A large floating disc multi-sensor inclination fusion monitoring and dynamic early warning method

Abstract

The application relates to the field of instruments and intelligent monitoring technology, and discloses a large floating disc multi-sensor inclination fusion monitoring and dynamic early warning method.The method comprises the following steps: arranging high-precision double-axis inclination sensors on a floating disc in a grid manner and realizing microsecond-level time synchronization; constructing a three-dimensional inclination field model; performing abnormal elimination, temperature drift compensation and time alignment on original data; performing spatial consistency verification and radial basis function repair based on structural continuity; introducing an adaptive weighted fusion strategy to calculate an overall equivalent inclination and a maximum local inclination; combining with environmental disturbance factors to establish a dynamic evolution model, and setting three-level adjustable early warning thresholds.Through the technical scheme, the monitoring precision, robustness and early warning reliability are improved, and the operation safety of a large oil tank is ensured.

Description

Technical Field

[0001] This invention belongs to the field of instrumentation and intelligent monitoring technology, specifically relating to a method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating platforms. Background Technology

[0002] With the widespread application of large oil storage tanks in energy storage and chemical production, the floating roof, as a key component of internal floating roof tanks, directly affects the safety, sealing, and evaporation loss control of the tank. Floating roofs are constantly positioned above flammable and explosive media, and are highly susceptible to tilting or even jamming due to multiple factors such as liquid level fluctuations, wind load disturbances, structural deformation, and sediment accumulation. In severe cases, this can lead to fires, explosions, or major environmental pollution accidents. Therefore, high-precision and highly reliable real-time monitoring and early warning of the floating roof tilt angle has become a requirement for the safe operation and maintenance of storage tanks.

[0003] Large floating roofs can reach diameters of over 60 meters, with large structural spans and uneven stiffness distribution, making it difficult to reflect the overall attitude from tilt measurements at a single location. Existing monitoring solutions typically deploy multiple tilt sensors on the floating roof to cover key areas, but data from each sensor is often processed independently, lacking an effective mechanism for fusing multi-source information. Due to inherent sensor limitations such as installation errors, temperature drift, and local disturbances, simply taking the average or maximum value as the basis for judgment can easily lead to false alarms or missed alarms, failing to accurately characterize the true tilt state of the floating roof.

[0004] In existing technologies, tilt angle warnings generally employ fixed threshold triggering mechanisms, failing to consider the impact of dynamic operating conditions such as tank operation phase, ambient temperature, and liquid level on the floating roof's behavior. This results in frequent alarms under normal operating condition fluctuations, while lacking sensitivity to hidden risks such as slow, cumulative tilting. Furthermore, the system lacks self-diagnostic capabilities for sensor malfunctions; if a sensor fails or drifts, it directly affects the overall judgment logic.

[0005] Current monitoring systems lack intelligent linkage control logic with safety interlock devices such as emergency shut-off and nitrogen sealing protection of storage tanks. Even when a dangerous tilt angle is detected, rapid response and proactive intervention are difficult to achieve, failing to meet the high standards of inherent safety and proactive defense required by modern intelligent storage and transportation systems. Therefore, there is an urgent need for an intelligent monitoring and early warning method for floating roof tilt angle that integrates multi-sensor data, possesses dynamic threshold generation, fault self-diagnosis, and safety linkage capabilities. Summary of the Invention

[0006] This invention provides a multi-sensor tilt fusion monitoring and dynamic early warning method for large floating roofs, aiming to solve the technical challenge of accurately, in real-time, and globally sensing and warning of local tilting caused by structural deformation, uneven medium distribution, or external disturbances during operation of large oil tank floating roofs with diameters up to 60 meters. Existing technologies typically use single-point or a small number of discretely arranged tilt sensors for monitoring. The data only reflects the instantaneous attitude of a local area and cannot effectively characterize the spatial attitude field distribution of the entire floating roof surface. Furthermore, these sensors are susceptible to sensor drift, installation errors, or local interference, leading to false alarms or missed alarms. Existing systems lack the ability to perform spatiotemporal consistency calibration, spatial weight allocation, and dynamic evolution trend modeling of multi-source heterogeneous tilt data, making it difficult to support accurate assessment and graded early warning of the overall stability state of the floating roof.

