Ground transient electromagnetic exploration device

Abstract

The invention provides a ground transient electromagnetic exploration device, and relates to the technical field of geophysical exploration, and the device comprises a placement box, the front side of the placement box is provided with a storage groove, the interior of the storage groove is provided with a transient electromagnetic instrument body, the front side of the placement box is provided with an upper cover used for covering the storage groove, and the bottom of the placement box is provided with a moving mechanism. A monitoring assembly is arranged on the transient electromagnetic instrument body, the monitoring assembly is connected with a control module, the monitoring assembly is used for collecting three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data of the transient electromagnetic instrument body, and the control module is used for performing diagnosis according to related data. According to the invention, field operation convenience and equipment safety are improved through a portable moving mechanism and a multi-buffer system, collision risk intelligent identification and trend prediction are realized in combination with an improved LSTM algorithm, the equipment damage risk is effectively reduced, the exploration data quality is guaranteed, and the exploration efficiency and reliability under complex terrains are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of geophysical exploration technology, and in particular to a ground transient electromagnetic exploration device. Background Technology

[0002] Transient electromagnetic methods, also known as time-domain electromagnetic methods, utilize ungrounded or grounded sources to emit a primary pulse magnetic field into the ground. During the intervals between these pulses, the secondary eddy current field is observed using coils or grounded electrodes. Traditional transient electromagnetic methods are generally considered to work best when probing geological formations using central and overlapping loops, maximizing the reception of formation information under stable source conditions. However, existing ground-based transient electromagnetic exploration devices face significant challenges in practical applications: their transmitting and receiving units are typically large, making them inconvenient for personnel to carry in complex terrains such as mountains and hills, severely impacting exploration efficiency. More importantly, these precision devices contain numerous sensitive electronic components and precision coils, making them susceptible to damage from impacts during transport. Existing protective measures primarily rely on passive protection methods such as reinforced casings, which cannot fundamentally address the problems of internal component damage, loose connections, or calibration parameter drift caused by severe vibrations, collisions, or drops. In terms of data quality assurance, traditional methods typically only detect equipment anomalies after operation through calibration, by which time a large amount of invalid data may have already been collected, resulting in a serious waste of exploration resources. Although recent research has attempted to apply machine learning algorithms to equipment status monitoring, most are limited to simple threshold alarms or classification judgments based on instantaneous data, failing to fully utilize the temporal characteristics of multi-source sensor data and lacking the ability to predict risk evolution trends. Especially in the field of transient electromagnetic exploration, existing technologies rely entirely on the operator's experience to judge equipment status, unable to accurately assess the degree of damage when a collision occurs, nor predict potential risk escalation, leading to a serious lag in maintenance decisions. This technological deficiency not only increases equipment maintenance costs but may also lead to the failure of the entire exploration project due to data distortion, urgently requiring an innovative solution that integrates portable design, physical protection, intelligent monitoring, and proactive diagnostics. Therefore, designing a ground-based transient electromagnetic exploration device is essential. Summary of the Invention

[0003] In order to overcome the shortcomings of the prior art, the purpose of this invention is to provide a ground transient electromagnetic exploration device.

[0004] To achieve the above objectives, the present invention provides the following solution: This invention provides a ground transient electromagnetic exploration device, comprising: a placement box, a top cover, a moving mechanism, a buffer assembly, a monitoring assembly, and a control module. The placement box has a storage slot on its front side, and the transient electromagnetic instrument body is placed inside the storage slot. The top cover is located on the front side of the placement box to cover the storage slot. The moving mechanism is located at the bottom of the placement box. The monitoring assembly is mounted on the transient electromagnetic instrument body and is connected to the control module. The monitoring assembly is used to collect triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data of the transient electromagnetic instrument body. The control module is used to perform diagnostics based on the triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data.

[0005] Preferably, the moving mechanism includes a limiting shaft, a universal wheel, and a handle. The bottom front and rear sides of the placement box are provided with limiting grooves, the limiting shaft is disposed inside the limiting grooves, the universal wheel is disposed on the limiting shaft, and the handle is disposed on the top of the placement box.

