Underground fault mode identification method and system based on voltage sampling and holding

Abstract

The invention discloses an underground fault mode recognition method and system based on voltage sampling and holding, and relates to the field of oil and gas well underground measurement and control. The method comprises the steps that an adjustable sampling and holding circuit with a measurement channel and a reference compensation channel is introduced to an underground tool side, synchronously acquiring a measurement holding voltage sequence and a reference holding voltage sequence used for representing environment electric leakage / temperature drift; carrying out high-temperature reference drift adaptive compensation and point-by-point correction on the measurement sequence based on the reference sequence to obtain a more stable correction voltage sequence; and further constructing a sample holding feature vector from the correction sequence in a multi-time scale segmentation statistical mode, and realizing state category judgment through distance measurement matching with a preset fault mode template library. Therefore, reliable identification of various types of underground fault modes is realized under the conditions of high temperature, strong noise and limited bandwidth.

Description

Technical Field

[0001] This application relates to the field of downhole measurement and control technology for oil and gas wells, and in particular to a downhole fault mode identification method and system based on voltage sampling and holding. Background Technology

[0002] Downhole logging and measurement while drilling tools typically operate under extreme conditions such as high temperature, high pressure, strong vibration, and corrosive media for extended periods. Issues such as aging of electronic components, fatigue of welded structures, and sensor drift can significantly increase the probability of failure. Once downhole tools malfunction, it is often difficult to obtain sufficient information in a timely manner for location and handling, resulting in a high risk of downtime and disruption of operations.

[0003] Currently, the industry has gradually adopted the condition monitoring approach, using multi-source data such as current, voltage, vibration, and temperature to determine whether downhole tools are abnormal. However, under conditions of strong downhole noise, limited sampling period, and limited uplink transmission bandwidth, common threshold / rule-based judgments are difficult to distinguish between "normal fluctuations" and "fault symptoms," and the limitation of uploaded information will further increase the complexity of diagnosis.

[0004] In addition, sample-and-hold circuits are more often used in downhole measurements to "keep the measured value stable during switching or multiplexing" or to achieve rapid isolation of fault sections in conjunction with voltage thresholds. These solutions either focus on the continuity of readings or on the rapid detection of obvious abrupt changes. They usually lack the ability to identify atypical faults (such as vibration-induced fluctuations, slow drift, etc.) in a patterned manner, and are also difficult to cope with the distortion of holding voltage caused by transient events and high-temperature drift. Summary of the Invention

[0005] This application provides a downhole fault mode identification method, system, storage medium, computer program product, and electronic device based on voltage sampling and holding, which at least solves the problem in the current related technologies that it is difficult to reliably identify and classify atypical abnormal signs of downhole tools under extreme downhole operating conditions and limited uplink information.

[0006] In a first aspect, embodiments of this application provide a downhole fault mode identification method based on voltage sample-and-hold (SHB) circuitry. The method includes: sending sampling configuration parameters to a downhole tool terminal and receiving a measured hold voltage sequence and a reference hold voltage sequence uploaded by the downhole tool terminal; wherein the measured hold voltage sequence is obtained by discretizing an original analog signal via a measurement channel of a configured adjustable SHB circuit, and the reference hold voltage sequence is obtained by synchronously acquiring environmental leakage current characteristics via a reference compensation channel of the adjustable SHB circuit; performing a high-temperature reference drift compensation operation on the measured hold voltage sequence, and utilizing the reference hold voltage... The voltage drop caused by environmental temperature drift is calculated using a pressure sequence, and the measured holding voltage sequence is corrected point by point to generate a corrected voltage sequence. The corrected voltage sequence is then subjected to time-domain multi-scale segmentation, and statistical characteristic parameters are calculated for each segment to construct a sample holding feature vector containing variation patterns at different time scales. The distance metric between the sample holding feature vector and the corresponding feature representation of each fault template in a preset fault mode template library is calculated, and the target state category corresponding to the fault template with the smallest distance metric is determined as the target state category of the downhole tool end. The target state category is either a normal operating state or a preset fault mode type.

[0007] Secondly, embodiments of this application provide a downhole fault mode identification system based on voltage sample-and-hold (SHB) technology. The system includes: a sampling configuration and data receiving unit, configured to send sampling configuration parameters to a downhole tool and receive a measured hold voltage sequence and a reference hold voltage sequence uploaded by the downhole tool; wherein the measured hold voltage sequence is obtained by discretizing the original analog signal through the measurement channel of a configured adjustable SHB circuit, and the reference hold voltage sequence is obtained by synchronously acquiring environmental leakage current characteristics through the reference compensation channel of the adjustable SHB circuit; and a high-temperature reference drift compensation unit, configured to perform high-temperature reference drift compensation operations on the measured hold voltage sequence, utilizing the reference hold... The voltage sequence calculation calculates the voltage drop caused by environmental temperature drift and performs point-by-point correction on the measured and held voltage sequence to generate a corrected voltage sequence. A multi-scale segmented feature construction unit performs time-domain multi-scale segmentation processing on the corrected voltage sequence and calculates statistical feature parameters within each segment to construct a sample holding feature vector containing variation patterns at different time scales. A template matching and discrimination unit calculates the distance metric between the sample holding feature vector and the corresponding feature representation of each fault template in a preset fault mode template library, and determines the target state category corresponding to the fault template with the smallest distance metric as the target state category of the downhole tool end, wherein the target state category is either a normal operating state or a preset fault mode type.

[0008] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the voltage sample-and-hold-based downhole fault mode identification method of any embodiment of the present application.

[0009] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, characterized in that, when the program is executed by a processor, it implements the steps of the downhole fault mode identification method based on voltage sampling and holding according to any embodiment of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the voltage sample-and-hold-based downhole fault mode identification method of any embodiment of this application.

[0011] The downhole fault mode identification method and system based on voltage sampling and holding provided in this application can achieve at least the following technical effects: (1) By setting up a synchronous acquisition mechanism between the measurement channel and the reference compensation channel, and using the environmental leakage current and temperature drift influence represented by the reference holding voltage as a benchmark quantity, the drift of the holding voltage sequence is estimated and corrected. This can significantly reduce the systematic shift and distortion of the holding voltage caused by environmental changes under extreme temperature conditions, so that the voltage changes relied upon by the subsequent identification process reflect more the true state evolution of the measured signal itself rather than environmental disturbances. As a result, the input sequence for state discrimination has better comparability and consistency, which can reduce the risk of state confusion caused by the superposition of high temperature drift and improve the stability of monitoring results across time periods and operating conditions.

[0012] (2) After obtaining the corrected voltage sequence, the fluctuations, trends, and local abrupt changes of the voltage sequence at different time granularities are transformed into structured feature representations through segmentation and statistical characterization oriented towards time scale differences. Furthermore, distance metric matching is performed with the feature representations of pre-set pattern templates, which can compress complex time-domain changes into discriminative pattern distance relationships, thereby outputting clear state category conclusions. Thus, by fusing multi-scale features to construct a discriminative link matching the feature template, the identification process no longer relies on single-point thresholds or single time windows, but is based on overall morphological similarity for classification. This is more conducive to achieving stable differentiation and verifiable judgment of abnormal signs under information-limited conditions.