[0007] To overcome the above problems, the present invention provides a method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks, comprising: dividing the surface of the large floating dock into a spatial grid and deploying a multi-sensor array, wherein the multi-sensor array includes multiple high-precision dual-axis tilt angle sensors, and each sensor corresponds to a monitoring unit with a defined geometric coordinate;

[0008] All tilt sensors are synchronized at the microsecond level using a unified clock source, and raw tilt data is collected in real time, including tilt values ​​in two orthogonal directions.

[0009] The original tilt angle data is preprocessed to remove outlier values ​​in the range, compensate for zero-point offset caused by temperature drift, and perform time alignment of asynchronous data based on the assumption of rigid body motion of the floating board.

[0010] Based on the physical installation positions of each sensor in the three-dimensional coordinate system of the floating table and the pre-processed tilt angle data, an initial model of the floating table tilt angle field is constructed.

[0011] Spatial consistency verification is performed on the initial model of the floating disk tilt field. The tilt gradient between adjacent monitoring units is calculated. If the gradient exceeds the maximum allowable threshold, it is determined that there is a local anomaly. The spatial interpolation method based on radial basis function is used to repair the data of the abnormal unit.

[0012] Based on the structural mechanical properties of the floating roof, adaptive spatial weight coefficients are assigned to each monitoring unit, and the tilt angle data of all monitoring units are weighted and fused according to the weight coefficients to calculate the overall equivalent tilt angle and the maximum local tilt angle of the floating roof.

[0013] A dynamic evolution model of the floating roof tilt angle is constructed. The overall equivalent tilt angle and the maximum local tilt angle are sampled by a sliding window to form a time series. A disturbance-response relationship model is established in combination with environmental disturbance factors to distinguish between normal fluctuations and structural anomalies.

[0014] A multi-level dynamic early warning threshold system is set up, and the early warning thresholds at each level are dynamically adjusted according to the intensity of the current environmental disturbance. The corresponding level of early warning response process is triggered based on the maximum local tilt angle and its rate of change.

[0015] Preferably, the surface of the large floating table is divided into a spatial grid and a multi-sensor array is deployed, including:

[0016] The surface of the floating table is divided into regular hexagonal or square monitoring units;

[0017] At least one high-precision dual-axis tilt sensor is installed in each monitoring unit;

[0018] Record the position coordinates of each sensor in a three-dimensional coordinate system with the geometric center of the floating disk as the origin, the X-axis along the east-west direction, the Y-axis along the north-south direction, and the Z-axis vertically upward. , For the first The X-axis coordinates of each sensor. For the first The Y-axis coordinate values ​​of each sensor. The design height for the floating roof is considered a constant.

[0019] Preferably, all tilt sensors are synchronized at the microsecond level using a unified clock source, and raw tilt data is acquired in real time, including:

[0020] A high-stability crystal oscillator is provided by the central processing unit as a unified clock source, and synchronization signals are distributed through wired or low-latency wireless protocols;

[0021] Raw tilt angle data is acquired at a fixed sampling frequency. The raw tilt angle data includes a unique sensor identifier, a timestamp, and the tilt angle in the X direction. Y-direction tilt angle and verification code;

[0022] Data is transmitted to the central processing unit via industrial Ethernet or an anti-interference wireless communication module.

[0023] Preferably, the original tilt angle data is preprocessed to remove outliers in the range, compensate for zero-point offset caused by temperature drift, and time-align the asynchronous data based on the assumption of rigid body motion of the floating disk, including:

[0024] when or If so, the original tilt angle data is deemed invalid and discarded;

[0025] Based on the ambient temperature measured by the built-in temperature sensor The tilt angle value is compensated using the factory-calibrated temperature-zero offset curve. The compensation formula is as follows: , For the first Tilt angle data in the X-axis direction of each sensor after temperature drift compensation For the first Tilt angle data in the Y-axis direction of each sensor after temperature drift compensation , This is the temperature drift coefficient. For calibration reference temperature;

[0026] Using the sensor at the geometric center of the floating disk as a reference, time alignment is performed on the data from other sensors using linear interpolation. The interpolation formula is as follows: , For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. , This represents the rate of change of tilt angle in the previous cycle. This is due to time deviation.