[0006] Preferably, the buffer assembly includes a buffer plate and a buffer spring. The buffer plate is installed on the upper wall, lower wall, top wall, bottom wall and rear wall of the receiving groove through the buffer spring, and the transient electromagnetic instrument body is placed on the buffer plate.

[0007] Preferably, the monitoring component includes an inertial measurement unit, an impact pressure sensor, and a coil impedance detection circuit. The inertial measurement unit is installed inside the transient electromagnetic instrument body for real-time acquisition of the device's triaxial acceleration and triaxial angular velocity. The impact pressure sensor is installed between the buffer plate and the transient electromagnetic instrument body for acquiring impact pressure data. The coil impedance detection circuit is integrated into the transmitting and receiving circuits of the transient electromagnetic instrument body for real-time monitoring of the resistance and inductance values ​​of the transmitting and receiving coils.

[0008] Preferably, the control module is equipped with a diagnostic algorithm for the ground transient electromagnetic exploration device, which specifically includes the following steps: Step 1: Acquire triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data, and perform data fusion and feature extraction on them; Step 2: Perform collision recognition based on the extracted standard feature vectors combined with the collision recognition model; Step 3: Perform comprehensive device monitoring and diagnosis based on standard feature vectors and recognition results.

[0009] Preferably, in step 1, triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data are acquired, and data fusion and feature extraction are performed on them, specifically as follows: Acquire triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data; Bandpass filtering and downsampling preprocessing were performed on the collected triaxial acceleration data, triaxial angular velocity data, and impact pressure data; Capture the event data window and extract time-domain and frequency-domain features from it; All extracted features are combined and standardized to generate a standard feature vector.

[0010] Preferably, in step 2, collision recognition is performed based on the extracted standard feature vector combined with the collision recognition model, specifically as follows: A collision recognition model is constructed based on an improved LSTM network structure; The collision recognition model is trained based on a pre-set dataset; The extracted standard feature vectors are input into the trained collision recognition model to perform collision recognition and output the risk level probability distribution of the collision event. The final risk level is determined based on the probability distribution.

[0011] Preferably, in step 3, the device monitoring and comprehensive diagnosis is performed based on the standard feature vector and the recognition results, specifically as follows: Obtain the coil impedance data and calculate its relative change with respect to the coil impedance baseline data; The risk level, impact location information, and relative change are combined and input into a rule-based expert system. The expert system executes a predefined diagnostic rule base and outputs a comprehensive diagnostic report.

[0012] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: This invention provides a ground transient electromagnetic exploration device, comprising a placement box, a top cover, a moving mechanism, a buffer assembly, a monitoring assembly, and a control module. The placement box has a storage slot on its front side, inside which the transient electromagnetic instrument body is placed. The top cover covers the storage slot. The moving mechanism is located at the bottom of the placement box. The monitoring assembly is mounted on the transient electromagnetic instrument body and connected to the control module. The monitoring assembly collects triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data from the transient electromagnetic instrument body. The control module performs diagnostics based on the triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data. This invention effectively solves the technical challenges of portability and reliability in traditional ground transient electromagnetic exploration devices through innovative structural design and deep integration of intelligent algorithms.

[0013] At the physical structure level, the device employs a meticulously optimized movement mechanism and cushioning system. Through the ingenious combination of bottom casters and retractable handles, the originally bulky exploration device can be moved as easily as a suitcase, greatly improving its mobility in complex terrain and significantly reducing the labor intensity of the staff. The multi-layer cushioning system constructs a comprehensive impact absorption network through highly elastic buffer layers and distributed buffer springs, which can effectively disperse and attenuate impact energy when the device encounters collisions. This significantly improves the equipment's resistance to damage from a physical perspective and provides an effective solution to the problem of existing devices being easily damaged during transport.

