Well lid state recognition device and method based on inclination-vibration dual-mode fusion

Abstract

This invention relates to a manhole cover status identification device and method based on tilt angle-vibration dual-mode fusion; belonging to the field of intelligent monitoring, the device includes a dual-mode MEMS sensing unit that synchronously outputs tilt angle detection signals and vibration detection signals based on the displacement of the same mass block; a processing unit is used to fuse and calculate tilt angle features based on the tilt angle detection signals and extract vibration features based on the vibration detection signals; a fusion decision unit is used to perform time-series correlation analysis based on tilt angle features and vibration features, initially classify vibration modes through multi-feature threshold matching, and correct them after time-series characteristics and multi-feature cross-validation, and then perform linkage mapping between the tilt angle change mode and the corrected vibration mode to output the manhole cover status identification result. This invention achieves accurate identification and low-power monitoring of various events such as manhole cover tilting and settlement, illegal opening, and external force damage through source reuse and progressive fusion decision.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring and Internet of Things technology, specifically relating to a manhole cover status recognition device and method based on tilt angle-vibration dual-mode fusion, which is applicable to intelligent manhole cover status recognition in scenarios such as underground power transmission channels and municipal utility tunnels. Background Technology

[0002] Manhole covers in urban underground power transmission channels, municipal utility tunnels, and other infrastructure are often left in an open environment, facing multiple safety risks such as illegal opening, theft, subsidence, and external impacts. Real-time monitoring of manhole cover status is of great significance for ensuring the safety of underground pipelines, preventing pedestrian falls, and maintaining urban public safety.

[0003] Existing manhole cover monitoring technologies mainly rely on a single type of sensor, which has the following technical shortcomings in practical applications: First, the false alarm rate is high. There is a lot of background vibration interference in the urban environment, such as road traffic and construction vibration. A single vibration sensor is difficult to effectively distinguish between such normal environmental vibrations and real abnormal events (such as illegal knocking or external damage), resulting in frequent false alarms from the system.

[0004] Second, the ability to recognize complex scenes is insufficient. A single sensor can only acquire physical quantity information in a single dimension, making it difficult to distinguish between events with similar features but different natures. For example, both "normal vehicle running over" and "illegal knocking" manifest as vibration signals, and both "slow settlement and tilting" and "instant illegal opening" manifest as angle changes. A single sensor lacks sufficient feature dimensions for accurate identification.

[0005] Third, there is a lack of a multimodal fusion decision-making mechanism. Existing solutions do not perform correlation analysis between tilt angle changes and vibration events in the time dimension. Event judgment relies solely on simple triggering of a single threshold, lacking cross-validation of multidimensional evidence and logical constraints, resulting in low reliability of the judgment results.

[0006] Fourth, there is a significant conflict between power consumption and reliability. To reliably capture transient abnormal events, the system needs to maintain high-frequency sampling and continuous operation, which leads to a significant increase in power consumption. However, if a low-power intermittent monitoring mode is adopted, it is easy to miss short-duration sudden abnormal events, making it difficult to achieve long-term stable operation under battery power.

[0007] Therefore, there is an urgent need for a method and device for intelligent identification of manhole covers that can adapt to complex vibration environments, achieve high accuracy, low false alarms, and controllable power consumption. Summary of the Invention

[0008] Based on the above analysis, the present invention aims to disclose a manhole cover status recognition device and method based on tilt angle-vibration dual-mode fusion; the method integrates a dual-mode MEMS sensor in hardware and constructs a time-series correlated multi-feature fusion decision model in software to achieve accurate recognition of various events such as normal state of manhole cover, tilt, settlement, illegal opening, and external force damage.

[0009] This invention discloses a manhole cover status recognition device based on tilt angle-vibration dual-mode fusion, comprising: The dual-modal MEMS sensing unit synchronously outputs tilt angle detection signals and vibration detection signals based on the displacement of the same mass block. The tilt angle detection signal includes X and Y axis acceleration components parallel to the manhole cover plane and a first Z-axis equivalent tilt angle signal perpendicular to the manhole cover plane. The vibration detection signal includes a second Z-axis acceleration component perpendicular to the manhole cover plane. The first Z-axis equivalent tilt angle signal characterizes the quasi-static angular displacement of the mass block caused by the tilt of the manhole cover, and the second Z-axis acceleration component reflects the dynamic acceleration of the mass block caused by external force impact. The processing unit, connected to the dual-mode MEMS sensing unit, is used to fuse and calculate tilt features based on tilt detection signals and extract vibration features based on vibration detection signals. The fusion decision unit, connected to the processing unit, is used for time-series correlation analysis based on tilt angle features and vibration features. It initially classifies vibration modes through multi-feature threshold matching, and after correction by time-series characteristics and multi-feature cross-validation, it performs linkage mapping by combining tilt angle change mode and the corrected vibration mode to output the manhole cover status identification result.

[0010] Another aspect of the present invention discloses a manhole cover state recognition method based on tilt angle-vibration dual-mode fusion, which uses the manhole cover state recognition device based on tilt angle-vibration dual-mode fusion as described above to perform the following steps: The system synchronously acquires multi-axis detection signals of a single mass block relative to the manhole cover; the multi-axis detection signals include X and Y axis acceleration components parallel to the plane of the manhole cover, and a first Z-axis equivalent tilt angle signal and a second Z-axis acceleration component perpendicular to the plane of the manhole cover; wherein, the first Z-axis equivalent tilt angle signal is detected by the annular fixed electrode group in conjunction with the third moving electrode, and the second Z-axis acceleration component is detected by the planar fixed electrode group in conjunction with the third moving electrode; Dual-modal parallel processing; the tilt angle features are calculated by fusing the static components of the equivalent tilt angle signals of the X, Y axes and the first Z axis, and the vibration features are extracted based on the dynamic components of the acceleration components of the second Z axis; Fusion judgment; based on the temporal correlation analysis of tilt angle features and vibration features, the vibration modes are initially classified through multi-feature threshold matching. After correction by temporal characteristics and multi-feature cross-validation, the tilt angle change mode and the corrected vibration mode are linked and mapped to output the manhole cover status recognition result.

