An adaptive data sampling rate adjustment system for inclinometers

Abstract

This invention discloses an adaptive data sampling rate adjustment system for inclinometers, relating to the field of inclinometer technology. It includes: a raw data acquisition module, used to collect observation data from multiple measuring points in real time at preset sampling intervals as the inclinometer probe moves along the inclinometer tube, and output the data in a structured format; and a data prediction module, used to preload historical observation data collected by the raw data acquisition module, determine potential deformation risks based on the correlation characteristics between depth and time, and output a predicted risk marker for a depth segment if the risk level of a certain depth segment exceeds a predicted risk threshold. This invention integrates multiple functional modules such as historical data prediction, current observation data analysis, and resource status awareness. Through information linkage between modules, it can achieve full-process adaptive control of sampling behavior, possessing higher responsiveness and sampling efficiency, and is suitable for various dynamic monitoring environments.

Description

Technical Field

[0001] This invention relates to the field of inclinometer technology, specifically to an adaptive adjustment system for the data sampling rate of an inclinometer. Background Technology

[0002] In applications such as geological engineering and safety monitoring, inclinometers are widely deployed in structures such as slopes, tunnels, and dams as an important underground deformation monitoring device to collect inclination information at different depths. Inclinometers are usually moved along the inclinometer tube by mechanical lowering or lifting and observe multiple measuring points at a set sampling interval to form a longitudinal inclination profile that reflects the spatial deformation pattern. In order to improve measurement efficiency and meet the data collection needs of areas with different risk levels, the sampling control strategy of inclinometer systems is gradually developing towards automation and self-adaptation.

[0003] In existing technologies, some inclinometer systems have achieved basic automatic sampling functions, enabling them to complete batch measurement tasks according to fixed sampling intervals. However, in practical applications involving long-term operation, complex geological environments, or multi-task scheduling, challenges such as resource constraints, superposition of interference signals, and difficulty in identifying data anomalies still exist. Therefore, how to dynamically adjust the sampling strategy by combining the characteristics of observation data and the status of system resources while ensuring measurement accuracy has become a key direction for the intelligent development of inclinometer systems.

[0004] In recent years, engineering sites have placed higher demands on the real-time performance, reliability, and resource adaptability of inclinometer systems, prompting the evolution of sampling strategies from "unified configuration" to "on-demand adjustment." Therefore, this invention proposes an adaptive adjustment system for the data sampling rate of inclinometers. Summary of the Invention

[0005] The purpose of this invention is to provide a data sampling rate adaptive adjustment system for inclinometers to solve the problems mentioned in the background art.

[0006] This invention can be achieved through the following technical solution: a data sampling rate adaptive adjustment system for an inclinometer, comprising a raw data acquisition module, a data prediction module, a disturbance identification module, a resource status assessment module, a dynamic sampling adjustment module, a sampling control execution module, and a data output module;

[0007] The raw data acquisition module is used to collect observation data of multiple measuring points in real time according to a preset sampling interval during the movement of the inclinometer probe along the inclinometer tube. The observation data includes measuring point depth, A-axis tilt angle, B-axis tilt angle, probe housing temperature, and lifting speed, and is output in a structured format.

[0008] The data prediction module is used to preload historical observation data collected by the original data acquisition module, judge potential deformation risks based on depth-time characteristics, and output the corresponding prediction risk label if the risk level of a certain depth segment exceeds the prediction risk threshold.

[0009] The disturbance identification module is used to identify abnormal signals caused by non-geological factors based on the current observation data collected by the original data acquisition module, and output abnormality marker information.

[0010] The resource status assessment module is used to detect the battery level, data cache capacity, communication status and remaining measurement time of the inclinometer during operation, and to generate resource constraint information based on resource usage rules;

[0011] The dynamic sampling adjustment module is used to comprehensively receive the predicted risk marker, anomaly marker information and resource constraint information, and generate sampling instructions according to preset sampling control rules. The sampling instructions include regular sampling, repeated sampling, encrypted sampling, delayed sampling or simplified sampling.

[0012] The sampling control execution module is used to control the real-time execution of sampling behavior according to the sampling instruction, and to attach a sampling behavior type identifier to each group of sampling data;

[0013] The data output module is used to record all observation data, prediction risk markers, anomaly marker information, resource constraint information, and sampling behavior type identifiers generated during the sampling process, and output structured data results for subsequent use.

[0014] A further technical improvement of the present invention is that: during the movement of the inclinometer probe along the inclinometer tube, the original data acquisition module collects the A-axis tilt angle, B-axis tilt angle, probe housing temperature and lifting speed values ​​of each measuring point in real time, and immediately performs multi-parameter joint fluctuation analysis on the current measuring point after acquisition to generate the corresponding confidence score;

[0015] The raw data acquisition module also pre-divides all measurement points into multiple fixed layered intervals according to depth, and dynamically updates the confidence statistics of each layered interval and calculates its consistency score during the sampling process of the probe into each layered interval.

[0016] When the confidence score of a certain measurement point is lower than the confidence threshold, a low confidence label is attached to that measurement point;

[0017] When the consistency score of a certain stratified interval is lower than the consistency score threshold, a low consistency label is added to the entire stratum.

[0018] Low confidence markers and low consistency markers are output along with the structured observation data.

[0019] A further technical improvement of the present invention is that: when the dynamic sampling adjustment module receives the observation data containing low consistency markers output by the original data acquisition module, it identifies the depth segment corresponding to the low consistency marker as a region with low signal consistency according to the preset sampling control rules, and directly generates an encrypted sampling command for the depth segment without relying on anomaly marker information or predicting risk markers.

[0020] A further technical improvement of the present invention is that: when the disturbance identification module receives the observation data output by the original data acquisition module, it first determines whether the current measuring point has been marked with a low consistency by the original data acquisition module;

[0021] If it belongs to the depth layer where the low consistency marker is located, the tilt angle change recognition threshold, peak disturbance recognition threshold and temperature drift offset recognition threshold are dynamically corrected according to the preset threshold adjustment rules before abnormal signal identification.