[0014] In terms of intelligent monitoring, this invention constructs a multi-dimensional diagnostic system based on an improved LSTM network. Through a sensor array composed of an inertial measurement unit, an impact pressure sensor, and a coil impedance monitoring circuit, it captures changes in the mechanical state and electrical parameters of the equipment in real time. Particularly noteworthy is that the improved LSTM network, by incorporating cell state memory functionality into a forgetting gate mechanism, significantly enhances its ability to extract temporal features. This not only accurately identifies the severity level of the current impact event but also predicts the future evolution trend of the risk, achieving a fundamental shift from passive response to proactive early warning. This proactive diagnostic capability allows the system to issue warnings before substantial damage to the equipment occurs, providing a valuable time window for taking protective measures and effectively avoiding the transient electromagnetic instrument damage caused by impacts in traditional devices.

[0015] In practical applications, this technology significantly improves the reliability and lifespan of the device. Through precise condition assessment and trend prediction, it avoids data distortion caused by equipment performance degradation, ensuring the accuracy and reliability of stratigraphic information acquired under center loop and overlap loop configurations. Simultaneously, the system's intelligent diagnostic function provides precise maintenance guidance after minor equipment damage, and combined with adaptive compensation algorithms to maintain detection accuracy, greatly improving operational adaptability and task completion rates in harsh environments. This invention truly realizes the transformation and upgrading of ground transient electromagnetic exploration devices from traditional bulky and vulnerable to intelligent and reliable ones, providing a complete and efficient technical support system for field geophysical exploration. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of a ground transient electromagnetic exploration device provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a standard LSTM network structure. Figure 3 A schematic diagram of the improved LSTM network structure; Figure 4 This is a diagnostic diagram of a ground transient electromagnetic exploration device provided in an embodiment of the present invention.

[0018] Reference numerals in the attached diagram: 1. Placement box; 2. Transient electromagnetic instrument body; 3. Limiting shaft; 4. Caster wheel; 5. Buffer spring; 6. Buffer plate. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The purpose of this invention is to provide a ground transient electromagnetic exploration device that improves the convenience of field operations and equipment safety through a portable mobile mechanism and a multi-buffer system. Combined with an improved LSTM algorithm, it realizes intelligent identification and trend prediction of collision risks, effectively reduces the risk of equipment damage, ensures the quality of exploration data, and significantly improves the exploration efficiency and reliability in complex terrain.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown, the present invention provides a ground transient electromagnetic exploration device, comprising: a placement box 1, a top cover, a moving mechanism, a buffer assembly, a monitoring assembly, and a control module. The placement box 1 has a storage slot on its front side, and a transient electromagnetic instrument body 2 is placed inside the storage slot. The top cover is located on the front side of the placement box 1 to cover the storage slot. The moving mechanism is located at the bottom of the placement box 1. The monitoring assembly is located on the transient electromagnetic instrument body 2 and is connected to the control module. The monitoring assembly is used to collect triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data of the transient electromagnetic instrument body 2. The control module is used to perform diagnostics based on the triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data.

[0023] The moving mechanism includes a limiting shaft 3, a universal wheel 4, and a handle. The bottom front and rear sides of the placement box 1 are provided with limiting grooves. The limiting shaft 3 is arranged inside the limiting grooves. The universal wheel 4 is arranged on the limiting shaft 3. The handle is arranged on the top of the placement box 1.

[0024] The buffer assembly includes a buffer plate 6 and a buffer spring 5. The buffer plate 6 is installed on the upper wall, lower wall, top wall, bottom wall and rear wall of the receiving groove through the buffer spring 5, and the transient electromagnetic instrument body 2 is placed on the buffer plate 6.

[0025] The monitoring components include an inertial measurement unit, an impact pressure sensor, and a coil impedance detection circuit. The inertial measurement unit is installed inside the transient electromagnetic instrument body 2 for real-time acquisition of the device's triaxial acceleration and triaxial angular velocity. The impact pressure sensor is installed between the buffer plate 6 and the transient electromagnetic instrument body 2 for acquiring impact pressure data. The coil impedance detection circuit is integrated into the transmitting and receiving circuits of the transient electromagnetic instrument body 2 for real-time monitoring of the resistance and inductance values ​​of the transmitting and receiving coils.

[0026] The outer shell of the placement box 1 and the top cover is made of engineering plastic and an internal metal frame, and rubber anti-collision strips are wrapped around key parts (such as corners) on the outer surface.