[0011] Compared with the prior art, the present invention has at least the following advantages: The present invention relates to a manhole cover status recognition device and method based on tilt angle-vibration dual-mode fusion. This method generates tilt angle detection signals and vibration detection signals synchronously based on the displacement of the same mass block, and achieves high-fidelity parallel processing of the dual-mode signals using static / dynamic component separation calculation. This eliminates data synchronization delay and redundant acquisition from heterogeneous sensors on a single sensitive structure. Furthermore, through a progressive fusion decision mechanism of "initial screening—time-sequence verification—linkage mapping," the tilt angle change mode and vibration mode are correlated and coupled in the time dimension. Utilizing the mutual constraints of a multi-dimensional evidence chain of position change and stress state, it effectively eliminates false alarms caused by environmental interference, significantly improving the accuracy and reliability of distinguishing complex events such as manhole cover tilting and settlement, illegal opening, normal rolling, and environmental interference, while also considering the system's low power consumption and miniaturization requirements. Attached Figure Description

[0012] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Figure 1 This is a schematic diagram showing the components and connections of the manhole cover status recognition device based on tilt angle-vibration dual-mode fusion in an embodiment of the present invention. Figure 2(a) is a top view of the dual-mode MEMS sensing unit in an embodiment of the present invention; Figure 2(b) is a side view cross-sectional view of the dual-mode MEMS sensing unit in an embodiment of the present invention. Figure 2(c) is a top view of the Z-axis electrode partitioning structure of the dual-mode MEMS sensing unit in an embodiment of the present invention; Figure 3 This is a flowchart of the manhole cover status recognition method in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the overall process of manhole cover status recognition in this embodiment of the invention. Detailed Implementation

[0013] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and, together with the embodiments of the present invention, serve to illustrate the principles of the present invention.

[0014] Example 1 One embodiment of the present invention discloses a manhole cover status recognition device based on tilt angle-vibration dual-mode fusion, such as... Figure 1 As shown, it includes: The dual-modal MEMS sensing unit synchronously outputs tilt angle detection signals and vibration detection signals based on the displacement of the same mass block. The tilt angle detection signal includes X and Y axis acceleration components parallel to the manhole cover plane and a first Z-axis equivalent tilt angle signal perpendicular to the manhole cover plane. The vibration detection signal includes a second Z-axis acceleration component perpendicular to the manhole cover plane. The first Z-axis equivalent tilt angle signal characterizes the quasi-static angular displacement of the mass block caused by the tilt of the manhole cover, and the second Z-axis acceleration component reflects the dynamic acceleration of the mass block caused by external force impact. The processing unit, connected to the dual-mode MEMS sensing unit, is used to fuse and calculate tilt features based on tilt detection signals and extract vibration features based on vibration detection signals. The fusion decision unit, connected to the processing unit, is used for time-series correlation analysis based on tilt angle features and vibration features. It initially classifies vibration modes through multi-feature threshold matching, and after correction by time-series characteristics and multi-feature cross-validation, it performs linkage mapping by combining tilt angle change mode and the corrected vibration mode to output the manhole cover status identification result.

[0015] It also includes a communication and power management unit, which supports low-power wireless communication NB-IoT and has a power control mechanism for deep sleep and event wake-up.

[0016] The power consumption control mechanism includes: a deep sleep mode, in which only the dual-mode MEMS sensing unit operates at a low sampling rate and is woken up by a vibration threshold interruption or a timer trigger; and an event wake-up mode, in which the processing unit and the fusion decision unit operate at full speed and return to the deep sleep mode after a set duration without any valid events.

[0017] It achieves ultra-low power consumption operation with a static current of less than 10μA, meeting the battery life requirements of more than five years for scenarios such as underground power transmission channels and municipal utility tunnels.

[0018] Specifically, the dual-mode MEMS sensing unit supports tilt angle measurement across a full range of 0°–90° and wideband vibration sensing from 0.1Hz to 500Hz, and features the ability to switch between high-resolution and low-power modes; it includes: a single mass block, an elastic suspension structure, and a capacitance detection structure; wherein, A single mass block is suspended by an elastic suspension structure, and the mass block has displacement degrees of freedom that are parallel to and perpendicular to the plane of the manhole cover; Define the two directions parallel to the plane of the manhole cover and orthogonal to each other as the X-axis and Y-axis, and the direction perpendicular to the plane of the manhole cover as the Z-axis; The capacitance detection structure includes a movable electrode group that moves synchronously with the mass block and a fixed electrode group that is fixed relative to the base. The movable electrode group and the fixed electrode group are arranged opposite to each other to form a detection capacitor. The capacitance detection structure includes an in-plane detection section and an out-of-plane detection section. The in-plane detection section is used to detect the acceleration component parallel to the plane of the manhole cover, and the out-of-plane detection section is used to detect the acceleration component perpendicular to the plane of the manhole cover. The out-of-plane detection section is arranged in radial sections, including a central detection area and a peripheral detection area. The central detection area responds to the dynamic displacement of the mass block in the vertical direction and outputs the second Z-axis acceleration component. The peripheral detection area responds to the quasi-static angular displacement of the mass block caused by tilting and outputs the first Z-axis equivalent tilt angle signal.

[0019] Specifically, the in-plane detection unit includes: The first movable electrode group is disposed on the outer surfaces of the mass block on opposite sides along the X-axis, and is in a comb-like shape extending along the X-axis. The first fixed electrode group is disposed at the corresponding position of the substrate and engages alternately with the first movable electrode group to form two sets of variable area differential capacitors. The second movable electrode group is disposed on the outer surfaces of the mass block on opposite sides along the Y-axis, and is in a comb-like shape extending along the Y-axis. The second fixed electrode group is disposed at the corresponding position of the substrate and engages alternately with the second movable electrode group to form two sets of variable area differential capacitors. The out-of-plane detection unit includes: The third movable electrode is a metallized layer formed on the outer surface of the mass block parallel to the plane of the well cover; A planar fixed electrode group is respectively disposed on the inner surface of the substrate and the upper cover plate opposite to the substrate, and is opposite to the central region of the third movable electrode to form a variable-pitch differential capacitor; A ring-shaped fixed electrode group is arranged around the periphery of the planar fixed electrode group and is opposite to the periphery of the third movable electrode, forming a variable-pitch differential capacitor. The ring-shaped fixed electrode group, in conjunction with the third moving electrode, outputs the first Z-axis equivalent tilt angle signal, and the planar fixed electrode group, in conjunction with the third moving electrode, outputs the second Z-axis acceleration component.

[0020] In the specific dual-mode MEMS sensing unit structure diagram provided in this embodiment, Figure 2(a) is a top view of the dual-mode MEMS sensing unit, showing the overall planar layout of the mass block, X / Y axis comb electrodes, and Z-axis planar fixed electrode group. Figure 2(b) is a side view of the dual-mode MEMS sensing unit, showing the stacked cross-sectional relationship between the mass block, Z-axis planar fixed electrode group, upper / lower fixed electrodes, and ring fixed electrodes. Figure 2(c) is a top view of the dual-mode MEMS sensing unit's Z-axis electrode partitioning, showing the radial partitioning layout of the Z-axis ring fixed electrode and the central fixed electrode, as well as the relative positions of the X / Y axis fixed electrodes.