[0013] This technical solution uses a reference sequence as an environmental benchmark to achieve voltage drift correction, and on this basis, it uses multi-scale statistical features and template distance matching to complete state classification. This constructs a complete identification process that gradually transitions from environmental disturbance suppression to pattern-based discrimination output, supporting the reliable implementation of downhole tool state identification under complex working conditions. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating an example of a voltage sample-and-hold-based downhole fault mode identification method according to an embodiment of this application is shown. Figure 2 A schematic diagram of the hardware structure of a downhole fault mode identification system based on voltage sample-and-hold according to an embodiment of this application is shown; Figure 3 This paper presents a schematic diagram comparing voltage waveforms acquired by a sample-and-hold circuit under different operating conditions according to an embodiment of this application. Figure 4 This illustration shows a schematic diagram of two-dimensional feature distribution clustering calculated based on the sample holding voltage sequence in an embodiment of this application; Figure 5 A structural block diagram of an example of a voltage sample-and-hold-based downhole fault mode recognition system according to an embodiment of this application is shown. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] In current related technologies, some studies have proposed diagnostic strategies based on multi-technology fusion or hierarchical fault trees to achieve more precise fault location. These methods typically divide the logging system into multiple levels, using analog signals to replace the upstream module output to test the downstream module, or combining time-frequency analysis to locate the fault source. However, such solutions often rely on complex hardware replacement interfaces or multi-step simulation testing procedures, making the diagnostic process cumbersome. They also struggle to quickly identify dynamic fault modes during real-time tool operation and have low utilization of signal details such as voltage transients.

[0018] For circuit monitoring of downhole tool strings, the industry currently commonly employs voltage threshold-based detection and isolation technologies. These solutions involve installing detection circuits between tool strings to monitor whether the voltage of the power supply path exceeds a preset threshold, thereby identifying short-circuit faults and controlling switches for physical isolation. While this method effectively prevents the entire tool string from losing power, it essentially still relies on detecting "hard faults" such as sudden voltage spikes or drops. It lacks the ability to identify atypical faults such as voltage fluctuations caused by vibration or slow drift caused by cable aging, and it typically cannot extract richer fault characteristics from the sampled signals. Furthermore, while some early measurement systems used sample-and-hold circuits, these were primarily designed to maintain reading continuity during alternating measurements of multiple parameters (such as pressure and temperature), with fixed sample-and-hold times, and were not designed for the analysis of complex fault modes.

[0019] With the development of data-driven technologies, some scholars have attempted to introduce machine learning algorithms (such as support vector machines) into downhole monitoring, using historical data to train classifiers to determine the health status of equipment. Although such methods can complete the diagnosis without modeling physical mechanisms, they rely excessively on a large amount of labeled sample data, making it difficult to cover the complex and ever-changing unknown working conditions downhole. In addition, the purely data-driven classification process often lacks physical interpretability, and its robustness to long-term drift noise in downhole environments still needs improvement.

[0020] Of particular note is the significant challenge posed by the high-temperature environment downhole (typically 150°C to 200°C or even higher) to the stability of electronic devices. High temperatures exacerbate capacitor leakage in sample-and-hold circuits and cause device parameter drift. Although some wide-temperature-range peak-hold circuit designs have emerged in related technologies to reduce the impact of temperature drift by improving circuit structure (such as using current mirrors), these improvements mainly focus on the stable holding of a single amplitude (peak value). They have not fully considered how to use the signal sequence after holding for in-depth fault mode identification, nor do they have a dynamic compensation mechanism for signal distortion during the holding process.

[0021] It should be understood that the above description of the relevant technologies is intended only to help the public better understand the inventive spirit and motivation of this application, and is not intended to limit this application. Furthermore, the technical solutions described in the above-mentioned relevant technologies are not prior art, and may also be undisclosed technical solutions, such as those under research or in the laboratory stage.

[0022] Figure 1 A flowchart illustrating an example of a voltage sample-and-hold-based downhole fault mode identification method according to an embodiment of this application is shown.

[0023] Regarding the execution subject of the method in the embodiments of this application, it can be any controller or processor with computing or processing capabilities, such as a well logging data processing platform controller. It establishes a two-way communication link with the downhole tool end to realize parameter distribution and data feedback, and completes the processing and discrimination of the feedback data on the controller / processor (or by calling the corresponding program instructions), thereby outputting the target status category of the downhole tool end and / or triggering the corresponding alarm and handling logic.

[0024] In some examples, it may be integrated into an electronic device or terminal through software, hardware, or a combination of both, and the type of terminal or electronic device may be diverse.

[0025] like Figure 1 As shown, in step S110, sampling configuration parameters are sent to the downhole tool terminal, and the measured holding voltage sequence and reference holding voltage sequence uploaded by the downhole tool terminal are received.

[0026] Here, the measured holding voltage sequence is obtained by discretizing the original analog signal through the measurement channel of the configured adjustable sample-and-hold circuit, and the reference holding voltage sequence is obtained by synchronously acquiring the environmental leakage current characteristics through the reference compensation channel of the adjustable sample-and-hold circuit.

[0027] In some implementations, sampling configuration parameters are used to controllably configure the adjustable sample-and-hold circuit at the downhole tool end, enabling it to output an engineering-usable hold voltage sequence under conditions of strong downhole noise, significant temperature drift, and limited uplink bandwidth. For example, sampling configuration parameters may include sampling period or sampling frequency, sample-and-hold capacitance level or equivalent hold time constant, sampling aperture / follow-up time and hold time, time-division multiplexing timing of the measurement channel and reference compensation channel (e.g., alternating sampling or inserting reference samples at a preset ratio), A / D quantization resolution and front-end gain, as well as frame length for sequence packet uploading, timestamp strategy, and packet loss retransmission strategy. After receiving the configuration parameters, the downhole tool end can perform discretized acquisition of the original analog signal through following, holding, and quantization in the measurement channel to form a measurement hold voltage sequence; simultaneously, the reference compensation channel can synchronously acquire electrical characteristics related to environmental leakage current to form a reference hold voltage sequence. To ensure that the reference sequence accurately reflects the environmental impact at the same moment, the two channels can share the same sampling and holding core structure (e.g., similar holding capacitors, similar switching structures, and similar buffered output paths). Timing control is used to ensure that the reference sampling and measurement sampling are completed under the same temperature and power / leakage conditions, thereby improving the consistency of the reference sequence's representation of environmental drift terms.

[0028] Regarding data backhaul, the downhole tool can upload the measured holding voltage sequence and the reference holding voltage sequence in a unified frame structure. For example, each sampling point can be accompanied by a sampling time index or timestamp, channel identifier, and quantization code value, and the alignment relationship between the two sequences can be maintained within a frame (e.g., reported in pairs as "measurement point-reference point" or reported at a fixed insertion position). In this way, the ground end can directly establish a point-to-point correspondence after receiving the two sequences, avoiding compensation deviations caused by alignment errors, and can obtain an additional reference observation sequence to characterize the impact of environmental leakage current / temperature drift without significantly increasing the amount of uplink data.

[0029] With configurable sample-and-hold parameters and a dual-channel synchronous feedback mechanism, the holding voltage sequence not only carries the discretized information of the measured signal, but also introduces reference observations that can be used to characterize environmental drift.

[0030] Figure 2 A schematic diagram of the hardware structure of a downhole fault mode recognition system based on voltage sampling and holding according to an embodiment of this application is shown.

[0031] like Figure 2 As shown, the system mainly includes components for the power supply bus voltage ( The adjustable sample-and-hold unit 1 (power supply) 100 is for the sensor output signal. The adjustable sample-and-hold unit 2 (sensor) 200, analog multiplexer 300, analog-to-digital converter 400, and downhole controller 500 are included.

[0032] The adjustable sample-and-hold unit 1 (power supply) 100 has an internally designed parallel dual-channel structure: via MOS switches With holding capacitor Construct a measurement channel for data acquisition. And output holding voltage Simultaneously, through MOS switching With reference capacitor This forms a reference compensation channel for acquiring a preset reference voltage. And output compensation voltage .

[0033] In the adjustable sample-and-hold unit 2 (sensor side) 200, the sensor output signal is used as... As input, and form a holding voltage output. and compensation voltage output .