[0027] The buffer plate 6 is a highly elastic buffer layer.

[0028] like Figure 4 As shown, the control module contains a diagnostic algorithm for ground transient electromagnetic exploration devices, which specifically includes the following steps: Step 1: Acquire triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data, and perform data fusion and feature extraction on them; Step 2: Perform collision recognition based on the extracted standard feature vectors combined with the collision recognition model; Step 3: Perform comprehensive device monitoring and diagnosis based on standard feature vectors and recognition results.

[0029] In step 1, triaxial acceleration data, triaxial angular velocity data, impact pressure data, and coil impedance baseline data are acquired, and then data fusion and feature extraction are performed on them, specifically as follows: 1. Synchronously acquire the following four types of raw data streams at a fixed sampling frequency (e.g., 1 kHz): (1) Triaxial acceleration data A raw =[a x (t), a y (t), a z(t): It is collected by an inertial measurement unit, and the unit is usually gravitational acceleration g. It is used to directly measure the linear impacts that the device receives in various directions. (2) Triaxial angular velocity data G raw =[ω x (t), ω y (t), ω z (t) is collected by an inertial measurement unit, with units of degrees / second or radians / second, and is used to measure the rotation and torsion of the measuring device. This is crucial for identifying non-head-on collisions (such as the tendency to flip over due to corner impacts). (3) Impact pressure data P raw =[p1(t),p2(t),...,p n [(t)]: n impact pressure sensors distributed between the buffer plate and the transient electromagnetic instrument body, in pressure units (e.g., Pa), used to locate the impact point and quantify the direct contact force; (4) Baseline data of coil impedance Z base =[R t x, L t x, R r x, L r [x]: From the coil impedance monitoring circuit, this is the resistance and inductance values ​​of the transmitting and receiving coils measured during system startup self-test or the last stable state, which will serve as the reference for subsequent comparisons.

[0030] 2. Data Preprocessing and Filtering The raw data contains high-frequency noise, thermal noise from the sensor itself, and low-frequency vibrations generated by the normal movement of the device, which must be filtered out.