[0021] The sensing unit in the figure includes: a rectangular silicon-based mass block, a rectangular silicon-based substrate frame, an elastic cantilever beam, an X-axis detection structure, a Y-axis detection structure, a first Z-axis detection structure, and a second Z-axis detection structure; The mass block is located within the substrate frame, and its outer corner is connected to the corresponding inner corner of the frame through the elastic cantilever beam, so that the mass block is elastically suspended in the center of the frame; wherein, the X-axis detection structure, the Y-axis detection structure and the first Z-axis detection structure together constitute the tilt detection structure, and the second Z-axis detection structure alone constitutes the vibration detection structure. The X-axis detection structure includes a first movable electrode group and a first fixed electrode group. The first movable electrode group consists of comb-shaped planar electrodes disposed on two outer surfaces of the mass block perpendicular to the X-axis and perpendicular to the Z-axis; the first fixed electrode group consists of comb-shaped planar electrodes disposed on the inner surface of the corresponding frame and perpendicular to the Z-axis. The first movable electrode group and the first fixed electrode group are interlocked to form two sets of variable-area first differential capacitors. The Y-axis detection structure includes a second movable electrode group and a second fixed electrode group. The second movable electrode group consists of comb-shaped planar electrodes disposed on two outer surfaces of the mass block perpendicular to the Y-axis and perpendicular to the Z-axis; the second fixed electrode group consists of comb-shaped planar electrodes disposed on the inner surface of the corresponding frame and perpendicular to the Z-axis. The second movable electrode group and the second fixed electrode group are interlocked to form two sets of variable-area second differential capacitors. In the first Z-axis detection structure and the second Z-axis detection structure, metallization layers are respectively provided on the two outer surfaces of the mass block perpendicular to the Z-axis, serving as third moving electrodes; on the inner surface of the frame opposite to the two outer surfaces, planar fixed electrodes aligned with the center of the mass block are respectively provided, as well as annular fixed electrodes sleeved around the planar fixed electrodes; wherein, the annular fixed electrodes are arranged opposite to the third moving electrodes as fixed electrodes to form a variable-pitch third differential capacitor, which constitutes the first Z-axis detection structure; the planar fixed electrodes are arranged opposite to the third moving electrodes as fixed electrodes to form a variable-pitch fourth differential capacitor, which constitutes the second Z-axis detection structure; The first differential capacitor, the second differential capacitor, and the third differential capacitor are used for tilt angle detection, and the fourth differential capacitor is used for vibration detection, thereby realizing the parallel and synchronous operation of tilt angle detection and vibration detection at the physical level. The first movable electrode group moves synchronously along the X-axis with the mass block, while the first fixed electrode group remains stationary. When the mass block is subjected to acceleration in the X direction, it drives the first movable electrode group to move along the X-axis, changing the overlap area with the first fixed electrode group, thus causing a difference in the first differential capacitor. This difference is proportional to the acceleration in the X direction; The voltage value obtained by detection Difference with differential capacitor The following relationship must be satisfied: ; in, This is the voltage value obtained from the first differential capacitor detection. It is the capacitor-to-voltage conversion gain of the circuit; The acceleration in the X direction is: ; in, For acceleration in the X direction, Let X be the stiffness coefficient of the elastic cantilever beam. For the mass of the mass block, The vacuum permittivity, The distance between the plates is the distance between the plates when in equilibrium; N is the number of comb tooth pairs, and h is the thickness of the comb tooth.

[0022] Similarly, the second movable electrode group moves synchronously along the Y-axis with the mass block, while the second fixed electrode group remains fixed. When the mass block is subjected to acceleration in the Y direction, it drives the second movable electrode group to move along the Y-axis, changing the overlap area with the second fixed electrode group, causing the second differential capacitor to generate a difference, which is proportional to the acceleration in the Y direction. The third moving electrode moves synchronously along the Z-axis with the mass block, while the annular fixed electrode remains fixed to the planar fixed electrode. When the mass block is subjected to acceleration in the Z-direction, it causes the third moving electrode to translate along the Z-axis, changing the distance between it and the planar fixed electrode, thus generating a difference in the fourth differential capacitor, which is proportional to the Z-direction acceleration. When the device tilts, the mass block undergoes angular displacement relative to the base, causing an asymmetrical change in the distance between the third moving electrode and the annular fixed electrode, resulting in a difference in the third differential capacitor, which is proportional to the tilt angle.

[0023] When the equipment tilts, the distance between the Z-axis annular fixed electrode and the mass block changes asymmetrically, and the difference in the third differential capacitor is proportional to the tilt angle. The voltage value obtained by detection Difference with differential capacitor The following relationship must be satisfied: ; in, This is the voltage value obtained from the third differential capacitor. It is the capacitor-to-voltage conversion gain of the circuit; Differential capacitor tilt angle measurement value It can be represented as: ,in, The sensitivity coefficient of the sensor represents the change in differential capacitance corresponding to the sine value of a unit tilt angle.

[0024] When the mass block is subjected to acceleration in the Z direction, it causes the third moving electrode to translate along the Z-axis, changing the distance between it and the fixed plane electrode, and the difference in the fourth differential capacitor value... It is proportional to the Z-axis acceleration; ; in, This is the voltage value obtained from the fourth differential capacitor detection. It is the capacitor-to-voltage conversion gain of the circuit; The acceleration in the Z direction is: ; in, Z-axis acceleration The Z-axis stiffness coefficient of the elastic cantilever beam is given. For the mass of the mass block, The vacuum permittivity, The area of ​​the electrode plate. The electrode spacing is at equilibrium.

[0025] The dual-mode MEMS sensing unit can be packaged in a standard LGA package with dimensions of 3mm × 3.25mm × 1.06mm, an operating voltage of 1.7V to 3.6V, vibration detection ranges of ±2g, ±4g and ±8g, and a power consumption of 18μA in normal operating mode.

[0026] The dual-mode MEMS sensing unit synchronously outputs multi-axis detection signals at a sampling frequency of not less than 100Hz, triggering a data readout every 10ms. This sampling rate can fully capture transient events such as illegal opening (tilt angle change > 10° / s) and instantaneous impact (lasting < 0.5s), avoiding signal loss due to excessively long sampling intervals (such as traditional low sampling rate sensors that may miss the peak of high-frequency impacts).

[0027] This embodiment's dual-modal MEMS sensing unit generates tilt and vibration detection signals synchronously based on the displacement of the same mass block. It employs a radially partitioned structure for the out-of-plane detection section (the central detection area outputs dynamic acceleration components, and the outer detection area outputs quasi-static tilt components), achieving parallel physical-level operation of tilt and vibration detection on a single sensing structure. Compared to traditional dual-sensor splicing schemes, this eliminates data synchronization delays and PCB wiring signal attenuation between dissimilar sensors, reducing package size. Furthermore, by separating and multiplexing the static / dynamic components of the same source signal, redundant acquisition is avoided, significantly reducing system power consumption.

[0028] Specifically, the processing unit, connected to the dual-modal MEMS sensing unit, includes: a preprocessing module, an tilt angle calculation module, and a vibration extraction module.