[0022] After completing the threshold correction, the A-axis inclination value, B-axis inclination value, probe housing temperature value and lifting speed value of the current measuring point are analyzed in real time to identify non-geological abnormal signals caused by local abnormal soil compression, probe operation disturbance or temperature change.

[0023] If a signal that meets the anomaly detection criteria is detected, the corresponding anomaly marker information is output.

[0024] If it does not belong to the depth layer where the low consistency marker is located, the default tilt change identification threshold, default spike disturbance identification threshold, and default temperature drift offset identification threshold are directly used to perform abnormal signal identification and output abnormal marker information.

[0025] Anomaly detection criteria include:

[0026] a1. If the absolute value of the change in the A-axis tilt angle or the absolute value of the change in the B-axis tilt angle between the current measuring point and its previous depth adjacent measuring point is greater than the preset tilt angle change identification threshold, it is determined to be an isolated tilt angle change anomaly.

[0027] a2. If the rate of change of the tilt angle of the A-axis or B-axis calculated based on the acquisition time interval and tilt angle difference between the current measuring point and the previous measuring point exceeds the preset peak disturbance identification threshold, it is determined to be a short-term peak disturbance anomaly.

[0028] a3. In an analysis window centered on the current measurement point and containing at least N consecutive measurement points, if the rate of change of the probe housing temperature exceeds the preset temperature change rate threshold, and the absolute value of the correlation coefficient between the A-axis or B-axis tilt sequence and the temperature sequence in the analysis window exceeds the preset correlation coefficient threshold, then it is determined to be an abnormal tilt drift caused by a sudden temperature change.

[0029] For a measurement point that meets any of the anomaly determination conditions a1, a2, or a3, anomaly marker information containing the anomaly type is generated and output.

[0030] A further technical improvement of the present invention is that: during the generation of sampling instructions, the dynamic sampling adjustment module performs data quality processing operations on the observation data of measurement points with attached low confidence indicators, including:

[0031] Based on the confidence score of the measurement point, the participation weight of the measurement point in the sampling instruction generation process is determined according to the preset mapping relationship between confidence and participation weight.

[0032] Alternatively, if the confidence score is lower than a preset rejection threshold, the observation data of the measurement point will be removed from the processing flow of the sampling instruction generation to avoid low-confidence data interfering with the sampling instruction.

[0033] A further technical improvement of the present invention lies in: a method for the resource status assessment module to generate resource constraint information, comprising:

[0034] The remaining battery power, remaining data buffer capacity, communication connection status, and remaining execution time of a single observation task are collected during the operation of the inclinometer.

[0035] The remaining battery power is compared with a preset power safety threshold, the remaining data cache capacity is compared with a preset cache capacity safety threshold, the communication connection status is compared with a preset communication stability threshold, and the remaining execution time of a single observation task is compared with a preset task time safety threshold to generate power status determination results, cache capacity status determination results, communication status determination results, and task time status determination results.

[0036] The determination result is processed according to resource restriction rules, which include: when the power status determination result shows that the remaining battery power is lower than the power safety threshold, resource constraint information containing the maximum allowed number of samplings is generated;

[0037] When the cache capacity status determination result indicates that the remaining data cache capacity is lower than the cache capacity safety threshold, resource constraint information containing sampling data structure compression parameters or sampling interval limits is generated.

[0038] When the communication status determination result indicates that the communication connection status is lower than the communication stability threshold, resource constraint information containing abnormal resampling restrictions or data transmission frequency adjustments is generated.

[0039] When the task time status determination result indicates that the remaining execution time is lower than the task time safety threshold, resource constraint information including sampling priority adjustment is generated.

[0040] Resource constraint information is output to the dynamic sampling adjustment module.

[0041] A further technical improvement of the present invention is that the sampling control rules include:

[0042] z1: If the current depth location contains a predicted risk marker, then generate an encrypted sampling instruction first;

[0043] z2: If the current measurement point contains abnormal marker information, generate a repeat sampling instruction;

[0044] z3: If the resource constraint information indicates that the system resources are below the stress threshold, then a simplified sampling instruction is generated for the measurement points that do not meet rules z1 and z2;

[0045] If none of the above rules apply, a standard sampling instruction will be generated.

[0046] Among them, rule z1 has a higher priority than rule z2, and rule z2 has a higher priority than rule z3.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] This invention integrates multiple functional modules such as historical data prediction, current observation data analysis, and resource status perception. Through information linkage between modules, it can achieve full-process adaptive control of sampling behavior, with higher responsiveness and sampling efficiency, and is suitable for various dynamic monitoring environments.

[0049] Furthermore, the system of this invention introduces multiple sampling decision factors, such as predicted risk markers, anomaly markers, and resource constraint information, and achieves unified strategy fusion in the dynamic sampling adjustment module. It can flexibly generate various sampling instructions, such as encrypted sampling, repeated sampling, simplified sampling, or regular sampling, based on the risk level, signal stability, and current resource conditions of different areas. This mechanism significantly improves the sampling density in critical areas while avoiding resource waste, exhibiting good adaptability and scalability.

[0050] On the other hand, by establishing a priority structure for sampling control rules, such as risk first, anomaly second, and resource constraints last, the present invention achieves a comprehensive balance of sampling behavior under multiple objectives and constraints. In addition, the system also adds a sampling behavior type identifier to each group of sampling data, which facilitates subsequent data analysis, quality control and scheduling review, and provides a complete data foundation and decision support for the intelligent utilization and automatic evaluation of inclination data. Attached Figure Description

[0051] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0052] Figure 1This is a schematic diagram of the system logic of the present invention. Detailed Implementation

[0053] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0054] Example 1

[0055] An adaptive data sampling rate adjustment system for an inclinometer includes a raw data acquisition module, a data prediction module, a disturbance identification module, a resource status assessment module, a dynamic sampling adjustment module, a sampling control execution module, and a data output module.