[0031] (1) Bandpass filtering: For A raw and G raw Apply a bandpass filter (such as a Butterworth filter) in the range of 0.5Hz to 200Hz. High-pass filter (0.5Hz): Used to remove DC components and extremely low-frequency signals caused by temperature drift or slow device tilt; Low-pass section (200Hz): Since impacts are usually instantaneous events, their energy is mainly concentrated in the low and mid frequencies. The 200Hz cutoff frequency can effectively preserve the impact characteristics while filtering out useless high-frequency noise. For P raw Similarly, a bandpass filter is used to eliminate steady-state pressure (such as the pressure difference between the inside and outside of the chamber) and high-frequency noise; (2) Downsampling: After filtering, the data can be downsampled from 1kHz to 500Hz without losing the characteristics of the impact event, so as to reduce the amount of data and computational load in subsequent calculations. After this step, we obtain the preprocessed data: A filt G filt and P filt ; 3. Event Triggering and Data Slicing To avoid continuous high-load operations, the system employs an event-triggered mechanism; The trigger condition is: when the absolute value of the acceleration on any axis |a i (t) | or the reading p of any impact pressure sensor j (t) When the pressure exceeds a preset low threshold (e.g., 0.5g or the pressure of a slight impact), it is determined to be a potential event; Data slicing: Once triggered, the system will automatically extract 100 milliseconds before and 400 milliseconds after the trigger point, for a total of 500 milliseconds of data A. filt G filt P filt The data forms an "event data window," which retains data prior to the event to help capture the complete dynamic process of the event; 4. Temporal Feature Extraction For each sensor data sequence within the "Event Data Window", calculate the following time-domain characteristics: (1) For triaxial acceleration and angular velocity (6 sequences in total): Peak value: The largest absolute value in the sequence. accx =max(|a x (t)|), which is the most direct indicator for measuring impact intensity; Peak-to-peak value: The difference between the maximum and minimum values, reflecting the dynamic range of the signal; Root mean square (RMS): The root mean square value of a signal, representing the average energy level of the signal; Impact duration: The time span during which the signal amplitude exceeds 50% of its peak value; The integral of the signal (area under the curve): ∫|a(t)|dt, which can be approximated as the "impulse" of the impact, measuring the total change in momentum of the impact; Skewness and kurtosis: Skewness describes the asymmetry of signal distribution; a strong, positive impact will cause the acceleration signal distribution to be skewed in the positive direction. Kurtosis describes the steepness of signal distribution; a sharp, transient impact will produce a signal with high kurtosis. (2) Regarding impact pressure data: Maximum pressure value: The maximum value among all pressure sensors; Impact location identification: Identifies which sensor(s) responded first and to the greatest extent, which can pinpoint the impact point (e.g., "bottom left"). Pressure duration: Similar to acceleration duration; 5. Frequency Domain Feature Extraction The time-domain signal within the "Event Data Window" is converted to the frequency domain using a Fast Fourier Transform, and its spectral characteristics are analyzed. (1) For triaxial acceleration signals: (2) Dominant frequency: The frequency component with the highest energy in the spectrum. Different types of impacts (such as blunt impact vs. sharp impact) have different dominant frequencies. Spectral centroid: the center frequency of the spectral energy distribution, FC = ∑(f i P i ) / ∑P i , where P i It is frequency f i The power spectral density at that point, and the impact with more high-frequency components, has a higher spectral centroid. Spectrum bandwidth: measures the extent to which a spectrum can be extended; Band energy ratio: Divide the spectrum into several key frequency bands (e.g., 0-50Hz, 50-100Hz, 100-200Hz), calculate the proportion of energy in each frequency band to the total energy, which can finely characterize the distribution pattern of impact energy; 6. Construct feature vectors All the extracted features are combined into a one-dimensional, standardized feature vector. Feature Combination: Concatenate all time-domain and frequency-domain features of an event (e.g., extract a total of 50 features from 6 IMU sequences and a stress sequence) into a long feature vector F. raw ; Feature standardization: Using the mean and standard deviation of pre-stored training data, standardize the feature vector F. raw Z-score standardization is performed to obtain the final feature vector F. final for: F finali =(F rawi -mean i ) / std i (1) This step is crucial because it eliminates the influence of different feature units and numerical ranges, enabling deep learning models to converge quickly and stably.

[0032] In step 2, collision recognition is performed based on the extracted standard feature vectors combined with the collision recognition model, specifically as follows: 1. Model Input Input data: The final feature vector F from step 1 final ; Data format: A matrix of shape [T, N], where T is the time step (i.e., the window size for continuous observations, e.g., containing the 10 most recent data slices), and N is the feature vector F for each time step. final The dimension; 2. Improved LSTM model architecture First, this invention introduces the LSTM network structure: like Figure 2 As shown, a standard LSTM neural network consists of a four-layer interactive structure, with each layer corresponding to one memory cell. The information transfer functions tanh and sigmoid are denoted as tanh and σ, respectively. For any time t, the input cell is x. t The cell state is C t The output unit is h t The previous time step t-1 corresponds to the input x. t-1 Hidden state C t-1 Output h t-1 The cell structure design in LSTM networks incorporates multiple gating mechanisms such as input gates, forget gates, and output gates, enabling precise control over the flow of information.