[0034] The outputs of both channels are impedance-matched via buffers before being connected to analog multiplexer 300. Analog multiplexer 300 is used to select between multiple hold / compensate voltages, and analog-to-digital converter 400 is used to convert the selected analog voltage into a digital sequence. .

[0035] Furthermore, the two adjustable sample-and-hold units each include components controlled by a sampling clock. , Controlled MOS switches (such as , and , ), holding capacitor (such as , ) and reference capacitor (e.g. , The corresponding holding voltage and compensation voltage are output through buffers (such as A1, AR1, A2, AR2). The downhole controller 500 (such as MCU / FPGA) includes an adjustable clock generator, a compensation and preprocessing unit, and a core processing unit, which are used to control the sampling timing, compensate and preprocess the digital sequence, and upload the processing results to the logging data processing platform controller via the uplink communication module 600.

[0036] In step S120, a high-temperature reference drift compensation operation is performed on the measurement holding voltage sequence. The voltage drop caused by ambient temperature drift is calculated using the reference holding voltage sequence, and the measurement holding voltage sequence is corrected point by point to generate a corrected voltage sequence.

[0037] Here, the core of high-temperature reference drift compensation lies in treating the reference holding voltage sequence as the observation channel of the "environmental drift term", using it to estimate the holding voltage drop caused by temperature drift and leakage current, and mapping this drop onto the measured holding voltage sequence for point-by-point correction.

[0038] In some implementations, the reference holding voltage sequence can be preprocessed to improve the stability of the drop estimation. For example, denoising and smoothing (such as moving average or median filtering to suppress occasional spikes) can be performed on the reference sequence, and invalid points can be removed based on timestamp alignment. Subsequently, a reference value of the reference sequence in a certain initial stabilization phase can be selected as the reference baseline (e.g., the mean of the first N reference samples can be used as the baseline), and the reference drop at any given time can be defined as the offset of the reference sample relative to the baseline, thereby obtaining a drop sequence that varies over time. Since the reference compensation channel and the measurement channel are consistent in electrical structure and timing, this drop sequence can characterize the systematic decay trend of the holding voltage under the same environmental conditions.

[0039] When applying drop values ​​to the measurement sequence, a point-by-point correction relationship can be established: for each measured holding voltage sample, compensation is performed based on the reference drop value aligned with it, so that the corrected voltage sequence is restored as close as possible to the "holding voltage after removing the effects of environmental drift".

[0040] Preferably, to avoid overcompensation in cases of severe transients or missing sampling, reasonable variation constraints can be set on the drop (e.g., limiting the maximum variation of the drop per unit time), or neighboring point interpolation can be used to maintain correction continuity when the reference point is missing. Through the above processing, the correction voltage sequence will more centrally reflect the changes in the measured signal itself, rather than being dominated by the systematic downward trend of the voltage under high temperature conditions.

[0041] In step S130, the time-domain multi-scale segmentation process is performed on the corrected voltage sequence, and the statistical characteristic parameters in each segment are calculated to construct a sample feature vector containing the variation law of different time scales.

[0042] Here, the purpose of multi-scale segmentation is to explicitly encode the variation of the correction voltage sequence at different time scales into a computable and comparable feature representation.

[0043] In some implementations, multiple segmentation scale parameters (e.g., short window, medium window, long window) can be pre-set, and the corrected voltage sequence can be divided into multiple time segments accordingly. The segmentation method can be non-overlapping segmentation or sliding segmentation. When using sliding segmentation, a step size can be set to balance time resolution and computational complexity. For each segment, statistical characteristic parameters describing the segment's shape can be calculated. These statistical characteristic parameters may include at least: segment mean and segment variance (used to characterize level position and fluctuation intensity), segment peak-to-peak value or range (used to characterize the range of abrupt changes), the mean / variance of the segment's first-order difference or the segment slope statistics (used to characterize the trend and jitter), and energy-related indicators within the segment (used to reflect the degree of fluctuation accumulation). By repeatedly calculating the above statistical characteristics on different scale segments, a multi-scale feature set that simultaneously covers short-term fluctuations, medium-term disturbances, and long-term trends can be formed.

[0044] In terms of feature vector organization, statistical features of each segment at each scale can be concatenated in a preset order to form sample feature vectors. Furthermore, normalization processing can be performed on the feature vectors to ensure comparability of different feature dimensions in distance calculations. Through multi-scale segmentation and statistical feature construction, the key morphological information of the corrected voltage sequence is stably encoded into a vectorized representation.

[0045] In step S140, the distance metric between the sample holding feature vector and the feature representation of each fault template in the preset fault mode template library is calculated, and the target state category corresponding to the fault template with the smallest distance metric is determined as the target state category of the downhole tool end. The target state category is either normal operation state or preset fault mode type.

[0046] In some implementations, a preset fault mode template library is used to store feature representations corresponding to different target state categories, including normal operation state and one or more preset fault mode types.

[0047] Regarding the construction of the template library, for each type of state, representative holding voltage sequence samples can be collected under controllable conditions, and sample holding feature vectors can be generated according to the same process as online. Subsequently, the feature representations of similar samples can be aggregated to form the feature representation of the fault template of that type (for example, the mean vector of the feature vectors of similar samples can be used as the template center, and the variance / covariance can be optionally saved to characterize the distribution range of the same type).

[0048] In this way, during online identification, a distance metric can be calculated between the current sample feature vector and the feature representations of each fault template in the template library. The smaller the distance, the more similar they are. The form of the distance metric can be selected according to implementation requirements (e.g., Euclidean distance or weighted distance). More specifically, the category corresponding to the fault template with the smallest distance metric can be determined as the target state category, and the corresponding identification result can be output for subsequent diagnosis, treatment, or recording and storage.

[0049] Preferably, the minimum distance value or the difference between the minimum distance and the second minimum distance can also be output as a quantitative basis for the strength of similarity to assist in subsequent manual review or thresholding strategies (e.g., to distinguish between "obvious match" and "general match"), but the final category is still based on the category corresponding to the minimum distance.

[0050] In this embodiment of the application, by using the feature representation of the template library and the distance metric matching mechanism, the state discrimination of the online acquisition sequence can be transformed into a computable and verifiable similarity judgment process, outputting a clear state category, which facilitates the formation of stable pattern recognition conclusions under the conditions of complex downhole working conditions and limited information.

[0051] Regarding the implementation details of step S110, in some examples of embodiments of this application, the sample-and-hold period of the adjustable sample-and-hold circuit is configured. It also generates a synchronous clock signal to control the switching action of the measurement channel and the reference compensation channel.

[0052] Combination Figure 2 The hardware structure shown has a system configuration for the power supply bus voltage ( The adjustable sample-and-hold unit 1 (power supply) 100 is for the sensor output signal. Adjustable sample-and-hold unit 2 (sensor) 200.

[0053] Taking the adjustable sample-and-hold unit 1 (power supply) 100 as an example, it internally constructs a physically parallel dual-loop structure: consisting of MOS switches With holding capacitor The series connection forms a "measurement channel" specifically designed to capture the signal under test; it is composed of MOS switches. With reference capacitor They are connected in series to form a "reference compensation channel" for capturing environmental benchmarks.

[0054] Specifically, the downhole controller 500 receives configuration instructions from the host computer or logging data processing platform controller to set the sampling and holding period. (For example, set to 20ms to capture fault characteristics in a specific frequency band). The adjustable clock generator inside the controller then generates a high-precision synchronization control signal. (For Unit 1) and (For Unit 2). Thus, through physical parallelism at the hardware level and strict synchronization at the clock level, complete alignment of the measurement signal and the reference signal on the time axis is ensured.

[0055] During the sampling phase The first analog switch of the closed measurement channel and the second analog switch of the reference compensation channel load the original analog signal onto the measurement holding capacitor. And apply the preset reference potential to the reference holding capacitor. .