[0056] The raw data acquisition module is used to collect observation data of multiple measuring points in real time according to the preset sampling interval as the inclinometer probe moves along the inclinometer tube. The observation data includes the measuring point depth, A-axis inclination angle, B-axis inclination angle, probe housing temperature, and lifting speed, and is output in a structured format.

[0057] During the movement of the inclinometer probe along the inclinometer tube, the raw data acquisition module collects the A-axis tilt angle, B-axis tilt angle, probe housing temperature, and lifting speed values ​​of each measuring point in real time. Immediately after the acquisition, it performs multi-parameter joint fluctuation analysis on the current measuring point and generates the corresponding confidence score.

[0058] The raw data acquisition module also pre-divides all measurement points into multiple fixed layered intervals according to depth, and dynamically updates the confidence statistics of each layered interval and calculates its consistency score during the sampling process of the probe into each layered interval.

[0059] When the confidence score of a certain measurement point is lower than the confidence threshold, a low confidence label is attached to that measurement point;

[0060] When the consistency score of a certain stratified interval is lower than the consistency score threshold, a low consistency label is added to the entire stratum.

[0061] Low confidence markers and low consistency markers are output along with the structured observation data.

[0062] Specifically, the A-axis and B-axis tilt angles are obtained in real time through a dual-axis tilt sensor installed inside the inclinometer probe. This sensor uses the probe's central axis as a reference frame and has sensing units arranged along two orthogonal planes perpendicular to the axial direction, enabling real-time detection of changes in the probe's tilt angle along the A-axis and B-axis. Every sampling time interval, the sensor outputs two tilt angle measurements for the current measurement point, which are then used as the A-axis and B-axis tilt angle values, respectively.

[0063] The probe housing temperature is acquired in real time by a high-sensitivity temperature sensor mounted on the surface of the probe housing or in close contact with the inner metal housing. This temperature sensor can be a thermistor or a digital temperature sensing chip, and its sampling period is synchronized with the tilt sensor. During the lowering or raising of the probe, the temperature sensor continuously collects temperature data from the current environment and outputs the temperature value corresponding to the current measuring point in real time, which is used as the probe housing temperature value at that point.

[0064] The lift velocity is calculated in real time by combining the depth information from the inclinometer probe with the sampling time interval. Specifically, at each sampling point, the system records the current depth value and sampling timestamp of the probe; when the probe moves to the next sampling point, the system synchronously records a new depth value and timestamp. Dividing the depth difference between two adjacent samples by the time interval yields the depth displacement rate per unit time, which is the lift velocity.

[0065] When calculating the confidence score, for the current measuring point, the numerical differences between the measuring point and its adjacent measuring points are calculated for each of the four physical quantity dimensions: A-axis tilt angle, B-axis tilt angle, temperature, and lifting speed. For each physical quantity, the difference between the value at this measuring point and the value at the previous measuring point is obtained and recorded as the "current difference". Simultaneously, the mean and standard deviation of this physical quantity are calculated among multiple measuring points within a certain depth range (e.g., 1 meter before and after) centered on the current measuring point.

[0066] Subsequently, the aforementioned "current difference" is normalized to the mean and standard deviation of its corresponding physical quantity within that local region, yielding the normalized deviation value for that physical quantity. The normalized deviation value is calculated by subtracting the local mean from the current difference and then dividing by the local standard deviation. This operation allows fluctuations between different physical quantities to be compared on a uniform numerical scale.

[0067] The normalized deviations of the A-axis tilt angle, B-axis tilt angle, probe housing temperature, and lifting speed are weighted and summed to obtain the comprehensive disturbance value for the measurement point. Different weighting coefficients can be set for different physical quantities during the weighting process. For example, the tilt angle can have a higher weight than temperature and speed. The specific weight ratio can be set according to actual application requirements, such as tilt angle 0.3, temperature 0.2, speed 0.2, with a total weight of 1.

[0068] The aggregated perturbation value is then converted into a confidence score using a mapping model. This mapping model is a monotonically decreasing function, designed using the tail mapping method of the standard normal distribution: the larger the perturbation value, the lower the corresponding confidence score. Specifically, on a pre-defined standard normal distribution curve, the cumulative probability corresponding to the aggregated perturbation value is found, and the confidence score is calculated as "1 minus this probability value". This method allows for a mapping of confidence levels between 0 and 1, with scores closer to 1 indicating higher reliability.

[0069] Furthermore, in some applications, to avoid misjudgments caused by data coupling due to temperature fluctuations, the correlation between the tilt angle change sequence and the temperature change sequence at the current measuring point can be analyzed. If the correlation coefficient exceeds a specific threshold (e.g., 0.8), it is determined that there is co-variation of temperature-driven signals. In this case, the tilt angle deviation value can be weighted down or the perturbation value can be corrected to reduce the confidence reduction caused by physical coupling.

[0070] The fixed-layer interval division operation includes: after the original data acquisition is completed, the system spatially divides the overall depth range of the inclinometer tube based on the depth information of each measuring point. This implementation uses a fixed division length to evenly divide the total depth range of the inclinometer tube into multiple fixed-layer intervals. Each layer interval has the same depth span; for example, if the division is done in 1-meter units, then 0–1 meter is the first layer interval, 1–2 meters is the second layer interval, and so on, until the entire measurement depth range is covered.

[0071] To improve the continuity of data analysis at the boundaries, a sliding window mechanism can be introduced during the partitioning process. For example, by overlapping upwards by 0.5 meters on top of each 1-meter layer, the second layer interval would range from 0.5 to 1.5 meters, and the third layer interval from 1 to 2 meters, forming a certain overlapping coverage area. This overlapping strategy can enhance the stability of consistency calculations between layers and avoid scoring bias caused by sparse distribution of boundary measurement points.