[0033] Traditional LSTM networks suffer from significant shortcomings in their gating mechanism design, particularly in the failure to fully consider the crucial regulatory role of cell states in information filtering during hidden layer construction. This directly impacts the network's ability to model long-term dependencies. Specifically, the gating system in the LSTM architecture exhibits two main drawbacks: first, the independent input and forget gate mechanisms make it difficult for the network to effectively assess the importance of historical information; second, insufficient interaction between the gating unit and the cell state reduces the efficiency and accuracy of feature extraction. To address these issues, this invention proposes an improved LSTM network structure, such as... Figure 3 As shown, the new architecture enhances the utilization of historical information by reconstructing the gating mechanism, introducing dynamic feedback of cell state in the forget gate, establishing a collaborative working mechanism between the input gate and the forget gate, improving the accuracy of feature selection, optimizing the cell state update strategy, and improving the modeling ability of long-term dependencies. The improved LSTM network differs fundamentally from the standard LSTM network in its architectural design. The forget gate mechanism of the traditional LSTM network only integrates the current input x. t and the hidden state h from the previous moment t-1 As an input feature, this design has significant limitations in temporal modeling. To improve network performance, the forget gate was structurally optimized, innovatively incorporating the cell state C from the previous time step. t-1The theoretical basis for incorporating the input feature set into the forget gate is that, in an LSTM network, the cell state C... t-1 As the core carrier of memory units, it fully records the temporal feature evolution law during network training, through C t-1 By introducing a forget gate decision mechanism, the network can more accurately assess the importance of historical information, thereby significantly improving its ability to model time series data. In terms of gating system design, besides retaining the standard input gate pair x... t and h t-1 In addition to the basic processing, a collaborative working mechanism connecting the forget gate and the input gate was established.

[0034] In the transmission of time-series information, information retention and forgetting are essentially complementary. Based on this, the rule of "input gate = 1 - forget gate" was designed to ensure that the input of new information strictly follows the logical order of forgetting before updating. This design not only conforms to the natural laws of information processing, but also logically guarantees the consistency of gating decisions, thereby significantly improving the reliability and stability of the method in practical applications. The optimized network model is defined as follows: (2) (3) 3. Model Output Output Data 1: Probability distribution of current risk level: P current =[p0, p1, p2, p3], representing the probability of belonging to each risk level at the current moment; Output Data 2: Risk Trend Prediction: P trend =[p up p stable p down ],in: p up The probability that the risk level will rise in the near future (e.g., in the next 5 time steps); p stable The probability that the risk level will remain unchanged; p down The probability of a decrease in risk level; Final judgment: Current risk level = argmax(P) current ); Risk trend = argmax(P) trend ); This invention provides an example for illustration, wherein: Risk levels are categorized into four levels: no risk, low risk, medium risk, and high risk. Risk Trend: Rising: The model predicts that even though the current shock is mild, the sequence pattern suggests that a more severe impact may follow, or that the equipment stability is declining. Stable: The current level of shock is expected to persist for some time; Decline: The impact event is isolated or its intensity is weakening.

[0035] In step 3, a comprehensive diagnostic assessment of the device is performed based on the standard feature vector and the recognition results, specifically as follows: The core objective of step 3 is to integrate instantaneous status and trend prediction, combined with physical measurement data, to conduct a comprehensive and forward-looking assessment of the equipment's health status. This involves receiving dual predictions of the current risk level and risk trend from step 2, cross-validating them with real-time hardware monitoring data, and forming the final diagnostic decision. Specifically: 1. Diagnostic Input (1) Prediction results from step 2: Current risk level (Level) current ): Immediate assessment of collision and impact risks; Risk trend: The direction of change in risk level over a future period of time (rising, stabilizing, or falling). Probability vector: the confidence level of the above results; (2) Real-time hardware status data: Coil impedance data: Real-time measured coil resistance and inductance values ​​and their changes relative to a baseline (ΔR) tx ΔL tx Δ Rrx ΔL rx ); (3) Equipment historical health records: Recent impact incident records, historical performance baselines, and maintenance records.

[0036] 2. Dynamic diagnostic logic and rule base This invention employs an enhanced expert system, whose rule base has been expanded to include diagnostic logic related to risk trends. Detailed explanation of the diagnostic process: (1) Trend perception and early warning Rule T1 (Upward Trend Warning): IF Level current ==Slight AND Trend ==Upward; THEN added a warning to the diagnosis: "The risk trend is on the rise, and a more severe impact may be imminent. It is recommended to prepare countermeasures." Rule T2 (High-Risk Continuous Warning): IF Levelcurrent >= Moderate AND Trend == Stable; THEN added a "Warning: The equipment is in a high-risk environment with a high risk of cumulative damage" to the diagnosis.

[0037] (2) Cross-validation of hardware status and prediction results Rule R1 (Coil Deformation Risk - Trend Verification): IF Level current >= Moderate AND (ΔL) tx or ΔL rx (Absolute value > deformation threshold) AND Trend != decreasing; THEN The diagnosis is "high confidence risk of coil deformation, and the shock environment may continue; immediate examination is recommended."