[0029] The preprocessing module is used to perform preprocessing on the multi-axis detection signal, including filtering, noise reduction, outlier removal, and quantization. Filtering: The vibration path uses a digital bandpass filter with a passband frequency of 0.5Hz–200Hz; the lower cutoff frequency of 0.5Hz is used to filter out low-frequency DC drift and sensor zero-bias temperature drift caused by slow temperature changes, and the upper cutoff frequency of 200Hz is used to filter out high-frequency electromagnetic interference and quantization noise above the effective vibration frequency band, while retaining the quasi-static tilt component and vibration dynamic component required for manhole cover status identification; the tilt path uses a low-pass filter (such as 0–5 Hz or 0–10 Hz) to retain the quasi-static component; Noise reduction: A sliding average smoothing process is adopted, with the sliding window length set to 10 sampling points (corresponding to a 0.1s window). While preserving the edge features of the effective signal, it suppresses the high-frequency spikes caused by the internal electronic noise of the sensor and the micro-vibration of the environment, thereby reducing the noise floor of subsequent feature extraction. Outlier removal: If a single peak value exceeds the set outlier threshold (e.g., 10g) within the window, but the frequency domain energy distribution of the other sampling points within the window does not show the corresponding frequency band concentration characteristics, then the peak value is determined to be a pseudo feature caused by sensor noise or sudden electromagnetic pulse, and is removed and not included in subsequent calculations. Quantification: Convert all features into standardized values ​​(e.g., normalize peak values ​​to the 0–10g range) to avoid unit or magnitude differences affecting the judgment.

[0030] The static tilt angle calculation module is used to calculate the pitch and roll angles based on the static components of the preprocessed signal using the biaxial component ratio method, and to calculate the acceleration tilt angle measurement value by combining the tangents of the pitch and roll angles; in order to eliminate the signal-to-noise ratio collapse problem of a single reference component at large tilt angles, and achieve high-precision static tilt angle output across the entire range.

[0031] The tilt angle calculation module receives the first Z-axis equivalent tilt angle signal output from the X-axis detection structure, Y-axis detection structure, and first Z-axis detection structure (ring-shaped fixed electrode) after static filtering; after converting the first Z-axis equivalent tilt angle signal into equivalent Z-axis acceleration components, the pitch angle is calculated in real time using the dual-axis composite component ratio method. With roll angle And calculate the rate of change of tilt angle. ;;in, The formula is: ; ; ; , , These are the acceleration components along the X, Y, and Z axes, respectively.

[0032] The principle of large tilt angle calculation: The commonly used pitch and roll angle formulas simplify the calculation of dual-axis or single-axis systems by using only the components of 1 to 2 axes and ignoring the information of the third axis. Since the "reference component" (such as the single Z-axis gravity component) tends to be close to 0 near 90°, the signal-to-noise ratio of "input signal / reference signal" drops sharply. Therefore, it is suitable for small tilt angle ranges (such as 0° to 60°) and has poor adaptability to large tilt angles.

[0033] This invention employs a dual-axis component calculation. Taking the pitch angle as an example, it uses the X-axis component and the Y / Z-axis component. Utilizing the components of all three axes, it calculates the angle using the ratio of "single-axis / dual-axis component" to cover the entire range (0°~90°), maintaining stability even at large tilt angles. Essentially, it integrates the gravity components of the two axes (Y and Z) perpendicular to the X-axis into an equivalent reference. The angle is then calculated using the ratio of the X-axis component to this equivalent reference. At this point, the angle calculation no longer depends on the component of a single axis, but rather covers the entire angle range through the synergy of the single-axis active component and the dual-axis reference component.

[0034] The tilt fusion module is used to receive the acceleration tilt angle measurement value output by the static tilt angle calculation module and the differential capacitance tilt angle measurement value output by the ring fixed electrode and the third moving electrode. Based on the motion intensity, it adaptively switches between the acceleration tilt angle measurement value and the differential capacitance tilt angle measurement value. In static and quasi-static conditions, the acceleration tilt angle measurement value is used as the main factor to eliminate capacitance drift. In dynamic disturbance conditions, the capacitance tilt angle measurement value is used as the main factor to suppress motion interference, and outputs a high-precision fused tilt angle under all working conditions. The tilt angle value calculated by acceleration is highly accurate in static conditions and exhibits no long-term drift. The tilt angle value obtained through differential capacitance, based on quasi-static angular displacement, directly detects the tilt angle, making it immune to motion acceleration and less susceptible to its influence. The fused signal is primarily contributed by acceleration in the low-frequency range, thus eliminating the drift of the capacitive sensor; in the high-frequency range, it is primarily contributed by the capacitive sensor, thus suppressing motion acceleration interference from the accelerometer.

[0035] Set an exercise intensity To determine the intensity of motion, g represents the acceleration due to gravity. ; ; When the object is stationary, M is 0. When M < 0.2, the tilt angle is calculated using acceleration. When M ≥ 0.2, it indicates that the carrier is subjected to significant non-gravitational acceleration, and the tilt angle measurement of the accelerometer is affected by vibration. In this case, the tilt angle value obtained by differential capacitance is used, and its immunity to motion acceleration is used to ensure the accuracy of tilt angle detection.

[0036] The vibration extraction module is used to extract time-domain and frequency-domain features from the Z-axis acceleration component perpendicular to the manhole cover plane output by the second Z-axis detection structure (planar fixed electrode). The time-domain features include peak value, root mean square (RMS), and duration, while the frequency-domain features include dominant frequency, frequency band energy distribution, and high-frequency energy proportion. By transforming multi-dimensional features such as vibration intensity, energy, frequency distribution, and duration into quantifiable and comparable judgment conditions, the limitations of misjudgment caused by single features are avoided.

[0037] The processing unit in this embodiment uses a biaxial composite component ratio method to calculate the pitch and roll angles based on the equivalent tilt angle signals of the X, Y, and first Z axes. It integrates the gravity components of the two axes perpendicular to the active axis into an equivalent reference. Through the synergy of the single-axis active component and the biaxial reference components, it covers the entire tilt angle range of 0°–90°. Compared to traditional simplified biaxial or single-axis algorithms that suffer from signal-to-noise ratio collapse at large tilt angles due to the reference component approaching zero, this solution maintains calculation stability and accuracy even at large tilt angles, avoiding angle misjudgments when the manhole cover is fully open or undergoes significant settlement.

[0038] The fusion decision unit, connected to the processing unit, is used to receive the tilt features output by the tilt angle calculation module and the vibration features output by the vibration extraction module; the tilt features include real-time pitch angle, roll angle and tilt angle change rate, and the vibration features include time domain features (peak value, root mean square, duration) and frequency domain features (dominant frequency, frequency band energy distribution, high frequency energy ratio).