[0027] Preferably, a spatial consistency check is performed on the initial model of the floating disk tilt field, the tilt gradient between adjacent monitoring units is calculated, and if the gradient exceeds the maximum allowable threshold, a local anomaly is determined to exist. A spatial interpolation method based on radial basis functions is then used to repair the data of the anomaly units, including:

[0028] Calculate any adjacent monitoring units and The gradient of the inclination between , For the first The tilt vector of each sensor, No. The tilt vector of each sensor, Denotes the Euclidean norm. The Euclidean distance between the two sensors;

[0029] Set the maximum allowed gradient threshold ,like This identifies the unit most likely to be abnormal. ;

[0030] Gaussian radial basis functions are used to pair Perform the repair; the repair formula is: , The reasonable tilt angle value of the abnormal element k after radial basis function interpolation is given. for The normal neighborhood set, For smoothness parameters, For normalized weights.

[0031] Preferably, an adaptive spatial weighting coefficient is assigned to each monitoring unit based on the structural mechanical characteristics of the floating roof, and the tilt angle data of all monitoring units are weighted and fused according to the weighting coefficient to calculate the overall equivalent tilt angle and the maximum local tilt angle of the floating roof, including:

[0032] Calculate the first Weighting coefficients of each monitoring unit , Its historical dip angle data standard deviation, The structural stiffness coefficient, The normalized distance to the center of the floating platform. , , To meet The positive real number adjustment parameter;

[0033] For all Normalization processing ;

[0034] Calculate the overall equivalent tilt angle of the floating roof And record the maximum local tilt angle. .

[0035] Preferably, a dynamic evolution model of the floating roof tilt angle is constructed, and the overall equivalent tilt angle and the maximum local tilt angle are sampled by a sliding window to form a time series. A disturbance-response relationship model is established in conjunction with environmental disturbance factors to distinguish between normal fluctuations and structural anomalies, including:

[0036] Using a 300-second sliding time window and Perform continuous sampling;

[0037] calculate and instantaneous rate of change and :

[0038] ;

[0039] For time, The sampling interval;

[0040] Access wind speed Liquid level change rate and temperature gradient Through multivariate linear regression model Calculate the expected dip angle. , , , For regression coefficients, if the actual and A deviation exceeding three times the standard deviation is considered abnormal.

[0041] Preferably, a multi-level dynamic early warning threshold system is set up, and the early warning thresholds at each level are dynamically adjusted according to the current environmental disturbance intensity. Based on the maximum local tilt angle and its rate of change, a corresponding level of early warning response process is triggered, including:

[0042] Set a Level 1 warning threshold The level is two, which is the second-level warning threshold. For the last 30 days 99%, Level 3 warning threshold for The speed is 10 degrees per second and the duration exceeds 10 seconds;

[0043] when meters per second, and Increase by 15%; when meters per minute Increase by 10%;

[0044] when A level one warning is triggered at this time. A level-two warning is triggered at this time. The conditions are met and A level-three warning is triggered at this time;

[0045] The early warning response includes highlighting over-limit units with different colors on the monitoring interface, generating an early warning report containing a tilt field heat map and trend analysis, triggering audible and visual alarms, and recording the event to a historical database.

[0046] Preferably, the structural stiffness coefficient The normalized distance is predetermined by finite element analysis of the floating roof. , To monitor the actual distance from the monitoring unit to the center of the floating table, Where is the radius of the floating platform.

[0047] Preferably, the scaling parameter in the Gaussian radial basis function The stiffness distribution of the floating roof structure is dynamically adjusted, with smaller values ​​used in the central region to enhance the preservation of local details and larger values ​​used in the edge region to improve smoothness.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] 1. By implementing a gridded multi-point deployment of tilt sensors on the surface of a large floating table, and combining microsecond-level time synchronization and spatial coordinate mapping, a complete floating table tilt field spatial model was constructed, overcoming the limitation that single-point monitoring cannot reflect the global attitude.