[0072] Perform consistency score calculation for stratified intervals, including:

[0073] Once the probe enters any fixed stratified interval, the system statistically analyzes the confidence scores of all measurement points within that interval and calculates the consistency score for that interval. The consistency score reflects the overall quality and volatility of the data within that interval, and its calculation includes the following three indicators:

[0074] Average confidence index: The arithmetic mean of the confidence scores of all measurement points within the stratified interval. The higher the value, the higher the overall data quality of the area.

[0075] Low confidence ratio indicator: The ratio of the number of measurement points with confidence scores below the confidence threshold (e.g., 0.4) to the total number of measurement points in this stratum. The lower the ratio, the fewer abnormal measurement points there are in this area.

[0076] Confidence score standard deviation index: Calculate the standard deviation of all confidence scores. The smaller the standard deviation, the more uniform the data quality among the measurement points.

[0077] The three indicators mentioned above correspond to the dimensions of "quality level," "abnormality rate," and "uniformity," respectively. A weighted combination of these three indicators yields a consistency score. The calculation logic is: Consistency Score = Average Confidence Value × (1 - Low Confidence Rate) × (1 - Normalized Standard Deviation). The normalized standard deviation is the ratio between the current standard deviation and the global maximum possible standard deviation, used to standardize the range of dimensions. The final score is controlled within the range of 0 to 1; a higher score indicates better consistency of the data within that stratified interval.

[0078] Finally, based on the above analysis results, the system performs structured labeling on the observation data:

[0079] When the confidence score of a certain measurement point is lower than the confidence threshold, a low confidence label is attached to that measurement point.

[0080] When the consistency score of a certain hierarchical interval is lower than the consistency score threshold, a low consistency mark is added to that hierarchical interval (for example, an identifier for that interval is added to the system record).

[0081] The data prediction module is used to preload historical observation data collected by the original data acquisition module, and judge potential deformation risks based on depth-time characteristics. If the risk level of a certain depth segment exceeds the prediction risk threshold, the corresponding prediction risk label is output.

[0082] Specifically, the historical observation data includes at least the A-axis and B-axis inclination values ​​at multiple measuring points at different depths, along with the corresponding acquisition timestamps and measuring point depth information. The data can be read from a local cache or imported from an external storage device and is structured within the data prediction module to ensure that multiple time-series data at the same depth can be indexed and accessed.

[0083] The data prediction module calculates the dip angle change rate over time for each depth segment based on depth and time information from historical observation data, using this as a fundamental indicator to measure potential deformation risk. In this embodiment, the dip angle change rate over time can be obtained as follows: For each depth segment, select the two most recent observation data records at that depth, extract the corresponding A-axis dip angle values ​​and B-axis dip angle values ​​respectively, calculate their absolute values ​​of change, and use the acquisition time interval as the denominator to obtain the dip angle change rate per unit time. Specifically, if the A-axis dip angle values ​​acquired twice at a certain depth segment are θ1 and θ2, and the acquisition times are t1 and t2 respectively, then the A-axis dip angle change rate for that depth segment is |θ2-θ1| / |t2-t1|, and the B-axis dip angle is obtained similarly.

[0084] The data prediction module compares the aforementioned rate of change results with a preset prediction risk threshold. In this embodiment, the prediction risk threshold for the rate of change of tilt angle can be set to 0.05 degrees per hour, or it can be set according to engineering requirements. When the rate of change of tilt angle in any axial direction at a certain depth segment exceeds this threshold, it is considered that there is a potential deformation risk in that depth segment.

[0085] After risk assessment, the data prediction module outputs a predicted risk marker for depth segments where the risk exceeds the limit. This marker can be represented by structured fields, including but not limited to "depth segment number," "risk assessment time," "exceeding axis," and "rate of change value." In this embodiment, the predicted risk marker can be output to the subsequent processing module in real time, or pre-stored in the observation data structure and loaded and invoked along with subsequent data acquisition tasks to assist in sampling instruction generation or other decision-making processes.

[0086] The disturbance identification module is used to identify abnormal signals caused by non-geological factors based on the current observation data collected by the raw data acquisition module, and output abnormality marker information;

[0087] When the disturbance identification module receives the observation data output by the raw data acquisition module, it first determines whether the current measuring point has been marked with a low consistency by the raw data acquisition module.

[0088] If it belongs to the depth layer where the low consistency marker is located, the tilt angle change recognition threshold, peak disturbance recognition threshold and temperature drift offset recognition threshold are dynamically corrected according to the preset threshold adjustment rules before abnormal signal identification.

[0089] After completing the threshold correction, the A-axis inclination value, B-axis inclination value, probe housing temperature value and lifting speed value of the current measuring point are analyzed in real time to identify non-geological abnormal signals caused by local abnormal soil compression, probe operation disturbance or temperature change.

[0090] If a signal that meets the anomaly detection criteria is detected, the corresponding anomaly marker information is output.

[0091] If it does not belong to the depth layer where the low consistency marker is located, the default tilt change identification threshold, default spike disturbance identification threshold, and default temperature drift offset identification threshold are directly used to perform abnormal signal identification and output abnormal marker information.

[0092] Anomaly detection criteria include:

[0093] a1. If the absolute value of the change in the A-axis tilt angle or the absolute value of the change in the B-axis tilt angle between the current measuring point and its previous depth adjacent measuring point is greater than the preset tilt angle change identification threshold, it is determined to be an isolated tilt angle change anomaly.

[0094] a2. If the rate of change of the tilt angle of the A-axis or B-axis calculated based on the acquisition time interval and tilt angle difference between the current measuring point and the previous measuring point exceeds the preset peak disturbance identification threshold, it is determined to be a short-term peak disturbance anomaly.

[0095] a3. In an analysis window centered on the current measurement point and containing at least N consecutive measurement points, if the rate of change of the probe housing temperature exceeds the preset temperature change rate threshold, and the absolute value of the correlation coefficient between the A-axis or B-axis tilt sequence and the temperature sequence in the window exceeds the preset correlation coefficient threshold, then it is determined to be an abnormal tilt drift caused by a sudden temperature change.