[0039] The fusion decision unit has a built-in tilt angle cache queue and a vibration event cache queue, and maintains a timing correlation window of ±2 seconds based on the system clock to ensure that tilt angle anomalies and vibration events have temporal consistency at the physical causal level.

[0040] The fusion decision unit is internally divided into a vibration mode initial screening module, a timing verification and correction module, and a dual-mode linkage decision module.

[0041] The vibration mode initial screening module is used for a three-level progressive screening based on peak threshold, root mean square stability threshold and dominant frequency range threshold to initially classify vibration modes into traffic crushing, continuous vibration, instantaneous impact and environmental interference; avoiding misjudgment based on a single feature.

[0042] The vibration mode primary screening module includes: The peak threshold delimitation submodule is used to quickly delimit vibration modes based on the peak value and form candidate vibration modes; The peak threshold is set to two levels: 1g and 5g. If the vibration peak is below 1g, it is directly nominated as the environmental interference mode and will not proceed to further screening. If the vibration peak is between 1g and 5g, it is nominated as the traffic crushing mode or the continuous vibration mode. If the vibration peak is above 5g, it is nominated as the instantaneous impact mode.

[0043] The root mean square stability verification submodule is used to screen the candidate vibration modes for stability based on the root mean square value within the sliding window and the fluctuation between windows. For the candidate modes output in the first level, calculate the root mean square (RMS) value and the fluctuation between windows within a sliding window (window duration 0.5s). Candidate traffic compaction modes must have an RMS value between 0.8g and 2g; candidate continuous vibration modes must have a stable RMS value and a fluctuation between windows of less than 0.3g; candidate instantaneous impact modes must have an RMS value less than 1g. If a candidate mode does not meet the corresponding RMS conditions, it is downgraded to an environmental disturbance mode.

[0044] The main frequency range locking submodule is used to lock the final vibration mode category based on the frequency domain energy concentration band; For candidate modes selected through the root mean square stability verification submodule, the dominant frequency and energy concentration band are extracted based on fast Fourier transform. Candidate traffic compaction modes must match a dominant frequency in the 10Hz to 30Hz range; candidate continuous vibration modes must match a dominant frequency in the 1Hz to 20Hz range; and candidate instantaneous impact modes must match a dominant frequency higher than 80Hz. If the dominant frequency deviates from the corresponding range, the candidate mode is corrected to an environmental disturbance mode or a similar vibration category.

[0045] After progressive screening by the above three sub-modules, preliminary classification results are output, including four candidate vibration modes: traffic crushing, continuous vibration, instantaneous impact, and environmental disturbance, along with the confidence score corresponding to each mode.

[0046] The timing verification and correction module is used to perform timing consistency verification and feature coupling verification on the preliminary classification results based on duration, repetition frequency, frequency band energy distribution and energy concentration, and to backtrack and correct boundary cases that have contradictions after verification, so as to obtain the corrected vibration mode.

[0047] The timing verification and correction module includes: The timing consistency verification submodule is used to verify whether the duration and repetition frequency of candidate vibration modes conform to the physical timing pattern of such vibration events; For example, if the initial classification is traffic crushing mode, it is necessary to verify whether its duration is in the range of 0.3s-1s and whether the number of recurrences within 1 minute is greater than or equal to 3; if there is only a single occurrence of a peak value of 3g and a main frequency of 20Hz, but no recurrence record, it should be corrected to environmental interference mode (such as heavy object falling). The feature coupling verification submodule is used to verify whether the spectral energy distribution and energy concentration of candidate vibration modes conform to the frequency domain fingerprint characteristics of such vibration modes. For example, if initially classified as a transient impact mode, it is necessary to verify whether its duration is less than 0.5s and whether the high-frequency band energy accounts for more than 70%; if the peak value is greater than 5g but the duration exceeds 1s, it should be corrected to a continuous vibration mode (such as continuous impact from heavy machinery). Similarly, if initially classified as a continuous vibration mode, it is necessary to verify whether its vibration amplitude fluctuation is less than 0.3g and whether the frequency band energy in the 1Hz–20Hz range accounts for more than 80%; if the vibration amplitude fluctuation is greater than 0.5g, it should be corrected to an environmental interference mode (such as random impact from construction dust). Furthermore, if initially classified as an environmental interference mode, it is necessary to verify whether it meets all the characteristics of being irregular and without repetition; if a peak value of 0.8g and a dominant frequency of 15Hz occur without repetition, the environmental interference mode classification should still be maintained. The boundary backtracking correction submodule is used to backtrack and correct boundary cases that have contradictions after time-series consistency verification and feature coupling verification. When a candidate vibration mode meets the timing conditions but not the frequency domain and amplitude conditions, or meets the frequency domain conditions but not the timing conditions, the candidate mode is corrected to a secondary vibration category that matches its actual multi-characteristic performance; when a candidate vibration mode does not meet both the timing and frequency domain conditions, it is corrected to an environmental interference mode and marked as an invalid anomaly and discarded.

[0048] The dual-modal linkage decision module is used to define the tilt angle change mode based on the absolute value of the tilt angle change rate, and jointly map the tilt angle change mode with the corrected vibration mode. When the tilt angle change mode is a sudden change and the corrected vibration mode is an instantaneous impact, based on the coupling relationship between the number of sign flips in the tilt angle change direction and the final change in tilt angle, the module distinguishes between successful illegal opening, illegal prying, and attempted opening behaviors in order of severity, and outputs the manhole cover status identification result.

[0049] The dual-modal linkage decision module includes: The tilt change mode definition submodule is used to define the tilt change mode based on the absolute value of the tilt change rate; the tilt change mode is defined as three modes: abrupt change, slow change, and no change. For example, a change greater than 10° / s is defined as a sudden change, a change between 2° / s and 10° / s is defined as a slow change, and a change less than 2° / s is defined as no change.

[0050] The vibration-tilt joint mapping submodule is used to trigger the corresponding decision process based on the combination of the tilt change mode and the corrected vibration mode. If the tilt angle change mode is abrupt and the corrected vibration mode is instantaneous impact, then the illegal opening depth determination process is triggered. If the tilt angle change pattern is slow and the corrected vibration pattern is continuous vibration, the settlement risk classification determination process is triggered. If the tilt angle change mode is unchanged and the corrected vibration mode is traffic crushing, it is determined to be a normal traffic scenario, no alarm is triggered, and only the traffic log is recorded. If the tilt angle change mode is unchanged and the corrected vibration mode is due to environmental interference, it is determined that there is no abnormal event. After completing the current calculation cycle, the system enters a deep sleep state and waits for the next sensor interruption to wake it up.