[0050] 2. By introducing spatial consistency verification and a data repair mechanism based on radial basis function, the negative impact of local sensor failures or interference on the overall judgment is suppressed, and data reliability is improved.

[0051] 3. By establishing an adaptive spatial weighted fusion strategy based on structural mechanical properties, the fusion results more realistically reflect the overall stability state of the floating roof, rather than a simple average.

[0052] 4. By constructing a dynamic evolution model that includes environmental disturbance factors and a multi-level adjustable early warning threshold system, the false alarm rate and false alarm rate have been reduced, realizing the transformation from passive response to proactive prediction.

[0053] 5. This method provides high-precision, robust, and highly adaptable tilt angle monitoring and intelligent early warning capabilities for large oil tank floating roofs with diameters up to 60 meters, ensuring the safety and continuity of oil tank operation, extending the service life of the floating roof structure, and providing technical support for preventing major safety accidents. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention;

[0055] Figure 2 This is a schematic diagram of the core principle framework of the adaptive spatial weighted fusion strategy based on structural mechanical properties in this invention;

[0056] Figure 3 This is a flowchart illustrating the logical process of multi-sensor spatiotemporal synchronous acquisition and tilt field initial model construction in this invention.

[0057] Figure 4 This is a logical flowchart of the floating disk tilt angle dynamic evolution modeling and multi-level dynamic early warning threshold system in this invention;

[0058] Figure 5 This is a schematic diagram of the multi-level interaction relationship and data flow between the terminal sensor network and the central processing unit in this invention. Detailed Implementation

[0059] Reference Appendix Figures 1 to 5 This invention provides a multi-sensor tilt fusion monitoring and dynamic early warning method for large floating roofs, applicable to oil tank floating roof structures with diameters up to 60 meters. This method constructs a tilt sensing network covering the entire floating roof surface by deploying multiple high-precision dual-axis tilt sensors in a grid pattern on the floating roof surface. Combined with microsecond-level time synchronization, spatial coordinate mapping, adaptive weighted fusion, and a dynamic threshold early warning mechanism, it achieves high-precision, robust, real-time monitoring and tiered early warning of the overall attitude state of the floating roof.

[0060] The method includes the following steps:

[0061] S1, spatial grid division of the surface of a large floating table and deployment of a multi-sensor array;

[0062] S2 performs spatiotemporal synchronous acquisition and transmission of raw tilt angle data from multiple sensors;

[0063] S3, preprocess the raw tilt angle data to eliminate outliers, temperature drift and time asynchrony effects;

[0064] S4, Construct an initial model of the floating dock tilt field based on the physical location of the sensors;

[0065] S5 performs space consistency checks and repairs local abnormal data.

[0066] S6 employs an adaptive spatial weighting strategy based on structural mechanical properties for global tilt angle fusion calculation;

[0067] S7, Construct a dynamic evolution model of the floating roof tilt angle and introduce environmental disturbance factors;

[0068] S8 sets up a multi-level dynamic early warning threshold system and executes a graded early warning response process.

[0069] In step S1, the surface of the large floating roof is divided into several monitoring units with clearly defined geometric coordinates. Each monitoring unit is a regular hexagonal or square area, with its side length determined based on the floating roof diameter, structural stiffness distribution, and expected monitoring accuracy; a typical value is 5 to 10 meters. Each monitoring unit is equipped with at least one high-precision dual-axis tilt sensor with a measurement range of -90 degrees to +90 degrees, a resolution of not less than 0.01 degrees, and a repeatability error of less than 0.05 degrees. All sensors are bolted to a rigid support structure on the upper surface of the floating roof, ensuring that their measurement axes are strictly parallel to the local plane of the floating roof. After the sensors are installed, the position coordinates of each sensor in a three-dimensional coordinate system with the geometric center of the floating roof as the origin, the X-axis along the east-west direction, the Y-axis along the north-south direction, and the Z-axis vertically upward are recorded. , For the first The X-axis coordinates of each sensor. For the first The Y-axis coordinate values ​​of each sensor. The design height of the floating platform is considered constant. The three-dimensional coordinate system of the floating platform has its origin at the geometric center of the platform, with the X-axis along the east-west direction, the Y-axis along the north-south direction, and the Z-axis vertically upward. A total of no fewer than 36 sensors are required to ensure a sufficiently dense sensing network on the 60-meter diameter floating platform.