[0096] For a measurement point that meets any of the anomaly determination conditions a1, a2, or a3, anomaly marker information containing the anomaly type is generated and output.

[0097] Specifically, in this embodiment, after receiving the structured observation data output by the raw data acquisition module, the disturbance identification module first identifies whether the current sampling point has a low consistency marker. If the identification result indicates that the current sampling point belongs to a stratified interval with a low consistency marker, it means that the historical data consistency of that depth segment is poor, and there are potential abnormal signals or data interference sources. To adapt to the identification needs under this special data background, the disturbance identification module dynamically corrects multiple judgment threshold parameters used for subsequent anomaly identification according to preset threshold adjustment rules.

[0098] The threshold adjustment rule can be based on the consistency score value within the stratified interval, and the adjustment direction and magnitude can be determined according to a nonlinear response function. Specifically, when the consistency score value is in the low to medium range (e.g., between 0.4 and 0.6), the system tends to lower the anomaly identification threshold to improve identification sensitivity and ensure that no possible real anomaly signals are missed. Conversely, when the consistency score decreases further, indicating that the background noise level of the data may be too high, the system can appropriately raise the identification threshold to reduce the risk of false positives and improve the reliability of the output anomaly labels. This strategy ensures that the anomaly identification mechanism has adjustable adaptability and robustness under different data consistency conditions.

[0099] After the threshold adjustment is completed, the disturbance identification module performs multi-type anomaly identification operations for the current measurement point. Anomaly identification includes the following three modes:

[0100] To identify isolated abrupt changes in dip angle, the disturbance identification module acquires the A-axis and B-axis dip angle values ​​of the current measuring point and performs difference calculations with the corresponding dip angle values ​​of adjacent measuring points at the previous depth, calculating the absolute value of the dip angle change in both directions. If the absolute value of the change in either direction is greater than the dynamically adjusted dip angle abrupt change identification threshold, it can be determined that the measuring point has an isolated abrupt change in dip angle.

[0101] To identify short-term spike disturbance anomalies, the disturbance identification module calculates the rate of change of tilt angle along the A-axis and B-axis directions based on the acquisition time interval and tilt angle change between the current measuring point and the previous measuring point. Specifically, the tilt angle difference between the two measuring points is divided by the sampling time interval. If the rate of change in either direction exceeds the dynamically adjusted spike disturbance identification threshold, the measuring point is considered to have a short-term spike disturbance anomaly.

[0102] To identify tilt drift anomalies caused by sudden temperature changes, the disturbance identification module constructs an analysis window centered on the current measuring point, encompassing the current measuring point and several measuring points above and below it. The window length is no less than N sampling points. Within this analysis window, a linear regression operation is performed on the probe housing temperature value sequence and the corresponding sampling time sequence, using the regression slope as the temperature change rate of the current segment. Simultaneously, the Pearson correlation coefficient between the A-axis tilt angle value sequence and the temperature sequence within the window is calculated. If the temperature change rate exceeds a set threshold, and the absolute value of the correlation coefficient exceeds the temperature drift identification correlation threshold, it is considered that the current measuring point may be affected by a sudden temperature change, resulting in tilt drift anomalies.

[0103] If any of the above anomaly identification conditions are met, the disturbance identification module will generate an anomaly marker containing anomaly type and location information, and bind the anomaly marker with the structured observation data of the corresponding measurement point for output, which can be used as a reference by the subsequent data analysis and processing or sampling instruction generation module.

[0104] If the current measurement point does not belong to the stratification interval where the low consistency marker is located, the disturbance identification module directly uses the default threshold configuration, performs the three types of anomaly identification operations according to the same judgment logic described above, and outputs the corresponding anomaly marker information. This mechanism ensures that the system maintains efficient detection capability under standard conditions and has adjustable sensitivity capability in complex or unstable data areas, improving the overall stability and practicality of the identification.

[0105] The resource status assessment module is used to detect the battery level, data cache capacity, communication status and remaining measurement time of the inclinometer during operation, and to generate resource constraint information based on resource usage rules;

[0106] The method for generating resource constraint information by the resource status assessment module includes:

[0107] The remaining battery power, remaining data buffer capacity, communication connection status, and remaining execution time of a single observation task are collected during the operation of the inclinometer.

[0108] The remaining battery power is compared with a preset power safety threshold, the remaining data cache capacity is compared with a preset cache capacity safety threshold, the communication connection status is compared with a preset communication stability threshold, and the remaining execution time of a single observation task is compared with a preset task time safety threshold to generate power status determination results, cache capacity status determination results, communication status determination results, and task time status determination results.

[0109] The judgment result is processed according to resource restriction rules, which include: when the power status judgment result shows that the remaining battery power is lower than the power safety threshold, resource constraint information containing the maximum allowed number of samplings is generated;

[0110] When the cache capacity status determination result indicates that the remaining data cache capacity is lower than the cache capacity safety threshold, resource constraint information containing sampling data structure compression parameters or sampling interval limits is generated.

[0111] When the communication status determination result indicates that the communication connection status is lower than the communication stability threshold, resource constraint information containing abnormal resampling restrictions or data transmission frequency adjustments is generated.

[0112] When the task time status determination result indicates that the remaining execution time is lower than the task time safety threshold, resource constraint information including sampling priority adjustment is generated.

[0113] Resource constraint information is output to the dynamic sampling adjustment module.

[0114] Specifically, after the inclinometer enters the observation operation state, the resource status assessment module periodically or event-triggeredly collects key resource indicator data during the inclinometer's operation, including: current remaining battery power (in percentage or absolute milliampere-hours), remaining data cache capacity (in bytes or number of storage blocks), communication connection status (quantified by signal strength, packet loss rate, or stability score), and remaining execution time for a single observation task (estimated by observation progress and estimated total time). This data collection process covers four aspects: power supply, storage, communication, and task scheduling, and is the foundation for building resource awareness capabilities.