[0056] Here, in the sampling phase Downhole controller drive control signal The signal transitions to a high level, simultaneously closing the first analog switch of the measurement channel. The second analog switch with the reference compensation channel At this time, the external original analog signal (such as...) Quickly measure the holding capacitance Charging causes its terminal voltage to follow the signal changes; simultaneously, a highly stable preset reference potential ( (Regarding the reference holding capacitor) Charge it.

[0057] During the holding phase, both the first analog switch and the second analog switch are simultaneously disconnected, utilizing the reference holding capacitor. Simulated measurement of holding capacitance Charge discharge characteristics under current downhole temperature conditions.

[0058] More specifically, during the hold phase, the controller drive toggles to a low level. Two switches and Simultaneous disconnection. In some cases, due to downhole ambient temperatures reaching 175 degrees Celsius or even higher, leakage current from the MOS switch and dielectric leakage current from the capacitor itself are unavoidable. The key to this embodiment is that... Selected as Devices with the same temperature coefficient and dielectric properties, and arranged adjacently in a consistent thermal environment, are thus utilized using a reference holding capacitor. The natural voltage decay during the holding period allows for accurate simulation and measurement of the holding capacitance. The charge discharge characteristics at the current temperature. Thus, the unpredictable "temperature drift noise" is transformed into an observable "reference decay signal", thereby physically separating the real signal fluctuations from the device's thermal drift.

[0059] At the end of the holding phase, the voltage values ​​across the holding capacitor are read and measured. The voltage across the reference holding capacitor And construct a measurement holding voltage sequence in chronological order. With reference holding voltage sequence .

[0060] At the end of the hold phase, the charge on the hold capacitor already includes signal information and leakage error. At this point, the hold voltage of the measurement channel... Compensation voltage with reference channel The outputs are respectively via high-impedance buffers (A1, AR1) to prevent the subsequent circuit from causing a load effect on the holding capacitor.

[0061] Under the control of the downhole controller, the analog multiplexer 300 sequentially sends these analog voltages to the analog-to-digital converter 400 for quantization. The preprocessing unit within the controller processes the measured and held voltage values ​​acquired at the same time. With reference holding voltage value Pairing and timestamping are performed, and the data is stored sequentially in chronological order to construct a pair of measured hold voltage sequences. With reference holding voltage sequence Thus, not only was the conversion from analog to digital domain completed, but more importantly, the complete original data containing temperature drift information was preserved. This allows subsequent algorithms to utilize the correlation between the two sequences to accurately reconstruct the true voltage waveform before distortion through differential or recursive models.

[0062] Regarding the implementation details of the high-temperature reference drift compensation operation in step S120, in some examples of the embodiments of this application, the cumulative drift compensation amount is initialized. The value is set to 0, and a leakage current proportionality coefficient is set to characterize the leakage current difference between the measurement channel and the reference compensation channel. .

[0063] Here, the high-temperature reference drift compensation uses the "holding period" as the update granularity. The well logging data processing platform controller recursively estimates the reference term of the measured holding voltage, thereby separating the slow voltage drop caused by high-temperature drift and leakage from the measurement sequence.

[0064] Specifically, before entering the compensation cycle, the fault identification terminal first needs to initialize parameters based on the current downhole environment and hardware characteristics. Specifically, the system initializes the cumulative drift compensation amount. Set it to 0; simultaneously, set the key adjustment parameter, leakage current ratio coefficient. It is used to characterize the mapping ratio of leakage current / temperature drift between the reference compensation channel and the measurement channel. It can be obtained by comparison sampling under the same temperature and pressure conditions during the experimental calibration stage, or a preset value range can be given according to different tool batches and different device temperature zones and kept fixed during operation.

[0065] Here, the leakage current proportionality factor It is a parameter with a clear physical meaning, expressing the slight, unavoidable differences in hardware manufacturing between the measurement channel and the reference compensation channel. Although the capacitance of the two channels ( and While the switching devices are designed to be compatible, slight deviations in leakage current characteristics may occur in actual manufacturing processes. (Coefficient) To correct this bias, the "pure ambient leakage" observed in the reference channel is accurately mapped to the scale of the measurement channel, ensuring the physical consistency of the compensation amount.

[0066] Then, regarding the current hold period Calculate the change in the reference holding voltage and, in conjunction with the leakage current proportional coefficient, calculate the cumulative drift compensation at the current moment using the following cumulative formula. : Equation (1) In the formula, For the first The cumulative drift compensation over each holding period. and They represent the first The and the first The reference holding voltage value for each holding cycle.

[0067] Here, the system uses the change in the reference holding voltage sequence to quantify the current leakage current level, and calculates the cumulative drift compensation at the current moment using equation (1) to achieve historical reconstruction. Wherein, the term... The voltage drop of the reference channel in the current cycle relative to the previous cycle was calculated. Since the reference channel input is at a constant potential, this drop is entirely caused by charge discharge (leakage) due to ambient high temperature. The system multiplies this single-step drop by a coefficient. Estimate the voltage increment lost due to leakage in the measurement channel within the same time interval, and add it to the total compensation at the previous moment. In the middle, a "reservoir" for leakage current is constructed through a recursive process, which records in real time how much voltage the signal has lost due to environmental factors since the start of sampling.

[0068] Then, the sum of the current measured holding voltage and the cumulative drift compensation is calculated to obtain the corrected voltage value, which is then obtained from the sequence. Constructing the correction voltage sequence: Equation (2) In the formula, For the first The measured holding voltage value for each holding cycle.

[0069] It should be noted that the voltage across the holding capacitor was measured due to leakage caused by high temperature. The drift gradually decreases over time (i.e., the measured value is lower than the true value). In this embodiment, the calculated cumulative drift compensation amount is used to compensate for the drift. Adding it back to the current measurement is equivalent to "virtually charging" the capacitor in the digital domain.

[0070] By employing a point-by-point addition correction mechanism, the system generates a correction voltage sequence. It can counteract the voltage drop trend that accumulates over time, and pull the voltage waveform that was originally "tilted at a large angle" or "collapsed" due to high temperature back to the normal horizontal baseline, thereby restoring the true amplitude and fluctuation characteristics of the original analog signal, and providing a high-fidelity, low-noise data foundation for subsequent fault feature extraction.

[0071] Regarding the implementation details of constructing the sample-and-hold feature vector in step S130 above, in some examples of embodiments of this application, the adjustable sample-and-hold circuit is configured with... Different maintenance periods The corrected voltage sequences corresponding to different time scales were obtained respectively.

[0072] It should be noted that, for the correction voltage sequence, a "multi-hold period - multi-time scale" approach is used to extract the sample hold feature vector that can stably characterize downhole fault symptoms.

[0073] Here, maintain the cycle. The physical meaning of "hold" refers to the duration for which the sample-and-hold unit enters the hold state after a single sampling. The longer the hold period, the more fully the leakage / temperature drift effect of the hold node accumulates, making it easier to amplify slow drift and charge discharge characteristics; the shorter the hold period, the more sensitive it is to rapid fluctuations introduced by transient disturbances or vibrations. By setting multiple hold periods and acquiring correction voltage sequences separately, the differentiated performance of different hold periods can be covered simultaneously at the feature level (e.g., covering transient capture at the millisecond level to trend monitoring at the second level).

[0074] Regarding the first Correction voltage sequence under holding period Calculate the average value of the inter-segment amplitude difference reflecting slow voltage drift or sudden changes. And the inter-segment energy variance reflecting voltage oscillation fluctuations. : Equation (3) Equation (4) In the formula, For the first Number of sampling holding segments under a holding period To correct the voltage sequence of the first Each hold segment represents a correction voltage value (e.g., the mean, median, or last value of the correction voltage samples within that hold segment). This represents the average value of the correction voltage for each holding segment.