[0070] In step S2, all tilt sensors achieve microsecond-level time synchronization via a unified clock source. This clock source is provided by a high-stability crystal oscillator built into the central processing unit and distributed to each sensor node via a wired synchronization signal line or a low-latency wireless synchronization protocol. The synchronization accuracy is better than 10 microseconds. The sensors acquire raw tilt data at a fixed sampling frequency, typically 20 Hz. The raw data includes the tilt angle value in the X direction. Inclination angle with the Y direction , representing the tilt angles of the floating roof relative to the horizontal plane in the X and Y directions, respectively. All raw data is transmitted to the central processing unit in real time via industrial Ethernet or an interference-resistant wireless communication module, with a transmission delay of no more than 50 milliseconds. The data packet format includes a unique sensor identifier, a timestamp, and , And a verification code to ensure data integrity and traceability.

[0071] In step S3, the central processing unit performs preprocessing on all received raw tilt angle data. Outliers exceeding the sensor's range limits are removed, i.e., when... or If this occurs, the original tilt angle data is deemed invalid. Zero-point offset caused by temperature drift is compensated. Each sensor has a built-in temperature sensor to monitor the ambient temperature in real time. Based on the factory-calibrated temperature-zero offset curve, for and Compensation will be provided, and the compensation formula is as follows: ; For the first Tilt angle data in the X-axis direction of each sensor after temperature drift compensation For the first Tilt angle data in the Y-axis direction of each sensor after temperature drift compensation and This is the temperature drift coefficient. To calibrate the reference temperature, typically 25 degrees Celsius, a time alignment operation is performed. The sensor located at the geometric center of the floating disk is selected as the primary reference sensor, and its timestamp is used. As a benchmark. For other sensors If data timestamp and There is a deviation Then, a linear interpolation algorithm based on the assumption of rigid body motion of the floating disk is used to calculate the... Inclination angle at time:

[0072] ;

[0073] For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. and For sensors The rate of change of tilt angle during the previous sampling period is calculated using the first-order difference. If If the maximum allowed synchronization window (e.g., 100 milliseconds) is exceeded, the data is discarded and marked as a communication error.

[0074] In step S4, an initial model of the floating table's tilt field is constructed based on the physical installation positions of each sensor on the floating table and its preprocessed tilt angle data. This initial model of the floating table's tilt field uses the floating table surface as its domain and any point... Inclination vector at point This is the value range. For point The tilt angle value along the X-axis. For point The initial model for the tilt angle value along the Y-axis (north-south) treats each sensor location as a discrete sampling point and directly assigns its tilt angle data, forming a sparse tilt angle field representation. This initial model of the floating dock tilt angle field provides the basic data structure for subsequent spatial consistency verification and fusion calculations.

[0075] In step S5, a spatial consistency check is performed. This check is based on the physical constraint of the floating roof structure's continuity: the tilt angle change between adjacent regions should be smooth, and its gradient should not exceed the critical value for material yielding or structural instability. The calculation is performed between any two adjacent monitoring units. and The gradient of the inclination between Its definition is: ; For the first The tilt vector of each sensor, No. The tilt vector of each sensor, Denotes the Euclidean norm. The Euclidean distance between the two sensors. Set the maximum allowable gradient threshold. The typical value is 0.5 degrees per meter. If Then the determination unit or At least one of them contains a local anomaly.

