifically, the private cloud edge receives differential signals through a streaming data access gateway and samples them using the gateway's ADC to generate a standardized digital signal stream. After real-time preprocessing such as noise filtering, electromagnetic interference cancellation, and invalid data removal, the stream is dynamically allocated a transmission channel based on task priority and transmitted directly to the GPU cloud processing center via TCP / IP.

[0064] Step 4: The GPU cloud processing center performs parallel heterogeneous operations on the preprocessed standardized digital signal stream and generates a spatiotemporal aligned tensor. It then generates a geological monitoring task corresponding to the spatiotemporal aligned tensor for each preset time period, and submits the geological monitoring task to the virtual machine for execution.

[0065] In this embodiment, the streaming data access gateway is segmented according to sensor type, driving the GPU to perform parallel preprocessing:

[0066] For the preprocessed, standardized digital signal stream corresponding to seismic wave data, the GPU cloud processing center initiates CUDA kernel functions to perform db4 wavelet basis denoising, thereby filtering out high-frequency noise while preserving signal characteristics. For the preprocessed, standardized digital signal stream corresponding to gamma-ray spectrum data, the GPU cloud processing center initiates parallel Savitzky-Golay peak correction, thereby smoothing the spectrum curve and enhancing peak position identification. For the preprocessed, standardized digital signal stream corresponding to infrared video stream data, the GPU cloud processing center activates the NVDEC hardware decoder to extract the thermal matrix, thereby converting the video stream into a numerical tensor.

[0067] Secondly, the spatiotemporal alignment tensor is generated, including: spatiotemporally aligning the results of parallel heterogeneous operations based on the DTW algorithm to generate the spatiotemporal alignment tensor. During the calculation process of the DTW algorithm, the calculation of the time warp path is decomposed into GPU thread blocks, and the shared memory access mode is optimized.

[0068] This embodiment achieves microsecond-level spatiotemporal alignment based on the improved DTW (Dynamic Time Warping) algorithm. By decomposing the time warping path calculation into GPU thread blocks (Block size=256) and optimizing shared memory access, the algorithm complexity is reduced from O(n²) to O(n), and the output is a standardized spatiotemporal alignment tensor with an alignment error ≤1μs.

[0069] For each preset time period, a corresponding geological monitoring task is generated based on the spatiotemporal alignment tensor and submitted to the virtual machine for execution. When executing the geological monitoring task, the virtual machine determines whether there is a geological risk in the corresponding time period based on the corresponding spatiotemporal alignment tensor and outputs the geological monitoring results. When a geological risk is determined to exist, a corresponding alarm can also be issued. The duration of the preset time period is set according to the actual monitoring needs, and this embodiment does not impose any restrictions on it.

[0070] Step 5: Calculate the success probability of the virtual machine executing the geological monitoring task, and determine whether the success probability is greater than the probability threshold. If yes, the virtual machine executes the corresponding geological monitoring task. If no, calculate the maximum allowable time when the success probability is equal to the probability threshold, and trigger the active restart protection process when the maximum allowable time is reached.

[0071] It is understandable that the above steps constitute a predictive proactive protection process. Specifically, in a real-world cloud environment, cloud services may experience both virtual machine (VM) and server hardware (PS) failures. Assume that the time elapsed between the occurrence of either failure and the start of task execution follows a Poisson process. For each execution of a geological monitoring task by a virtual machine, let the time required for the virtual machine to successfully execute the geological monitoring task be... The failure rate of virtual machines is The server hardware failure rate The virtual machine failure occurred at the time of The time of the server hardware failure was The running time before the fault occurred was ,but ,when This indicates that the geological monitoring task could not be executed successfully, and the probability of the geological monitoring task failing to execute is due to virtual machine failure or server hardware failure. It can be represented as:

[0072] ;

[0073] The success rate of a virtual machine executing a geological monitoring task can be expressed as:

[0074] ;

[0075] in, express The probability, express and The probability, This indicates the probability of a virtual machine successfully executing a geological monitoring task. This indicates the failure rate of the virtual machine. The failure rate of server hardware can be obtained through analysis of historical operation logs, hardware inspection reports, and other methods. This indicates the time required to perform a geological monitoring task. Represents the natural constant.

[0076] In practical applications, when the success probability When the probability exceeds the threshold, the virtual machine directly executes the corresponding geological monitoring task. Calculate the success probability when it is less than or equal to the probability threshold. The maximum allowed time is set at the probability threshold. When the maximum allowed time is reached, the active restart protection process is triggered.

[0077] In this embodiment, the maximum allowable time The calculation formula is as follows:

[0078] ;

[0079] in, Indicates the maximum allowed time. This represents the probability threshold, which can be set according to actual monitoring needs, such as 95%.