[0075] In equation (3), by calculating the absolute value of the voltage difference between two adjacent holding segments and averaging it, the first-order differential statistical characteristics of the signal are essentially extracted, quantifying the "roughness" or rate of change of the voltage signal. The average level of the "variation amplitude between segments" is characterized by the absolute difference of the corrected voltage between adjacent sampling segments. When the downhole tool is under stronger drift, intermittent sudden changes, or vibration shock, the voltage difference between adjacent segments will increase as a whole, thus... The difference tends to be smoother under relatively stable operating conditions; conversely, it tends to be more gradual under relatively stable operating conditions.

[0076] In equation (4), firstly... The deviation of the relative mean from the "energy" is described, and the energy term is further strengthened and averaged to make it more sensitive to non-stationary characteristics such as "intermittent strong fluctuations, sharp disturbances, and concentrated bursts of energy", thereby distinguishing different oscillation intensities and fluctuation patterns.

[0077] Set voltage anomaly threshold The average length of the continuous holding intervals in the statistical sequence where the voltage amplitude change exceeds the voltage anomaly threshold is obtained. Transient duration characteristics of the holding period .

[0078] At the same time, set a voltage anomaly threshold. (This can be determined from calibration data or statistical quantiles), the average length of the continuous hold-out segments in the statistical sequence where the amplitude change exceeds the threshold is obtained. This feature is used to characterize the persistence of abnormal morphology. If the abnormality only manifests as a short spike, the continuous overthreshold segment is relatively short; if the abnormality manifests as clustered oscillations or a continuous drift phase, the continuous overthreshold segment is relatively long.

[0079] The characteristic parameters calculated under various holding periods are concatenated to generate the sample holding feature vector. .

[0080] Finally, the results obtained under each holding period , and By cascading and constructing sample feature vectors, the variation amplitude, fluctuation energy and anomaly persistence under multiple time scales are uniformly encoded into a structured representation. This can retain key discriminative information about fault symptoms even under conditions of limited uplink information, and improve the separability between different state modes.

[0081] Regarding the implementation details of determining the target state category based on the distance metric in step S140, after obtaining the sample feature vector... Then, a quantifiable similarity comparison is performed between it and various fault templates in a pre-set fault mode template library, and the target state category of the downhole tool end is output accordingly. In some examples of embodiments of this application, the fault mode template is obtained from the fault mode template library. Feature mean vector of fault modes Covariance matrix and weight matrix .

[0082] In some implementations... Used to characterize the "typical center" of this failure mode across various feature dimensions; Used to characterize the degree of dispersion and correlation of the failure mode in each feature dimension and between dimensions, so as to represent the natural fluctuation range of similar samples; It is usually a diagonal weight matrix, whose diagonal elements are used to reflect the importance of each feature dimension in distinguishing the fault mode.

[0083] It should be noted that the above template parameters can be generated based on the labeled sample set during the offline stage. For example, for the sample set belonging to the... The mean value of the historical sample feature vector set of the failure mode is obtained. And calculate the covariance according to the statistical definition to obtain Simultaneously, weights can be assigned to each dimension based on the information entropy or discriminant index of each dimension's features, forming... This allows feature dimensions with higher information content and greater sensitivity to classification to have a greater impact on distance calculation. As a result, the template library not only preserves the category center, but also explicitly introduces the range of fluctuations within the category and the importance of feature dimensions, which is beneficial for maintaining recognition stability when there are downhole noise disturbances and operating condition fluctuations.

[0084] The entropy-weighted Mahalanobis distance algorithm is used to calculate the feature vector of the sample to be tested. With the Distance metric between different failure modes .

[0085] Equation (5) In the formula, This represents the Hadamard product, which is the product operation of corresponding matrix elements; This is a moment vector transpose operation; It is a diagonal weight matrix, whose diagonal elements are derived from the first... The information entropy of each feature dimension of the fault-like mode is determined, and there is an inverse correlation between the information entropy of the feature dimension and the weight of the corresponding diagonal element.

[0086] In equation (5), This represents the deviation vector between the feature to be tested and the center of the template of this type; This is used to "normalize the shape of the distribution by category" of the bias, thereby avoiding the over-amplification of certain naturally volatile feature dimensions in the distance, and enabling the distance metric to take into account the correlation between different dimensions. Further weight modulation is applied to each feature dimension, emphasizing feature dimensions with lower information entropy (i.e., more stable and more discriminative) or stronger discriminative power, so that distance calculation is more focused on the feature subset that contributes more to recognition; This is the Hadamard product, used to implement the weight matrix and... Element-level fusion is used to introduce category-oriented feature importance adjustment while maintaining the Mahalanobis distance scale normalization capability.

[0087] Then, iterate through all categories in the fault mode template library and select those that make the following possible. The smallest category index is used as the recognition result.

[0088] Therefore, this distance metric not only compares "how far away from the template center", but also considers the fluctuation range of this type of fault template in the feature space and the importance of key feature dimensions. Thus, even when downhole high-temperature drift has been corrected but residual disturbances may still exist, it improves the robustness and verifiability of pattern matching, reduces the probability of misjudgment caused by abnormal fluctuations in individual feature dimensions, and enhances the ability to distinguish different fault mode types.

[0089] Regarding the diagonal weight matrix diagonal elements in The calculation details, in some examples of embodiments of this application, are based on the first Historical sample data of failure modes, the first The value range of each feature dimension is divided into: A preset interval is defined, and samples falling into the specified interval are counted. The probability of a preset interval .

[0090] Here, the diagonal weight matrix Its function is to: target the first For each type of failure mode, we select and strengthen the feature dimensions that are "more stable and more discriminative for this type of mode" from the various feature dimensions of the sample feature vector.

[0091] Therefore, based on the first Historical sample data for the first type of failure mode, targeting the The value range of each feature dimension is first discretized. Specifically, the value interval of this dimension in historical samples (e.g., determined by the minimum and maximum values, or using quantile ranges to suppress the influence of extreme values) is divided into... A preset interval is defined, and samples falling into the specified interval are counted. The probability of each interval .

[0092] In terms of statistical implementation, it can be Defined as "falling into the interval" Number of samples / number of the first "Total number of samples in each class", and can introduce a minimal smoothing term when there is a zero-count interval (e.g., add to the count for each interval). This is to avoid the instability of the values ​​of subsequent logarithmic terms.

[0093] Subsequently, the calculation of the first Class Fault Mode in the Information entropy in each feature dimension : Equation (6) In equation (6), the concept of thermodynamic entropy is used to quantify the "disorder" or "uncertainty" of the characteristic distribution. If a certain characteristic (e.g., voltage drop amplitude) always falls stably within a certain specific interval during a short-circuit fault, its probability distribution is concentrated, and the calculated entropy value... The entropy value will be very small; conversely, if a feature is affected by downhole noise and fluctuates violently and is distributed haphazardly, its entropy value will be very large.

[0094] According to information entropy Calculate normalized weights This allows feature dimensions with lower information entropy to receive higher weights: Equation (7) In the formula, This indicates the total number of preset intervals. Indicates the index of the preset interval; This represents the total number of feature dimensions in the sample feature vector. This represents the feature dimension index used to calculate the normalized denominator; This represents the sum of the complementary information entropy values ​​of all feature dimensions, used for weight normalization.

[0095] In particular, This indicates that it has been divided. The theoretical maximum information entropy value (i.e., the entropy value under uniform distribution) for discrete intervals. Therefore, the numerator term In physics, it represents the first The complementary information entropy value (or information redundancy) of each feature dimension, which must be non-negative, is used to quantify the degree to which the feature deviates from the random noise distribution. (Denominator) This represents the sum of the complementary information entropy values ​​of all feature dimensions, used to normalize the weights to the [0, 1] interval.