[0115] After collecting the above four resource indicators, the resource status assessment module compares them sequentially with their corresponding preset safety thresholds to generate four types of resource status judgment results. Specifically: the current remaining battery power is compared with the power safety threshold (e.g., 20%); if it is lower than this threshold, the power status judgment result is "tight"; the current remaining cache capacity is compared with the cache capacity safety threshold (e.g., 10MB); if it is lower than this threshold, the cache capacity status judgment result is "tight"; the current communication connection status is compared with the communication stability threshold (e.g., signal quality score less than 60 points); if it is lower than this threshold, the communication status judgment result is "unstable"; and the remaining execution time is compared with the task time safety threshold (e.g., 5 minutes); if it is lower than this threshold, the task time status judgment result is "insufficient". These judgment results form a set of standardized resource usage status assessment information for subsequent processing.

[0116] The resource status assessment module further processes the above judgment results according to the resource restriction rules, and generates resource constraint information that can be directly called by the sampling control.

[0117] Resource constraint rules include: when the battery status determination result indicates that the remaining battery power is lower than the battery safety threshold, resource constraint information containing the "maximum allowed sampling number limit" parameter is generated, such as limiting the subsequent sampling number to no more than 50 times; when the cache capacity status determination result indicates that the remaining data cache capacity is lower than the cache capacity safety threshold, resource constraint information containing the "sampling data structure compression parameter" or "sampling interval limit" field is generated, such as setting the compression level of optional fields in structured observation data to 2, or adjusting the minimum sampling interval to 0.8 meters; when the communication status determination result indicates that the communication connection status is lower than the communication stability threshold, resource constraint information containing the "abnormal resampling restriction" or "data transmission frequency adjustment" field is generated, such as prohibiting more than 3 consecutive abnormal sampling retries, or reducing the transmission frequency to half of the original frequency; when the task time status determination result indicates that the remaining execution time is lower than the task time safety threshold, resource constraint information containing the "sampling priority adjustment" parameter is generated, such as increasing the sampling priority of high confidence areas and high-risk prediction areas by 1 level.

[0118] Finally, after extracting and structuring the aforementioned constraints, the resource status assessment module outputs the generated resource constraint information and synchronizes it to the dynamic sampling adjustment module. This information is then used by the dynamic sampling adjustment module to match sampling strategies and control the execution of sampling instructions when generating them, taking into account the current state of the measurement point, anomaly markers, and predicted risk levels. This implementation method enables dynamic adjustment of sampling behavior and intelligent resource control under various complex operating environments, such as critical power levels, unstable communication, or limited buffering. This extends the measurement cycle and enhances system resilience and adaptability while maintaining the integrity of critical data.

[0119] The dynamic sampling adjustment module is used to comprehensively receive predicted risk markers, anomaly markers and resource constraint information, and generate sampling instructions according to preset sampling control rules. The sampling instructions include regular sampling, repeated sampling, encrypted sampling, delayed sampling or simplified sampling.

[0120] During the generation of sampling instructions, the dynamic sampling adjustment module performs data quality processing operations on the observation data of measurement points that are marked with low confidence, including:

[0121] Based on the confidence score of the measurement point, the participation weight of the measurement point in the sampling instruction generation process is determined according to the preset mapping relationship between confidence and participation weight.

[0122] Alternatively, if the confidence score is lower than the preset rejection threshold, the observation data of that measurement point can be removed from the processing flow of the sampling instruction generation to avoid low-confidence data interfering with the sampling instruction.

[0123] In this embodiment, in order to improve the accuracy and stability of the sampling instruction generation process, the dynamic sampling adjustment module performs the following data quality processing operation on the observation data of the measuring points with low confidence labels during the processing of the observation data output by the original data acquisition module. This operation is closely integrated with the sampling instruction generation process to improve the stability and adaptability of the overall sampling control.

[0124] Specifically, after the dynamic sampling adjustment module receives the structured observation data containing low-confidence indicators, it extracts the corresponding confidence score for each measurement point identified as having low confidence. The confidence score is a continuous value between 0 and 1. The closer the value is to 1, the more stable and reliable the observation data at that measurement point is; conversely, a lower value indicates that the data at that measurement point has large fluctuations, abnormal deviations, or poor consistency in physical quantities.

[0125] After obtaining the confidence score, the dynamic sampling adjustment module determines the actual participation weight of the observation data at the measurement point in the sampling instruction generation process based on the preset mapping relationship between confidence and participation weight. In this embodiment, the mapping relationship is constructed using a linear piecewise function. For example, when the confidence score is between 0.9 and 1.0, the participation weight is set to 1.0; when the confidence score is between 0.7 and 0.9, the participation weight is 0.8; when the confidence score is between 0.5 and 0.7, the participation weight is 0.5; and when the confidence score is below 0.5, the participation weight decreases to 0.2 or even lower.

[0126] The participation weight plays several key roles in the sampling instruction generation process, including: (1) in the identification of encrypted sampling depth segments, it acts as a weight factor in conjunction with risk markers and anomaly markers, participating in the calculation of importance scores; (2) in the dynamic adjustment logic of sampling frequency, it is used to construct a depth segment priority ranking mechanism based on observation quality; (3) in resource-constrained situations, it serves as a decision reference for simplifying sampling, determining whether to implement skip sampling or merge sampling strategies for the measurement point. Through the above multi-point fusion, the participation weight effectively guides the sampling strategy to focus on measurement point areas with high data quality and high sampling value.

[0127] When the confidence score of a certain measuring point is lower than the preset rejection threshold (e.g., set to 0.3), the dynamic sampling adjustment module determines that the observation data of that measuring point does not have the basic data quality to participate in the decision-making process, and directly removes the data of that measuring point from the sampling instruction generation and processing flow. The rejection process includes deleting the depth information of the measuring point from the candidate sampling target sequence, and masking its dip angle change trend, temperature change and adjacent disturbance characteristics during statistical analysis, so as to avoid unreliable data from interfering with or misleading the decision results.