[0051] The illegal opening depth determination submodule is used to perform depth-level determination of illegal opening events based on the coupling relationship between the sign flip feature of the tilt angle change direction and the final change in tilt angle. In the module, after removing environmental micro-vibration noise through dead-zone filtering, the symbol flip information of the tilt angle change direction within the observation window is extracted. The symbol flip information is used to characterize the intermittent force mechanical characteristics caused by the illegal tool's inability to form a stable fit with the lock cylinder or the edge of the manhole cover. The density of the symbol flip is jointly mapped with the final change in tilt angle. Based on the judgment condition gradient that both satisfy, the illegal opening is distinguished from the most serious to the least serious, and the illegal prying and attempted opening behaviors are distinguished in order of severity, and the corresponding alarm response is triggered.

[0052] Because illegal tools (such as crowbars) cannot form a stable fit with the lock cylinder or the edge of the manhole cover, the force application process often exhibits a discontinuous characteristic of "applying force → getting stuck → breaking through → applying force again → getting stuck again", resulting in the manhole cover tilt angle exhibiting a repeated shaking phenomenon of "increasing → slightly retreating → increasing again".

[0053] The illegal opening depth determination submodule includes: The tilt angle change direction detection module is used to calculate the tilt angle difference between adjacent sampling times, and after filtering out false tilt angle jitter caused by environmental micro-vibrations and sensor noise based on the dead zone threshold, it determines the direction of change of the tilt angle at the current time relative to the previous time. The symbol flipping counting module is used to record the cumulative number of times the tilt angle change direction flips within the observation window; the flipping is used to characterize the intermittent force application and jitter characteristics caused by the illegal tool's inability to form a stable fit with the lock core or the edge of the manhole cover; The graded judgment module is used to output the judgment results of illegal opening success, illegal prying and attempted opening behavior in order of severity, based on the coupling relationship between the cumulative number of flips and the final change in tilt angle, and to trigger emergency alarm, early warning record or extended monitoring continuous tracking mode respectively.

[0054] Examples of processing procedures include: Calculate the difference between two adjacent tilt angle samples And set a dead zone threshold. Used to filter out sensor noise and false flips caused by environmental micro-vibrations; when At that time, if Then it is determined that the tilt angle has increased. Then it is determined that the tilt angle is decreasing; record the number of times the sign of the rate of change of tilt angle is flipped. Simultaneously monitor the final change in tilt angle. .

[0055] like and If the signal is not received, it is determined that the device was opened illegally and an emergency alarm is triggered. The alarm is then reported to the monitoring platform via the NB-IoT module. The reported data packet includes the event type, timestamp, tilt angle change value, vibration peak value, and number of flips. like and If the manhole cover is illegally pried open, the warning event will be recorded and reported, indicating the risk of attempted opening. like If the above threshold conditions are not met, it is determined to be an attempted opening behavior, triggering early warning monitoring and extending the observation window to continuously track the trend of tilt angle changes.

[0056] Furthermore, it also includes: The settlement risk classification and determination submodule is used to distinguish settlement events from surrounding construction vibrations; If the corrected vibration pattern is continuous vibration and the tilt angle change pattern is slow change, then the risk classification is further carried out according to the temporal distribution characteristics: if the continuous vibration occurs continuously and stably for 24 hours, it is preferentially identified as settlement; if it occurs intermittently during the day and disappears at night, it is preferentially identified as settlement risk accompanied by vibration from surrounding construction, and the risk level label is attached when reporting.

[0057] Characteristics of a single sudden sustained vibration (settlement / construction): peak value 1-3g; root mean square 0.5-1.5g; dominant frequency 1-20Hz; energy concentrated in 1-20Hz; frequency band duration >3s (settlement) or intermittent repetition (construction); stable vibration amplitude.

[0058] The boundary case verification submodule is used to verify boundary cases where there is a contradiction between the tilt angle change mode and the vibration mode; If the tilt angle changes abruptly but the vibration mode is due to environmental disturbance, or if the tilt angle remains unchanged but the vibration mode is due to instantaneous impact, the system outputs a "pending confirmation" status, extends the monitoring window to 5 seconds, and increases the sampling rate to 400Hz for verification. If no valid chain of evidence conforming to the joint mapping rules is found during the verification period, the no-anomaly judgment is maintained to avoid false alarms caused by single-dimensional anomalies.

[0059] The authorized operation verification submodule is used to synchronously query whether there is a valid unlocking record uploaded by the electronic key when an illegal opening alarm or warning is triggered. If a valid unlocking record exists, it is determined to be an authorized operation, and the illegal opening alarm or warning is canceled to avoid false alarms.

[0060] In the integrated judgment unit, a progressive judgment mechanism of "initial screening - temporal verification - linkage mapping" is used to quickly delineate vibration modes using multi-feature thresholds. Boundary cases are then corrected through temporal characteristics and multi-feature cross-verification. Finally, a linkage judgment is made by combining the temporal correlation between tilt angle change patterns and vibration modes. Utilizing the mutual constraints of multi-dimensional evidence chains of tilt angle and vibration, environmental interference such as traffic crushing, construction dust, and falling heavy objects is effectively eliminated, significantly reducing false alarms caused by single-dimensional anomalies. Based on the coupling relationship between the sign reversal feature of the tilt angle change direction and the final change in tilt angle, deep hierarchical identification of illegal opening events is performed. By capturing the intermittent force jitter features (increased tilt angle - slight retreat - further increased sign reversal) caused by the illegal tool's inability to form a stable fit with the lock cylinder or manhole cover edge, illegal opening, illegal prying, and attempted opening behaviors are distinguished in descending order of severity. Linked with electronic key authorization verification, false alarms of legitimate operations are eliminated, achieving accurate tracing and hierarchical response to manhole cover opening behavior.

[0061] Example 2 One embodiment of the present invention discloses a manhole cover status recognition method based on tilt angle-vibration dual-mode fusion, which uses the manhole cover status recognition device based on tilt angle-vibration dual-mode fusion from Embodiment 1 and performs the following steps: The system synchronously acquires multi-axis detection signals of a single mass block relative to the manhole cover. The multi-axis detection signals include X and Y axis acceleration components parallel to the plane of the manhole cover, and a first Z-axis equivalent tilt angle signal and a second Z-axis acceleration component perpendicular to the plane of the manhole cover. The first Z-axis equivalent tilt angle signal is detected by a combination of a ring-shaped fixed electrode and a third moving electrode located in the outer region of the mass block. The second Z-axis acceleration component is detected by a planar fixed electrode and a third moving electrode located in the central region. Dual-modal parallel processing; the tilt angle features are calculated by fusing the static components of the equivalent tilt angle signals of the X, Y axes and the first Z axis, and the vibration features are extracted based on the dynamic components of the acceleration components of the second Z axis; Fusion judgment; based on the temporal correlation analysis of tilt angle features and vibration features, the vibration modes are initially classified through multi-feature threshold matching. After correction by temporal characteristics and multi-feature cross-validation, the tilt angle change mode and the corrected vibration mode are linked and mapped to output the manhole cover status recognition result.