[0038] The ELSE IF condition is the same as above, but Trend == decreases; THEN The diagnosis is "risk of coil deformation, but the impact environment is becoming milder, and timely examination is recommended."

[0039] Rule R2 (Circuit Failure Risk - Instantaneous Verification): IF Level current >= Mild AND (ΔR) tx or ΔR rx >Connection failure threshold); The diagnostic conclusion includes "risk of connection failure in the transmit / receive coil circuit." This rule relies on transient hard damage; the trend is only used as a reference for maintenance urgency. (3) Cumulative effect analysis (enhanced) Rule R6 (Association of Fatigue Damage with Trend): If recent minor shocks are frequent AND Trend is stable or rising THEN The risk level of the diagnostic conclusion is automatically upgraded by one level, and it is emphasized that "high-frequency impacts are accompanied by adverse trends, and the structural fatigue life is reduced at an accelerated pace."

[0040] 3. Diagnostic output A structured comprehensive diagnostic report is generated as the output of this step. This invention provides an embodiment of the comprehensive diagnostic report as follows: { Diagnostic conclusion: High risk of physical deformation of the coil, and the risk trend is increasing; Key risk component: [transmitting coil] Risk level: Moderate Risk trend: On the rise. Confidence level: High Specific basis: { Triggering event: Moderate bump, impact on the left side. Forecast trend: The intensity of future shocks may increase. Coil status: Transmitting coil inductance change +3.5% (exceeding threshold). Historical context: Recent frequency of mild shocks }, Recommended action: [ Immediately perform coil calibration and fixation checks. It is recommended to suspend operations and avoid the current high-risk environment. Key monitoring areas in the next 5 operating cycles ] }

[0041] This invention provides an embodiment of an intelligent diagnostic case of a device encountering continuous impacts during field exploration; The background is that a geological exploration team was conducting transient electromagnetic exploration in a mountainous area. Due to the complex terrain, the exploration equipment encountered multiple collisions during the towing process, including a relatively serious side impact. 1. Test conditions: Initial state of the device: The baseline data of the coil impedance is stable; Test environment: rugged mountainous terrain with gravel roads; Test duration: 8 hours of continuous operation; Monitoring parameters: triaxial acceleration, angular velocity, impact pressure, coil impedance.

[0042] 2. Implementation process: Phase 1: Mild, continuous impact (first 2 hours) The device was dragged on a gravel road, and the inertial measurement unit continuously collected small vibrations. The impact pressure sensor recorded multiple impact events ranging from 0.3 to 0.8 g. The system triggered a mild risk warning, but the coil impedance change was within the normal range (ΔL<1%). The diagnostic report indicates that the equipment has experienced a series of mild impacts, and it is recommended to pay attention to the cumulative effect.

[0043] Phase Two: Moderate Lateral Impact (3rd Hour) The device struck a rock on its left side, generating a peak acceleration of 4.2g. The pressure sensor reading on the left side reached 85 kPa; Improved LSTM model recognition features: Time domain features: peak value 4.2g, duration 120ms, large impact impulse; Frequency domain features: spectral center of gravity 65Hz, main frequency components concentrated in 50-100Hz; Risk level: medium risk (probability 92%); Risk trend: rising (probability 78%).

[0044] Phase 3: Real-time Diagnosis and Validation Coil impedance monitoring revealed a change in the transmitting coil inductance of ΔLtx = +3.8%; The system executes dynamic diagnostic logic: Trigger rule R1: moderate risk + inductance change exceeds threshold + upward trend; Trigger rule T1: risk trend is rising; combined with historical records: there have been 15 mild impacts in the previous 2 hours.