[0128] Finally, after completing the aforementioned weight allocation or data removal processes, the dynamic sampling adjustment module, based on the valid participation data from all current measuring points, executes strategies such as sampling priority sorting, encrypted sampling area identification, and resource quota matching to ultimately generate sampling instructions. These instructions are then distributed to the actual sampling control process through the sampling control execution module. This mechanism, while ensuring the quality-based utilization of observation data, improves the responsiveness, accuracy, and resource allocation efficiency of the sampling instruction generation process, significantly enhancing the adaptive capability of inclinometer sampling control.

[0129] The sampling control execution module is used to control the real-time execution of sampling behavior according to sampling instructions, and to attach a sampling behavior type identifier to each group of sampled data;

[0130] Sampling control rules include:

[0131] z1: If the current depth location contains a predicted risk marker, then generate an encrypted sampling instruction first;

[0132] z2: If the current measurement point contains abnormal marker information, generate a repeat sampling instruction;

[0133] z3: If the resource constraint information indicates that the system resources are below the stress threshold, then a simplified sampling instruction is generated for the measurement points that do not meet rules z1 and z2;

[0134] If none of the above rules apply, a standard sampling instruction will be generated.

[0135] Among them, rule z1 has a higher priority than rule z2, and rule z2 has a higher priority than rule z3.

[0136] The data output module is used to record all observation data, prediction risk markers, anomaly markers, resource constraint information, and sampling behavior type identifiers generated during the sampling process, and output structured data results for subsequent use.

[0137] Example 2

[0138] An adaptive data sampling rate adjustment system for an inclinometer includes a raw data acquisition module, a data prediction module, a disturbance identification module, a resource status assessment module, a dynamic sampling adjustment module, a sampling control execution module, and a data output module.

[0139] The raw data acquisition module is used to collect observation data of multiple measuring points in real time according to the preset sampling interval as the inclinometer probe moves along the inclinometer tube. The observation data includes the measuring point depth, A-axis inclination angle, B-axis inclination angle, probe housing temperature, and lifting speed, and is output in a structured format.

[0140] During the movement of the inclinometer probe along the inclinometer tube, the raw data acquisition module collects the A-axis tilt angle, B-axis tilt angle, probe housing temperature, and lifting speed values ​​of each measuring point in real time. Immediately after the acquisition, it performs multi-parameter joint fluctuation analysis on the current measuring point and generates the corresponding confidence score.

[0141] The raw data acquisition module also pre-divides all measurement points into multiple fixed layered intervals according to depth, and dynamically updates the confidence statistics of each layered interval and calculates its consistency score during the sampling process of the probe into each layered interval.

[0142] When the confidence score of a certain measurement point is lower than the confidence threshold, a low confidence label is attached to that measurement point;

[0143] When the consistency score of a certain stratified interval is lower than the consistency score threshold, a low consistency label is added to the entire stratum.

[0144] Low confidence markers and low consistency markers are output along with the structured observation data.

[0145] The data prediction module is used to preload historical observation data collected by the original data acquisition module, and judge potential deformation risks based on depth-time characteristics. If the risk level of a certain depth segment exceeds the prediction risk threshold, the corresponding prediction risk label is output.

[0146] The disturbance identification module is used to identify abnormal signals caused by non-geological factors based on the current observation data collected by the raw data acquisition module, and output abnormality marker information;

[0147] The resource status assessment module is used to detect the battery level, data cache capacity, communication status and remaining measurement time of the inclinometer during operation, and to generate resource constraint information based on resource usage rules;

[0148] The dynamic sampling adjustment module is used to comprehensively receive predicted risk markers, anomaly markers, and resource constraint information;

[0149] Compared to Example 1, in Example 2, when the dynamic sampling adjustment module receives observation data containing low consistency markers from the original data acquisition module, it identifies the depth segment corresponding to the low consistency marker as a region with low signal consistency according to the preset sampling control rules, and directly generates encrypted sampling instructions for the depth segment without relying on anomaly marker information or predictive risk markers.

[0150] The sampling control execution module is used to control the real-time execution of sampling behavior according to sampling instructions, and to attach a sampling behavior type identifier to each group of sampled data;

[0151] The data output module is used to record all observation data, prediction risk markers, anomaly markers, resource constraint information, and sampling behavior type identifiers generated during the sampling process, and output structured data results for subsequent use.

[0152] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0153] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A data sampling rate adaptive adjustment system for an inclinometer, characterized in that, include: The raw data acquisition module is used to collect observation data of multiple measuring points in real time according to a preset sampling interval as the inclinometer probe moves along the inclinometer tube, and output the data in a structured format. The data prediction module is used to preload historical observation data collected by the original data acquisition module, judge potential deformation risks based on the correlation characteristics between depth and time, and output the predicted risk label of the depth segment if the risk level of a certain depth segment exceeds the predicted risk threshold. The disturbance identification module, based on the current observation data, identifies non-geological anomaly signals caused by local abnormal soil compression, probe operation disturbance, or temperature change, and outputs anomaly marker information; The resource status assessment module is used to detect the battery level, data cache capacity, communication status and remaining measurement time of the inclinometer during operation, and generate resource constraint information based on resource usage rules; The dynamic sampling adjustment module is used to comprehensively receive predicted risk markers, anomaly markers, and resource constraint information, and dynamically generate sampling instructions based on preset sampling control rules. Sampling control rules include: z1: If the current depth location contains a predicted risk marker, then generate an encrypted sampling instruction first; z2: If the current measurement point contains abnormal marker information, generate a repeat sampling instruction; z3: If the resource constraint information indicates that the system resources are below the stress threshold, then a simplified sampling instruction is generated for the measurement points that do not meet rules z1 and z2; If none of the above rules apply, a standard sampling instruction will be generated. Among them, rule z1 has a higher priority than rule z2, and rule z2 has a higher priority than rule z3; The sampling control execution module is used to control the real-time execution of sampling behavior according to the sampling instructions, and to attach a sampling behavior type identifier to each group of sampling data.