[0080] In this embodiment, the proactive restart protection process includes: freezing the process and capturing a runtime snapshot, generating snapshot data, compressing and encrypting the snapshot data, and storing it in a cross-node disaster recovery storage pool; scheduling the corresponding geological monitoring task to a healthy node, the healthy node obtaining the snapshot data from the cross-node disaster recovery storage pool, decrypting and decompressing the snapshot data, loading it into a new environment, and restoring the process state.

[0081] Specifically, when the success probability Less than or equal to the probability threshold At that time, The current computing process is frozen and a runtime snapshot, including video memory data, register status, and I / O operation pipeline, is captured and stored in a cross-node disaster recovery storage pool after compression and encryption. Then, the current execution instance is actively terminated and the computing task is scheduled to a pre-checked healthy node within 500 milliseconds. The runtime context is precisely reconstructed to achieve seamless resume from the last saved state and continue the execution of the geological monitoring task.

[0082] The above process enables accurate fault diagnosis and autonomous state transfer, avoiding fault recovery time caused by virtual machine or hardware failures and improving system reliability.

[0083] In summary, the cloud-edge collaborative deep-hole geological monitoring method provided in this embodiment utilizes an enhanced differential driver to eliminate strong electromagnetic interference in deep holes and increase signal power by standardizing, suppressing common modes, and enhancing the drive of multi-source sensor signals. This significantly improves signal quality and avoids signal distortion. Edge preprocessing compresses data volume, and scheduling strategies optimize bandwidth utilization, reducing the burden on the cloud while ensuring timely processing of critical data. GPU heterogeneous acceleration and tensor modeling enable rapid generation of spatiotemporally aligned tensors, achieving second-level detection of geological anomalies. By calculating the success probability of geological monitoring tasks, when the probability falls below a threshold, the corresponding geological monitoring task is proactively scheduled to a healthy node at a critical time point. This achieves accurate fault determination and autonomous state transfer, avoiding recovery time caused by virtual machine or hardware failures and improving system reliability.

[0084] Based on the above technical solution, this embodiment also proposes a cloud-edge collaborative deep borehole geological monitoring system to implement the cloud-edge collaborative deep borehole geological monitoring method described in this embodiment. Please refer to [link to relevant documentation]. Figure 3 The system includes:

[0085] Multi-source sensor, used to collect multi-source sensor signals inside deep holes;

[0086] An anti-interference transmission layer is used to transmit the multi-source sensor signals to the private cloud edge. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize the multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal.

[0087] The private cloud edge is used to convert the differential signal into a standardized digital signal stream. After real-time preprocessing of the standardized digital signal stream, it is transmitted to the GPU cloud processing center based on bandwidth scheduling.

[0088] The GPU cloud processing center is used to perform parallel heterogeneous operations on the preprocessed standardized digital signal stream and generate a spatiotemporal aligned tensor. It generates a geological monitoring task corresponding to the spatiotemporal aligned tensor in each preset time period, and the geological monitoring task is submitted to the virtual machine for execution.

[0089] The fault handling model is used to calculate the success probability of the virtual machine executing a geological monitoring task, determine whether the success probability is greater than a probability threshold, if so, the virtual machine executes the corresponding geological monitoring task; if not, it calculates the maximum allowable time when the success probability is equal to the probability threshold, and triggers an active restart protection process when the maximum allowable time is reached.

[0090] It is understood that since the cloud-edge collaborative deep borehole geological monitoring system described in this embodiment is a system for implementing the cloud-edge collaborative deep borehole geological monitoring method described in the embodiment, the system disclosed in the embodiment is relatively simple to describe because it corresponds to the method disclosed in the embodiment. For relevant parts, please refer to the description of the method, and it will not be repeated here.

Claims

1. A deep-hole geological monitoring method based on cloud-edge collaboration, characterized in that, The method includes: Acquire signals from multiple sensor sources inside deep holes; The multi-source sensor signals are transmitted to the private cloud edge based on the anti-interference transmission layer. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize the multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal. The private cloud edge converts the differential signal into a standardized digital signal stream, performs real-time preprocessing on the standardized digital signal stream, and then transmits it to the GPU cloud processing center based on bandwidth scheduling. The GPU cloud processing center performs parallel heterogeneous operations on the preprocessed standardized digital signal stream and generates a spatiotemporal aligned tensor. It then generates a geological monitoring task corresponding to the spatiotemporal aligned tensor for each preset time period, and the geological monitoring task is submitted to the virtual machine for execution. Calculate the success probability of the virtual machine executing the geological monitoring task, and determine whether the success probability is greater than the probability threshold. If it is, the virtual machine executes the corresponding geological monitoring task. If not, calculate the maximum allowable time when the success probability is equal to the probability threshold, and trigger the active restart protection process when the maximum allowable time is reached.

2. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 1, characterized in that, The method for calculating the success probability includes: Calculate the probability of success based on the time required to execute the geological monitoring task, the failure rate of the virtual machine, and the failure rate of the server hardware: ； in, Indicates the probability of success. This indicates the failure rate of the virtual machine. This indicates the failure rate of the server hardware. This indicates the time required to perform a geological monitoring task. Represents the natural constant.

3. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 2, characterized in that, The formula for calculating the maximum allowable time is as follows: ； in, Indicates the maximum allowed time. This represents the probability threshold.

4. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 1, characterized in that, The proactive restart protection process includes: Freeze the process and capture a runtime snapshot, generate snapshot data, compress and encrypt the snapshot data, and then store it in a cross-node disaster recovery storage pool. The corresponding geological monitoring task is scheduled to the healthy node. The healthy node obtains snapshot data from the cross-node disaster recovery storage pool, decrypts and decompresses the snapshot data, loads it into the new environment, and restores the process state.

5. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 1, characterized in that, The anti-interference cable comprises, from the inside out, a shielding layer, a silver-plated copper wire braided mesh, and a nano-magnetic wave-absorbing layer.

6. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 1, characterized in that, The enhanced differential driver includes a signal input interface, a preprocessing module, a level converter, a drive enhancement circuit, a differential signal generation module, and a signal output interface; The signal input interface is used to receive signals from multiple sensor sources; The preprocessing module is used to perform pre-emphasis adaptive processing and edge-jump compensation processing on multi-source sensor signals to generate preprocessed signals. The level converter is used to boost the preprocessed signal to generate a high-voltage signal; The drive enhancement circuit is used to amplify the power of the high-voltage signal to generate an enhanced high-voltage signal. The differential signal generation module is used to sample the enhanced high voltage signal in real time using a fully differential operational amplifier, extract the common-mode voltage, and generate a differential signal with a high common-mode rejection ratio. The signal output interface is used to output differential signals.

7. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 1, characterized in that, The multi-source sensor signals include sensor signals corresponding to seismic wave data, gamma spectrum data, and infrared video stream data.

8. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 7, characterized in that, The GPU cloud processing center performs parallel heterogeneous computations on the preprocessed standardized digital signal stream, including: For the preprocessed standardized digital signal stream corresponding to the seismic wave data, the GPU cloud processing center starts the CUDA kernel function to perform db4 wavelet basis noise reduction. For the preprocessed standardized digital signal stream corresponding to the γ energy spectrum data, the GPU cloud processing center initiates parallel Savitzky-Golay peak correction. For the preprocessed standardized digital signal stream corresponding to the infrared video stream data, the GPU cloud processing center activates the NVDEC hardware decoder to extract the heat matrix.

9. The deep borehole geological monitoring method based on cloud-edge collaboration according to claim 8, characterized in that, Generate a spatiotemporally aligned tensor, including: The results of parallel heterogeneous computation are spatiotemporally aligned based on the DTW algorithm to generate a spatiotemporally aligned tensor. During the computation of the DTW algorithm, the computation of time-warped paths is decomposed into GPU thread blocks, and the shared memory access mode is optimized.

10. A deep-hole geological monitoring system based on cloud-edge collaboration, characterized in that, For implementing the deep borehole geological monitoring method based on cloud-edge collaboration as described in any one of claims 1 to 9, the system comprises: Multi-source sensor, used to collect multi-source sensor signals inside deep holes; An anti-interference transmission layer is used to transmit the multi-source sensor signals to the private cloud edge. The anti-interference transmission layer includes an anti-interference cable and an enhanced differential driver. The enhanced differential driver is used to standardize the multi-source sensor signals, suppress common-mode interference, and drive enhancement processing to generate an enhanced anti-interference differential signal. The private cloud edge is used to convert the differential signal into a standardized digital signal stream. After real-time preprocessing of the standardized digital signal stream, it is transmitted to the GPU cloud processing center based on bandwidth scheduling. The GPU cloud processing center is used to perform parallel heterogeneous operations on the preprocessed standardized digital signal stream and generate a spatiotemporal aligned tensor. It generates a geological monitoring task corresponding to the spatiotemporal aligned tensor in each preset time period, and the geological monitoring task is submitted to the virtual machine for execution. The fault handling model is used to calculate the success probability of the virtual machine executing a geological monitoring task, determine whether the success probability is greater than a probability threshold, if so, the virtual machine executes the corresponding geological monitoring task; if not, it calculates the maximum allowable time when the success probability is equal to the probability threshold, and triggers an active restart protection process when the maximum allowable time is reached.

Images