[0096] In equation (7), by introducing The system implements deterministic adaptive weight allocation for low-entropy features (i.e., ...). much smaller This indicates that the feature is concentrated and highly stable across failure modes. A larger value results in a higher weight; for high-entropy features (i.e., ... near This indicates that the characteristic distribution is close to uniform random noise and has high uncertainty. The value of approaches 0, thus obtaining extremely low weight or even being suppressed.

[0097] Through this mechanism, the algorithm can automatically filter out key feature dimensions that are "highly consistent across samples and exhibit significant physical laws" within each type of failure mode, while filtering out dimensions that are severely affected by downhole environmental noise, thereby significantly improving the robustness of the distance metric algorithm.

[0098] In some examples of embodiments of this application, after completing the distance measurement based on the template library and obtaining the identification result, a "unknown abnormal pattern" judgment logic is further introduced to avoid forcibly classifying samples that do not conform to any known template distribution into a certain known category.

[0099] Specifically, after determining the target state category at the downhole tool end, the calculated minimum distance metric value will be... Compared with the preset unknown fault determination threshold Compare them.

[0100] For example, one can start with the distance metrics for each category. Determine the minimum distance metric value and its corresponding minimum distance category index; then... Compared with the preset unknown fault determination threshold Comparison. When When, it indicates that the sample to be tested holds a feature vector. Consistent with the statistical distribution of at least one known fault template, the target status category at the downhole tool end can be output according to the minimum distance category.

[0101] Here, threshold The distance distribution of historical samples in the template library can be determined offline (for example, taking the highest quantile of the minimum distance between known category samples as the upper limit), thus providing a clear statistical basis for the determination of "unknown anomalies".

[0102] If the minimum distance metric value is greater than the unknown fault determination threshold If the current state does not belong to any known type in the preset fault mode template library, the target state category of the downhole tool end will be marked as an unknown abnormal mode.

[0103] Here, when When the current state deviates significantly from any known category in the template library, it is determined that the current state does not belong to any known type in the preset fault mode template library, and the target state category of the downhole tool end is marked as an unknown abnormal mode.

[0104] Furthermore, in response to unknown abnormal patterns, the current sample feature vector is automatically adjusted. The data is stored in the database to be analyzed, and after obtaining expert annotation information for the unknown abnormal mode, the fault mode template library is iteratively updated to enable online learning of new fault modes.

[0105] In some implementations, the current sample feature vector can be... Minimum distance metric The distance ranking information for each category (e.g., the top candidate categories and their distance values), along with the necessary context related to feature generation, are written into the database to be analyzed. For example, the context may include at least sampling configuration parameters (e.g., hold period set, sampling clock settings), drift compensation related parameters (e.g., ... , Includes baseline estimation status, downhole tool identification, sampling time index / timestamp, etc., to ensure that experts can reproduce the calculation chain of the sample during retrospective analysis.

[0106] After obtaining expert annotation information for the unknown anomaly pattern, the fault mode template library is iteratively updated according to the annotation results. If it is labeled as a new fault type, the feature vector set of that type of sample is used to generate new template parameters (such as new mean vector, covariance matrix, and weight matrix) and added to the template library; if it is labeled as a variant of an existing type or a misclassified sample, the sample can be merged into the sample set of the corresponding category to update its statistical template parameters. By leveraging the closed-loop mechanism of expert annotation information, the system can avoid the misclassification of unknown states without sacrificing the stability of known category recognition, and continuously expand the template library's coverage of new fault modes, thereby achieving a gradual enhancement of the online learning and recognition capabilities for new fault modes.

[0107] In some examples of embodiments of this application, after determining the target status category at the downhole tool end, a coordinated decision-making process oriented towards handling can be further initiated to transform the "status identification result" into executable control commands or operational suggestions. In some implementations, a "status category-handling strategy" mapping table can be pre-established, and the target status category can be used as a trigger condition for branching processing.

[0108] Specifically, when the target status category indicates a short-circuit or open-circuit fault, a power cut-off command or a fault segment isolation command is generated. For example, this type of fault may cause power overcurrent, abnormal module overheating, or signal link interruption. To prevent the fault from escalating, a power cut-off command or a fault segment isolation command can be generated. The power cut-off command may include fields such as the target power supply branch identifier, cut-off duration, and recovery conditions; the fault segment isolation command may include fields such as the isolated module number, isolation priority, and the bypass power supply or derating power supply strategy after isolation, and can be sent to the downhole tool end via the uplink communication link, whereby the downhole controller drives the corresponding switching devices or relay units to perform power-off / isolation operations. Through this strategy, protective actions can be quickly taken when the fault type is clear and the risk level is high, reducing the cascading impact of electrical abnormalities on other components of the downhole tool and improving the certainty and safety of fault handling.

[0109] Secondly, when the target state category indicates abnormal vibration, operational parameter adjustment suggestions are generated to prompt adjustments to downhole drilling pressure or pump speed. Such faults can be interpreted as state events where fluctuations in downhole operating conditions affect the stability of the measurement link or mechanical structure, and operational parameter adjustment suggestions are generated to prompt on-site personnel to adjust downhole drilling pressure or pump speed. For example, the suggestion content can be formed based on the intensity level and duration characteristics of the current vibration anomaly, as well as comparison information with historical operating conditions. For instance, suggestions could include "Recommend reducing drilling pressure to a preset range" or "Recommend adjusting pump speed to avoid the resonance range," and could include reasons for the suggestions (such as "Vibration anomaly duration exceeds threshold" or "Oscillation energy characteristics significantly increased") to enhance interpretability.

[0110] Thirdly, when the target state category indicates that the voltage signal is subject to high-frequency aliasing interference, the sampling configuration parameters, including the updated sample-and-hold period, are regenerated and sent to the downhole tool to change the sampling frequency to adapt to the current signal variation characteristics. This type of high-frequency aliasing interference fault often indicates that the current sample-and-hold period does not match the signal variation characteristics, resulting in an insufficient sampling frequency to cover the effective bandwidth or susceptibility to folding interference. Therefore, the sampling configuration parameters, including the updated sample-and-hold period, can be regenerated and sent to the downhole tool via the communication link to dynamically change the sampling frequency or sampling timing, making the sampling process more compatible with the current signal spectrum characteristics. The updated sampling configuration parameters may include at least a new set of hold periods, a sampling clock division coefficient, and a corresponding MUX / ADC sampling order adjustment strategy, thereby ensuring that the measured hold voltage sequence and the reference hold voltage sequence remain synchronized. Through this adaptive reconfiguration mechanism, the contamination of the hold voltage sequence by aliasing interference can be suppressed without replacing the hardware, allowing subsequent identification to recover to a more stable feature extraction and template matching state under new sampling conditions, improving the system's continuous monitoring capability and robustness in complex downhole electromagnetic interference environments.

[0111] To verify the effectiveness of the proposed method, this embodiment designs a numerical simulation experiment. Considering that downhole measured data is difficult to obtain publicly, the experiment uses generated synthetic voltage sequences to simulate four operating conditions: normal operation, short circuit, open circuit, and abnormal vibration. For each type of operating condition, a slowly varying voltage baseline with 5 V as the reference is first constructed, and corresponding fault characteristic terms and random noise terms are superimposed on the baseline. Among them, the short circuit / open circuit condition is characterized by introducing voltage abrupt changes or distortion segments at preset times, the abnormal vibration condition is characterized by superimposing oscillation components with preset frequency and amplitude, and the noise term is used to simulate downhole measurement disturbances and superimposed on the original sequence according to a preset intensity.