2. The data sampling rate adaptive adjustment system for an inclinometer according to claim 1, characterized in that, The observation data includes the depth of the measuring point, the tilt angle of the A-axis, the tilt angle of the B-axis, the temperature of the probe housing, and the lifting speed.

3. The data sampling rate adaptive adjustment system for an inclinometer according to claim 2, characterized in that, During the movement of the inclinometer probe along the inclinometer tube, the raw data acquisition module collects the A-axis tilt angle, B-axis tilt angle, probe housing temperature, and lifting speed values ​​of each measuring point in real time. Immediately after the acquisition, it performs multi-parameter joint fluctuation analysis on the current measuring point and generates the corresponding confidence score. The raw data acquisition module also pre-divides all measurement points into multiple fixed layered intervals according to depth, and dynamically updates the confidence statistics of each layered interval and calculates its consistency score during the sampling process of the probe into each layered interval. When the confidence score of a certain measurement point is lower than the confidence threshold, a low confidence label is attached to that measurement point; When the consistency score of a certain stratified interval is lower than the consistency score threshold, a low consistency label is added to the entire stratum. Low confidence markers and low consistency markers are output along with the structured observation data.

4. The data sampling rate adaptive adjustment system for an inclinometer according to claim 3, characterized in that, When the dynamic sampling adjustment module receives observation data containing low consistency markers from the original data acquisition module, it identifies the depth segment corresponding to the low consistency marker as a region with low signal consistency according to the preset sampling control rules, and directly generates encrypted sampling instructions for the depth segment without relying on anomaly marker information or predictive risk markers.

5. The data sampling rate adaptive adjustment system for an inclinometer according to claim 3, characterized in that, When receiving the observation data output by the original data acquisition module, the disturbance identification module first determines whether the current measuring point has been marked with a low consistency by the original data acquisition module. If it belongs to the depth layer where the low consistency marker is located, the tilt angle change recognition threshold, peak disturbance recognition threshold and temperature drift offset recognition threshold are dynamically corrected according to the preset threshold adjustment rules before abnormal signal identification. After completing the threshold correction, the A-axis inclination value, B-axis inclination value, probe housing temperature value and lifting speed value of the current measuring point are analyzed in real time to identify non-geological abnormal signals caused by local abnormal soil compression, probe operation disturbance or temperature change. If a signal that meets the anomaly detection criteria is detected, the corresponding anomaly marker information is output. If it does not belong to the depth layer where the low consistency marker is located, the default tilt angle change recognition threshold, the default peak disturbance recognition threshold, and the default temperature drift offset recognition threshold are used to perform abnormal signal recognition and output abnormal marker information.

6. The data sampling rate adaptive adjustment system for an inclinometer according to claim 5, characterized in that, Anomaly detection criteria include: a1. If the absolute value of the change in the A-axis tilt angle or the absolute value of the change in the B-axis tilt angle between the current measuring point and its previous depth adjacent measuring point is greater than the preset tilt angle change identification threshold, it is determined to be an isolated tilt angle change anomaly. a2. If the rate of change of the tilt angle of the A-axis or B-axis calculated based on the acquisition time interval and tilt angle difference between the current measuring point and the previous measuring point exceeds the preset peak disturbance identification threshold, it is determined to be a short-term peak disturbance anomaly. a3. In an analysis window centered on the current measurement point and containing at least N consecutive measurement points, if the rate of change of the probe housing temperature exceeds the preset temperature change rate threshold, and the absolute value of the correlation coefficient between the A-axis or B-axis tilt sequence and the temperature sequence in the analysis window exceeds the preset correlation coefficient threshold, then it is determined to be an abnormal tilt drift caused by a sudden temperature change. For a measurement point that meets any of the anomaly determination conditions a1, a2, or a3, anomaly marker information containing the anomaly type is generated and output.

7. A data sampling rate adaptive adjustment system for an inclinometer according to claim 5, characterized in that, During the generation of sampling instructions, the dynamic sampling adjustment module performs data quality processing operations on the observation data of measurement points with low confidence markers, including: Based on the confidence score of the measurement point, the participation weight of the measurement point in the sampling instruction generation process is determined according to the preset mapping relationship between confidence and participation weight. Alternatively, if the confidence score is lower than the preset rejection threshold, the observation data of that measurement point can be removed from the processing flow of the sampling instruction generation to avoid low-confidence data interfering with the sampling instruction.

8. The data sampling rate adaptive adjustment system for an inclinometer according to claim 1, characterized in that, The method for generating resource constraint information by the resource status assessment module includes: Collect data on the remaining battery power, remaining data buffer capacity, communication connection status, and remaining execution time of a single observation task during the operation of the inclinometer. The remaining battery power is compared with a preset power safety threshold, the remaining data cache capacity is compared with a preset cache capacity safety threshold, the communication connection status is compared with a preset communication stability threshold, and the remaining execution time of a single observation task is compared with a preset task time safety threshold to generate power status determination results, cache capacity status determination results, communication status determination results, and task time status determination results. The judgment result is processed according to resource restriction rules, which include: when the power status judgment result shows that the remaining battery power is lower than the power safety threshold, resource constraint information containing the maximum allowed number of samplings is generated; When the cache capacity status determination result indicates that the remaining data cache capacity is lower than the cache capacity safety threshold, resource constraint information containing sampling data structure compression parameters or sampling interval limits is generated. When the communication status determination result indicates that the communication connection status is lower than the communication stability threshold, resource constraint information containing abnormal resampling restrictions or data transmission frequency adjustments is generated. When the task time status determination result indicates that the remaining execution time is lower than the task time safety threshold, resource constraint information including sampling priority adjustment is generated.

Images