[0045] The final diagnostic report was as follows: { Diagnostic conclusion: High risk of physical deformation of the coil; the impact environment continues to deteriorate. Key risk components: [transmitting coil, left-side buffer structure] Risk level: Moderate Risk trend: On the rise. Confidence level: High Specific basis: { Triggering event: Moderate lateral impact, force applied to the left side. Forecast trend: The intensity of the impact may continue to increase. Coil status: Transmitting coil inductance change +3.8% (exceeding the deformation threshold of 2%). Historical context: The structure had accumulated fatigue due to 15 mild impacts in the early stages. }, Recommended operation: [ Stop work immediately and inspect the left side of the casing and the cushioning assembly. The transmitting coil was calibrated and tested. Change course to avoid further impact. The coil performance will be the focus of monitoring over the next three operating cycles; ] }

[0046] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0047] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A ground transient electromagnetic prospecting device, characterized by, Include: The placing box, the upper cover, the moving mechanism, the buffer assembly, the monitoring assembly and the control module, the front side of the placing box is provided with a receiving groove, the inside of the receiving groove is provided with a transient electromagnetic instrument body, the front side of the placing box is provided with the upper cover, for covering the receiving groove, the bottom of the placing box is provided with the moving mechanism, the transient electromagnetic instrument body is provided with the monitoring assembly, the monitoring assembly is connected with the control module, the monitoring assembly is used for collecting three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data of the transient electromagnetic instrument body, the control module is used for diagnosing according to three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data.

2. The apparatus of claim 1, wherein, The moving mechanism includes a limiting shaft, a universal wheel and a hand lever, the bottom of the placing box is provided with a limiting groove on the front and back sides, the limiting shaft is arranged in the limiting groove, the universal wheel is arranged on the limiting shaft, and the hand lever is arranged on the top of the placing box.

3. The apparatus of claim 2, wherein, The buffer assembly includes a buffer plate and a buffer spring, the buffer plate is arranged on the upper wall, the lower wall, the top wall, the bottom wall and the rear wall of the receiving groove through the buffer spring, and the transient electromagnetic instrument body is arranged on the buffer plate.

4. The apparatus of claim 3, wherein, The monitoring assembly includes an inertial measurement unit, an impact pressure sensor and a coil impedance detection circuit, the inertial measurement unit is arranged in the transient electromagnetic instrument body, used for collecting three-axis acceleration and three-axis angular velocity of the device in real time, the impact pressure sensor is arranged between the buffer plate and the transient electromagnetic instrument body, used for collecting impact pressure data, and the coil impedance detection circuit is integrated in the transmitting and receiving circuit of the transient electromagnetic instrument body, used for monitoring the resistance and inductance value of the transmitting coil and the receiving coil in real time.

5. The apparatus of claim 4, wherein, The control module is internally provided with a diagnosis algorithm of the ground transient electromagnetic exploration device, which specifically includes the following steps: Step 1: obtain three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data, and perform data fusion and feature extraction on them; Step 2: based on the extracted standard feature vector, combine the knock identification model to identify the knock; Step 3: based on the standard feature vector and the identification result, the device monitoring comprehensive diagnosis is carried out.

6. The apparatus of claim 5, wherein, In step 1, three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data are obtained, and data fusion and feature extraction are performed, specifically: Obtain three-axis acceleration data, three-axis angular velocity data, impact pressure data and coil impedance baseline data; The collected three-axis acceleration data, three-axis angular velocity data and impact pressure data are preprocessed by band-pass filtering and downsampling; Extract time domain features and frequency domain features from the event data window; Combine and standardize all the extracted features to generate a standard feature vector.

7. The apparatus of claim 6, wherein, In step 2, based on the extracted standard feature vector, combine the knock identification model to identify the knock, specifically: An improved LSTM network structure is used to construct a knock identification model; The knock identification model is trained based on the preset data set; The extracted standard feature vector is input into the trained collision recognition model to perform collision recognition, and a risk level probability distribution of a collision event is output; According to the probability distribution, the final risk level is determined.

8. The apparatus of claim 7, wherein, In step 3, based on the standard feature vector and the recognition result, device monitoring comprehensive diagnosis is performed, specifically: Obtain the coil impedance data and calculate the relative change amount thereof relative to the coil impedance baseline data; In combination with the risk level, the impact direction information and the relative change amount, input into the rule-based expert system, and the expert system outputs a comprehensive diagnosis report by executing a predefined diagnosis rule library.

Images