[0112] Subsequently, the original voltage sequence is input into the sampling and holding circuit model and processed: the sampling and holding period is set ( A number of sampling points (the corresponding time scale is determined by the sampling rate) are used to complete sampling and holding within each hold period. During the hold phase, a voltage drop / holding error effect consistent with the hold circuit is introduced to obtain the hold sequence. When simulating the effects of high-temperature environments, a reference hold sequence is simultaneously generated to characterize the hold voltage drift caused by the environment, and the hold sequence is corrected according to the aforementioned compensation strategy. Finally, the sample hold feature vector is calculated according to the feature extraction process defined above and used for subsequent template distance measurement and state recognition evaluation.

[0113] Figure 3 This paper presents a schematic diagram comparing voltage waveforms acquired by a sample-and-hold circuit under different operating conditions according to an embodiment of this application.

[0114] Reference Figure 3 Part (a) shows the voltage waveform under normal operating conditions. It can be seen that during normal system operation, the sampled and held voltage signal, referenced to 5V, exhibits a basically stable state with only minor ripple fluctuations, indicating that the power supply or sensor output is within a stable range.

[0115] Reference Figure 3 Part (b) shows the voltage waveform under short-circuit fault conditions. The voltage remains around 5V before t = 40 ms, and at t = 40 ms, the voltage signal experiences a significant step drop, and then remains at a lower voltage level (about 4.0V). This significant level drop corresponds to the short-circuit fault mode in the circuit.

[0116] Reference Figure 3 Part (c) shows the voltage waveform under open-circuit fault conditions. The waveform shows that, based on the reference voltage, there are randomly occurring high-amplitude spike pulses. These spikes not only have amplitudes that significantly exceed the normal range, but also appear at random times, reflecting the transient high-voltage interference characteristics caused by open circuit or poor contact.

[0117] Reference Figure 3 Part (d) shows the voltage waveform under vibration fault conditions. Unlike the aforementioned faults, this waveform exhibits continuous high-frequency oscillation characteristics, with the voltage amplitude fluctuating violently around the reference value. This indicates that the downhole tool was subjected to strong mechanical vibration or fluid impact, resulting in a sinusoidal high-frequency modulation phenomenon in the sensor or power supply voltage.

[0118] Figure 4 The diagram shows a two-dimensional feature distribution clustering diagram obtained based on the sample holding voltage sequence in an embodiment of this application. The four different colored scatter points in the diagram represent the distribution positions of four typical operating conditions in the feature space.

[0119] like Figure 4 As shown, different fault modes form well-defined clusters in the two-dimensional feature space, verifying the effectiveness of the extracted features. Specifically, the normal operating condition (blue dots) is located in the lower left region near the origin, characterized by extremely low inter-segment amplitude difference and energy variance, consistent with the characteristics of stable voltage. Short-circuit faults (red dots) are distributed in the upper left region, characterized by low inter-segment amplitude difference but high inter-segment energy variance, which is consistent with the characteristic of low voltage but accompanied by high-frequency noise fluctuations after a short circuit. Open-circuit faults (green dots) are distributed in the upper right region, characterized by both high amplitude difference and high energy variance, reflecting the random high-amplitude spike pulses accompanying an open circuit. Vibration faults (purple dots) are distributed in the lower right region, exhibiting high inter-segment amplitude difference but relatively low energy variance.

[0120] Through the above Figure 4 The two-dimensional feature distribution relationship in the data can intuitively reflect the distinguishability of different fault modes in the feature space. Based on the separability of the aforementioned feature distribution, the entropy-weighted Mahalanobis distance algorithm can effectively identify various working conditions. Therefore, compared with traditional simple threshold detection algorithms that struggle to distinguish overlapping features, it has significantly superior fault mode identification capabilities.

[0121] This paper proposes a downhole fault mode identification method based on voltage sampling and holding. By introducing adjustable sampling and holding units into the power supply line and sensor output line of the downhole tool, a holding voltage sequence that can be used to characterize state changes is obtained. The method combines a reference compensation channel and an adaptive benchmark tracking mechanism to dynamically correct the effects of high temperature drift and leakage current. On this basis, multi-scale segmented statistical feature mining is performed on the corrected holding sequence to construct a sampling and holding feature vector with physical meaning and interpretability. Finally, entropy-weighted Mahalanobis distance and a fault mode template library are used to achieve real-time differentiation and identification of various typical fault modes such as short circuit, open circuit, and vibration.

[0122] The proposed solution balances statistical robustness and interpretability on the algorithm side, supports sensitive detection of unknown anomalies and iterative template expansion, and features a simple hardware structure on the engineering side that is easy to integrate with commonly used downhole instruments. It has low computational overhead and can be deployed in downhole embedded or above-ground platforms, thus possessing good scalability and promotional value. It can be used for downhole tool condition monitoring and fault identification, and has the potential to be migrated to similar extreme environment equipment health monitoring scenarios such as geothermal, mining and deep sea.

[0123] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of combined actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Secondly, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application. In the above embodiments, the descriptions of each embodiment have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0124] Figure 5 A structural block diagram of an example of a voltage sample-and-hold-based downhole fault mode recognition system according to an embodiment of this application is shown.

[0125] like Figure 5As shown, the downhole fault mode recognition system 5000 based on voltage sampling and holding includes a sampling configuration and data receiving unit 510, a high-temperature reference drift compensation unit 520, a multi-scale segmented feature construction unit 530, and a template matching discrimination unit 540.

[0126] The sampling configuration and data receiving unit 510 is used to send sampling configuration parameters to the downhole tool end and receive the measurement hold voltage sequence and reference hold voltage sequence uploaded by the downhole tool end; wherein, the measurement hold voltage sequence is obtained by discretizing the original analog signal through the measurement channel of the configured adjustable sample hold circuit, and the reference hold voltage sequence is obtained by synchronously acquiring the environmental leakage current characteristics through the reference compensation channel of the adjustable sample hold circuit.

[0127] The high-temperature reference drift compensation unit 520 is used to perform a high-temperature reference drift compensation operation for the measured holding voltage sequence, calculate the voltage drop caused by ambient temperature drift using the reference holding voltage sequence, and perform point-by-point correction on the measured holding voltage sequence to generate a corrected voltage sequence.

[0128] The multi-scale segmented feature construction unit 530 is used to perform time-domain multi-scale segmentation processing on the corrected voltage sequence and calculate the statistical feature parameters in each segment to construct a sample feature vector containing the variation law of different time scales.

[0129] The template matching and discrimination unit 540 is used to calculate the distance metric between the sample holding feature vector and the corresponding feature representation of each fault template in the preset fault mode template library, and to determine the target state category corresponding to the fault template with the smallest distance metric as the target state category of the downhole tool end, wherein the target state category is one of normal operation state or preset fault mode type.

[0130] In some embodiments, this application provides a non-volatile computer-readable storage medium storing one or more programs including execution instructions. The execution instructions can be read and executed by an electronic device (including but not limited to a computer, server, or network device) to perform the steps of any of the voltage sample-and-hold-based downhole fault mode recognition methods described above.

[0131] In some embodiments, this application also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of any of the above-described voltage sample-and-hold downhole fault mode identification methods.

[0132] In some embodiments, this application also provides an electronic device comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform steps of a voltage sample-and-hold-based downhole fault mode identification method.

[0133] The above-described product can perform the methods provided in the embodiments of this application, and has the corresponding functional modules and beneficial effects for performing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in the embodiments of this application.

[0134] The electronic devices in this application can exist in various forms, including but not limited to: mobile communication devices, ultra-mobile personal computer devices, portable entertainment devices, or other airborne electronic devices with data interaction functions.