[0062] The specific technical details and beneficial effects of this embodiment are the same as those of Embodiment 1. Please refer to them for details, and they will not be repeated here.

[0063] Example 3 One embodiment of the present invention discloses a manhole cover state recognition method based on tilt angle-vibration dual-mode fusion, which uses the manhole cover state recognition device based on tilt angle-vibration dual-mode fusion from Embodiment 1 and performs the following steps: S1, System initialization; Hardware initialization: Configure the MCU's SPI, UART and timer interfaces; initialize the dual-mode MEMS sensing unit to sampling mode (100Hz) and enable three-axis synchronous acquisition; configure the NB-IoT module to PSM (Power Saving Mode) mode and preset the heartbeat cycle.

[0064] Software initialization: Load preset threshold parameters (including peak thresholds at various levels, root mean square thresholds, dominant frequency range, tilt rate of change threshold, dead zone threshold). (etc.); create a tilt angle buffer queue and a vibration event buffer queue; initialize the system state to deep sleep ready state.

[0065] S2, Dual-modal data synchronous acquisition; Synchronous acquisition is triggered with a period of 10ms to read the X-axis and Y-axis acceleration components, the first Z-axis equivalent tilt angle signal (tilt angle detection path), and the second Z-axis acceleration component (vibration detection path) output by the dual-mode MEMS sensing unit. The equivalent tilt angle signals of the X, Y axes and the first Z axis are stored in the tilt angle buffer queue, and the acceleration component of the second Z axis is stored in the vibration buffer queue; the system timestamp is recorded synchronously.

[0066] S3, Parallel processing of dual-mode signals; Tilt path: Low-pass filtering, moving average noise reduction, and outlier removal are performed on the data in the tilt buffer queue; the pitch angle is calculated in real time using the dual-axis composite component ratio method. With roll angle And calculate the rate of change of tilt angle. If the tilt angle is greater than 20° and the rate of change is greater than 10° / s, mark it as "tilt angle abnormal" and record the timestamp.

[0067] Vibration path: The data in the vibration buffer queue is filtered and noise-reduced preprocessed; time-domain features (peak value, root mean square, duration) and frequency-domain features (dominant frequency, frequency band energy distribution, high-frequency energy ratio) are extracted.

[0068] S4. Integrating judgment and incident reporting; Input the tilt features (real-time pitch angle, roll angle, and tilt rate of change) and vibration features (time domain features and frequency domain features) into the fusion decision unit; The manhole cover status identification result is output after three levels of progressive judgment: vibration mode initial screening module, timing verification and correction module, and dual-mode linkage judgment module. Based on the identification results, the following actions are triggered: if unauthorized opening is successful, an emergency alarm is triggered and the NB-IoT is reported to the monitoring platform; if unauthorized prying is recorded as an early warning event and reported to the monitoring platform; if a settlement / tilt event is reported with its risk level attached; if normal passage is recorded in the log; if there are no abnormalities, proceed to step S5.

[0069] S5, low-power loop control; If there are no valid abnormal events in the current calculation cycle, the system will continue to run for 30 seconds and then enter deep sleep mode (current < 10μA), retaining only the low-power interrupt detection and timer wake-up function of the dual-mode MEMS sensing unit (5-minute cycle). After being interrupted by the sensor (detection of tilt angle exceeding the threshold or vibration peak reaching the standard) or woken up by the timer, return to step S2 and repeat the above process.

[0070] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A manhole cover status recognition device based on tilt angle-vibration dual-mode fusion, characterized in that, include: The dual-modal MEMS sensing unit synchronously outputs tilt angle detection signals and vibration detection signals based on the displacement of the same mass block. The tilt angle detection signal includes X and Y axis acceleration components parallel to the manhole cover plane and a first Z-axis equivalent tilt angle signal perpendicular to the manhole cover plane. The vibration detection signal includes a second Z-axis acceleration component perpendicular to the manhole cover plane. The first Z-axis equivalent tilt angle signal characterizes the quasi-static angular displacement of the mass block caused by the tilt of the manhole cover, and the second Z-axis acceleration component reflects the dynamic acceleration of the mass block caused by external force impact. The processing unit, connected to the dual-mode MEMS sensing unit, is used to fuse and calculate tilt features based on tilt detection signals and extract vibration features based on vibration detection signals. The fusion decision unit, connected to the processing unit, is used for time-series correlation analysis based on tilt angle features and vibration features. It initially classifies vibration modes through multi-feature threshold matching, and after correction by time-series characteristics and multi-feature cross-validation, it performs linkage mapping by combining tilt angle change mode and the corrected vibration mode to output the manhole cover status identification result.

2. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 1, characterized in that, The dual-modal MEMS sensing unit includes: a single mass block, an elastic suspension structure, and a capacitance detection structure; wherein... A single mass block is suspended by an elastic suspension structure, and the mass block has displacement degrees of freedom that are parallel to and perpendicular to the plane of the manhole cover; The capacitance detection structure includes a movable electrode group that moves synchronously with the mass block and a fixed electrode group that is fixed relative to the base. The movable electrode group and the fixed electrode group are arranged opposite to each other to form a detection capacitor. The capacitance detection structure includes an in-plane detection section and an out-of-plane detection section. The in-plane detection section is used to detect the acceleration component parallel to the plane of the manhole cover, and the out-of-plane detection section is used to detect the acceleration component perpendicular to the plane of the manhole cover. The out-of-plane detection section is arranged in radial sections, including a central detection area and a peripheral detection area. The central detection area responds to the dynamic displacement of the mass block in the vertical direction and outputs the second Z-axis acceleration component. The peripheral detection area responds to the quasi-static angular displacement of the mass block caused by tilting and outputs the first Z-axis equivalent tilt angle signal.

3. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 2, characterized in that, The in-plane detection unit includes: The first movable electrode group is disposed on the outer surfaces of the mass block on opposite sides along the X-axis, and is in a comb-like shape extending along the X-axis. The first fixed electrode group is disposed at the corresponding position of the substrate and engages alternately with the first movable electrode group to form two sets of variable area differential capacitors. The second movable electrode group is disposed on the outer surfaces of the mass block on opposite sides along the Y-axis, and is in a comb-like shape extending along the Y-axis. The second fixed electrode group is disposed at the corresponding position of the substrate and engages alternately with the second movable electrode group to form two sets of variable area differential capacitors. The out-of-plane detection unit includes: The third movable electrode is a metallized layer formed on the outer surface of the mass block parallel to the plane of the well cover; A planar fixed electrode group is respectively disposed on the inner surface of the substrate and the upper cover plate opposite to the substrate, and is opposite to the central region of the third movable electrode to form a variable-pitch differential capacitor; An annular fixed electrode group is respectively arranged around the periphery of the corresponding planar fixed electrode and on the inner surface of the substrate and the upper cover plate, opposite to the periphery of the third movable electrode, forming a variable-pitch differential capacitor; The ring-shaped fixed electrode group, in conjunction with the third moving electrode, outputs the first Z-axis equivalent tilt angle signal, and the planar fixed electrode group, in conjunction with the third moving electrode, outputs the second Z-axis acceleration component.

4. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 3, characterized in that, The processing unit includes: The static tilt angle calculation module is used to calculate the pitch angle and roll angle based on the static component of the preprocessed signal by the biaxial component ratio method, and to calculate the acceleration tilt angle measurement value by combining the tangents of the pitch angle and roll angle. The tilt fusion module is used to receive the acceleration tilt angle measurement value output by the static tilt angle calculation module and the first Z-axis equivalent tilt angle signal output by the annular fixed electrode group and the third moving electrode. Based on the motion intensity, it adaptively switches the acceleration tilt angle measurement value and the first Z-axis equivalent tilt angle signal. In static and quasi-static conditions, the acceleration tilt angle measurement value is used as the main factor to eliminate capacitance drift. In dynamic disturbance conditions, the first Z-axis equivalent tilt angle signal is used as the main factor to suppress motion interference, and the tilt angle feature with high accuracy under all working conditions is output. The vibration extraction module is used to extract time-domain features and frequency-domain features based on the second Z-axis acceleration component; the time-domain features include peak value, root mean square value and duration, and the frequency-domain features include dominant frequency, frequency band energy distribution and high-frequency energy ratio.

5. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 1, characterized in that, The fusion decision unit includes: The vibration mode initial screening module is used for a three-level progressive screening based on peak threshold, root mean square stability threshold and dominant frequency range threshold to initially classify vibration modes into traffic rolling, continuous vibration, instantaneous impact and environmental interference. The timing verification and correction module is used to perform timing consistency verification and feature coupling verification on the preliminary classification results based on duration, repetition frequency, frequency band energy distribution and energy concentration, and to backtrack and correct boundary cases that have contradictions after verification. The dual-modal linkage decision module is used to define the tilt angle change mode based on the absolute value of the tilt angle change rate, and jointly map the tilt angle change mode with the corrected vibration mode. When the tilt angle change mode is a sudden change and the corrected vibration mode is an instantaneous impact, based on the coupling relationship between the number of sign flips in the tilt angle change direction and the final change in tilt angle, the module distinguishes between successful illegal opening, illegal prying, and attempted opening behaviors in order of severity, and outputs the manhole cover status identification result.

6. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 5, characterized in that, The vibration mode primary screening module includes: The peak threshold delimitation submodule is used to quickly delimit vibration modes based on the peak value and form candidate vibration modes; The root mean square stability verification submodule is used to screen the candidate vibration modes for stability based on the root mean square value within the sliding window and the fluctuation between windows. The main frequency range locking submodule is used to lock the final vibration mode category based on the frequency domain energy concentration band.

7. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 5, characterized in that, The timing verification and correction module includes: The timing consistency verification submodule is used to verify whether the duration and repetition frequency of candidate vibration modes conform to the physical timing pattern of such vibration events; The feature coupling verification submodule is used to verify whether the spectral energy distribution and energy concentration of candidate vibration modes conform to the frequency domain fingerprint characteristics of such vibration modes. The boundary backtracking correction submodule is used to backtrack and correct boundary cases that have contradictions after time-series consistency verification and feature coupling verification.

8. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 5, characterized in that, The dual-modal linkage decision module includes: The tilt angle change mode definition submodule is used to define the tilt angle change mode based on the absolute value of the tilt angle change rate; the tilt angle change mode is defined as three modes: abrupt change, slow change and no change. The vibration-tilt joint mapping submodule is used to trigger the corresponding decision process based on the combination of the tilt change mode and the corrected vibration mode. If the tilt angle change mode is abrupt and the corrected vibration mode is instantaneous impact, then the illegal opening depth determination process is triggered. If the tilt angle change pattern is slow and the corrected vibration pattern is continuous vibration, the settlement risk classification determination process is triggered. If the tilt angle change mode is unchanged and the corrected vibration mode is traffic crushing, it is determined to be a normal traffic scenario, no alarm is triggered, and only the traffic log is recorded. If the tilt angle change mode is unchanged and the corrected vibration mode is environmental interference, it is determined that there is no abnormal event. After completing the current calculation cycle, the system enters a deep sleep state and waits for the next sensor interruption to wake it up. The illegal opening depth determination submodule is used to perform depth-level determination of illegal opening events based on the coupling relationship between the sign flip feature of the tilt angle change direction and the final change in tilt angle.

9. The manhole cover status recognition device based on tilt angle-vibration dual-mode fusion according to claim 8, characterized in that, The illegal opening depth determination submodule includes: The tilt angle change direction detection module is used to calculate the tilt angle difference between adjacent sampling times, and after filtering out false tilt angle jitter caused by environmental micro-vibrations and sensor noise based on the dead zone threshold, it determines the direction of change of the tilt angle at the current time relative to the previous time. The symbol flipping counting module is used to record the cumulative number of times the tilt angle change direction flips within the observation window; the flipping is used to characterize the intermittent force application and jitter characteristics caused by the illegal tool's inability to form a stable fit with the lock core or the edge of the manhole cover; The graded judgment module is used to output the judgment results of illegal opening success, illegal prying and attempted opening behavior in order of severity, based on the coupling relationship between the cumulative number of flips and the final change in tilt angle, and to trigger emergency alarm, early warning record or extended monitoring continuous tracking mode respectively.

10. A method for identifying the state of manhole covers based on tilt angle-vibration dual-mode fusion, characterized in that, Using the manhole cover status recognition device based on tilt angle-vibration dual-mode fusion as described in any one of claims 1-9, the following steps are performed: The system synchronously acquires multi-axis detection signals of a single mass block relative to the manhole cover; the multi-axis detection signals include X and Y axis acceleration components parallel to the plane of the manhole cover, and a first Z-axis equivalent tilt angle signal and a second Z-axis acceleration component perpendicular to the plane of the manhole cover; wherein, the first Z-axis equivalent tilt angle signal is detected by the annular fixed electrode group in conjunction with the third moving electrode, and the second Z-axis acceleration component is detected by the planar fixed electrode group in conjunction with the third moving electrode; Dual-modal parallel processing; the tilt angle features are calculated by fusing the static components of the equivalent tilt angle signals of the X, Y axes and the first Z axis, and the vibration features are extracted based on the dynamic components of the acceleration components of the second Z axis; Fusion judgment; based on the temporal correlation analysis of tilt angle features and vibration features, the vibration modes are initially classified through multi-feature threshold matching. After correction by temporal characteristics and multi-feature cross-validation, the tilt angle change mode and the corrected vibration mode are linked and mapped to output the manhole cover status recognition result.

Images