[0135] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0136] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A voltage sample-and-hold based downhole fault mode recognition method, characterized by, The method comprises: issuing a sampling configuration parameter to a downhole tool end and receiving a measured hold voltage sequence and a reference hold voltage sequence uploaded by the downhole tool end; wherein the measured hold voltage sequence is obtained by discretely collecting an original analog signal via a measurement channel of a configured adjustable sample and hold circuit, and the reference hold voltage sequence is obtained by synchronously collecting an environmental leakage characteristic via a reference compensation channel of the adjustable sample and hold circuit; performing a high-temperature reference drift compensation operation on the measured hold voltage sequence, calculating a voltage drop amount generated by an environmental temperature drift using the reference hold voltage sequence, and point-by-point correcting the measured hold voltage sequence to generate a corrected voltage sequence; performing time-domain multi-scale segmentation processing on the corrected voltage sequence and calculating statistical feature parameters in each segment to construct a sample holding feature vector containing different time scale change rules; calculating distance metric values between the sample holding feature vector and corresponding feature representations of each fault template in a pre-set fault mode template library, and determining a target state category corresponding to a fault template with the smallest distance metric value as a target state category of the downhole tool end, wherein the target state category is one of a normal operating state or a pre-set fault mode type.

2. The method of claim 1, wherein, The method further comprises: Configuring a sample-and-hold period of the adjustable sample-and-hold circuit and generating a synchronization clock signal to control the switching actions of the measurement channel and the reference compensation channel; In the sampling phase , the first analog switch closing the measurement channel and the second analog switch of the reference compensation channel load the original analog signal to the measurement holding capacitor , and load the preset reference potential to the reference holding capacitor ; In the hold phase, while the first analog switch and the second analog switch are simultaneously disconnected, the reference hold capacitor is utilized analoging the measurement hold capacitor charge drain characteristics at the current downhole temperature environment at the end of said holding phase, reading the voltage values across said measurement holding capacitor and said reference holding capacitor and constructing a sequence of measurement holding voltages in chronological order and a sequence of reference holding voltages .

3. The method of claim 2, wherein, The method further comprises: initializing the accumulated offset compensation quantity to 0 and setting a leakage proportionality coefficient for characterizing a difference in leakage between the measurement channel and the reference compensation channel ; For the current holding period , the change amount of the reference holding voltage is calculated, and in combination with the leakage ratio coefficient, the accumulated drift compensation amount at the current time is calculated by the following accumulation formula : ， wherein is the accumulated drift compensation amount for the th holding period, and respectively represent the reference holding voltage values for the th and the th holding period. calculating the sum of the current measured hold voltage value and the accumulated drift compensation value, resulting in a corrected voltage value, and outputting the corrected voltage value as a sequence composing the sequence of correction voltages ， In the formula, is the measured holding voltage value for the first holding period.

4. The method of claim 1, wherein, The method further comprises: In the adjustable sample-and-hold circuit a different holding period , respectively, to obtain a correction voltage sequence corresponding to different time scales For the first correction voltage sequence under a holding period , the average of the amplitude difference between segments reflecting the slow drift or mutation of the voltage , and the energy variance between segments reflecting the oscillation fluctuation of the voltage : , , In the formula, is the number of sampling and holding segments under the first holding period, is the number of sampling and holding segments under the first holding period, is the segment representative correction voltage value of the first sampling and holding segment in the correction voltage sequence, is the segment representative correction voltage value of the first sampling and holding segment in the correction voltage sequence, is the average of the segment representative correction voltage values of the sampling and holding segments. Setting a voltage abnormal threshold , counting the average length of the continuous holding section whose voltage amplitude change exceeds the voltage abnormal threshold in the sequence to obtain a transient duration characteristic of the first holding period ;​ The feature parameters calculated under various holding periods are concatenated to generate a sample feature vector .

5. The method of claim 4, wherein, The method further comprises: obtaining a feature mean vector of a class of failure modes obtaining a feature mean vector of a class of failure modes obtaining a feature mean vector of a class of failure modes obtaining a feature mean vector of a class of failure modes obtaining a feature mean vector of a class of failure modes Adopt the weighted Mahalanobis distance algorithm based on entropy weight, calculate the feature vector of the sample to be measured With the first Distance measure value between failure modes : , wherein denotes the Hadamard product, i.e. the element-wise multiplication of the matrices; is the matrix vector transpose operation; is a diagonal weight matrix whose diagonal elements are given by the The information entropy of the failure mode class in each feature dimension is determined, and the information entropy of the feature dimension and the corresponding diagonal element weight are inversely related. traversing all categories in the failure mode template library, selecting the smallest category index as the identification result. as the identification result.

6. The method of claim 5, wherein, determining diagonal elements in a diagonal weight matrix The determination process of the diagonal elements in the diagonal weight matrix​ Based on the Historical sample data of failure modes, the first The value range of each feature dimension is divided into: A preset interval is defined, and samples falling into the specified interval are counted. The probability of a preset interval ; Computing the information entropy of the class failure mode in the dimension of the feature : ， According to the information entropy Computing the normalized weight So that the feature dimension with lower information entropy can obtain higher weight: ， In the formula, represents the total number of preset intervals, represents the index of the preset interval; represents the total number of feature dimensions in the sample feature vector, represents the feature dimension index used to calculate the normalization denominator; represents the sum of the information entropy complementary values of all feature dimensions, used for weight normalization.

7. The method of claim 5, wherein, The method further comprises: The calculated minimum distance metric value is compared with a preset unknown fault determination threshold value is performed. If the minimum distance metric value is greater than the unknown fault decision threshold then determine that the current state does not belong to any known type in the pre-set fault mode template library, and mark the target state category of the downhole tool end as an unknown abnormal mode. in response to the unknown abnormal pattern, automatically storing the current sample feature vector to a database to be analyzed, and after obtaining expert annotation information for the unknown abnormal pattern, iteratively updating the fault pattern template library to realize online learning of new fault patterns.

8. The method of claim 1, wherein, After determining the target state category of the downhole tool end, the method further comprises: After determining the target state category of the downhole tool end, the method further comprises: when the target state category indicates a short circuit or open circuit fault, generating a power cut-off instruction or a fault segment isolation instruction; when the target state category indicates a vibration anomaly, generating an operation parameter adjustment suggestion to prompt adjustment of downhole drilling pressure or pump speed; 9. A voltage sample-and-hold based downhole fault mode recognition system, characterized by, when the target state category indicates that the voltage signal has high-frequency aliasing interference, re-generating a sampling configuration parameter containing an updated sample and hold period and issuing it to the downhole tool end to change the sampling frequency to adapt to the current signal change characteristics. The system comprises: The sampling configuration and data receiving unit is configured to send sampling configuration parameters to the downhole tool end and receive a measured hold voltage sequence and a reference hold voltage sequence uploaded by the downhole tool end; wherein the measured hold voltage sequence is obtained by discretely collecting original analog signals via a measurement channel of the configured adjustable sampling and holding circuit, and the reference hold voltage sequence is obtained by synchronously collecting environmental leakage characteristics via a reference compensation channel of the adjustable sampling and holding circuit; The high-temperature reference drift compensation unit is configured to perform a high-temperature reference drift compensation operation on the measured hold voltage sequence, calculate a voltage drop amount generated by environmental temperature drift using the reference hold voltage sequence, and perform point-by-point correction on the measured hold voltage sequence to generate a corrected voltage sequence; The multi-scale segmented feature construction unit is configured to perform time-domain multi-scale segmentation processing on the corrected voltage sequence and calculate statistical feature parameters in each segment to construct a sample holding feature vector containing different time scale change rules; The template matching discrimination unit is configured to calculate distance measurement values between the sample holding feature vector and corresponding feature representations of each fault template in a pre-set fault mode template library, and determine a target state category corresponding to a fault template with the smallest distance measurement value as a target state category of the downhole tool end, wherein the target state category is one of a normal operating state or a pre-set fault mode type.

Images