[0076] Furthermore, by comparing the gradient mean of a given cell with that of all its neighboring cells, the most likely anomalous cells are identified. Once the anomalous cells are identified... The data repair mechanism is activated. This mechanism employs a spatial interpolation method based on radial basis functions, utilizing... Reconstruct the dip angle data of all normal cells in the neighborhood. The appropriate tilt angle value for the location. A Gaussian kernel is chosen as the radial basis function, and its expression is: ; The reasonable tilt angle value of the abnormal element k after radial basis function interpolation is given. for The normal neighborhood set, and Units and The position vector, The scaling parameter is used to control the smoothness of interpolation. For normalized weight coefficients, satisfying The repaired data replaces the original outliers, updates the tilt field model, and ensures its spatial continuity and physical rationality.

[0077] In step S6, an adaptive spatial weighted fusion strategy based on the structural mechanical properties of the floating roof is introduced. This adaptive spatial weighted fusion strategy aims to dynamically allocate the weight of each monitoring unit in the global tilt angle calculation according to the importance of each unit, avoiding the central region information being overwhelmed by edge noise due to simple averaging. The formula for calculating the weight coefficient of each monitoring unit is: ; For the first The standard deviation of the tilt angle data of each monitoring unit in the most recent hour reflects its historical stability; The structural stiffness coefficient of the region where the element is located is predetermined by the finite element analysis of the floating roof, with a value of 1.0 for the central region and 0.6 for the edge region; The normalized distance from the cell to the geometric center of the floating disk. , This is the actual distance. The radius of the floating platform; , , The preset positive real number adjustment parameter satisfies Typical value All weight coefficients are normalized to make Based on this weight, the overall equivalent tilt angle of the floating roof is calculated. : ;

[0078] At the same time, the maximum value of the tilt angle amplitude in all monitoring units is recorded as the maximum local tilt angle. .

[0079] In step S7, a dynamic evolution model of the floating dock tilt angle is constructed. This dynamic evolution model of the floating dock tilt angle uses a sliding time window of 300 seconds to... and Continuous sampling is performed to form a time series. First-order difference is used to calculate... and instantaneous rate of change and : ; For time, The sampling interval is 50 milliseconds. Simultaneously, an external environmental monitoring system is connected to obtain real-time wind speed. Liquid level change rate and the surface temperature gradient of the floating table The perturbation-response relationship is established using a multivariate linear regression model: ;

[0080] The expected dip angle value, , , The regression coefficients were obtained through offline training using historical normal operating data. The actual... and If the deviation exceeds three times the standard deviation, it is judged as an abnormal change, which may be caused by structural jamming or local instability.

[0081] In step S8, a multi-level dynamic early warning threshold system is set. The first-level early warning threshold... Based on the floating roof design specifications, the typical value is level two; the level two warning threshold. Based on the statistical distribution of historical operating data, take The 99th percentile value over the past 30 days; Level 3 warning threshold. Based on the dynamic rate of change, when Triggered when the intensity of the disturbance exceeds 10 seconds per second. The thresholds at each level are dynamically adjusted based on the intensity of the current environmental disturbance. The specific adjustment rule is as follows: If... meters per second, then and Increase by 15% proportionally; if meters per minute, then Increase by 10%. The warning trigger logic is: when When, a Level 1 warning is triggered; when When, a level-two warning is triggered; when The conditions are met and When the event occurs, a Level 3 warning is triggered. The warning response process includes: highlighting the over-limit monitoring unit in yellow, orange, and red on the monitoring interface; generating a warning report containing the current tilt field heat map, over-limit values, the trend of change within 10 minutes, environmental disturbance parameters, and possible cause analysis; triggering the audible and visual alarm device; and completely recording the event to the historical database, including timestamps, all raw data from sensors, fusion results, warning level, and operator confirmation status.

[0082] The aforementioned method integrates a full-chain technology framework, including gridded multi-point sensing, spatiotemporal synchronization, spatial verification, adaptive weighting, dynamic modeling, and intelligent early warning, to achieve global, accurate, and robust monitoring of the tilt angle of large floating roofs. This method effectively overcomes the limitations of single-point monitoring, suppresses the influence of local interference, distinguishes between environmental disturbances and structural anomalies, and significantly reduces false alarms and missed alarms through dynamic thresholds, providing a reliable guarantee for the safe operation of large oil tanks.

[0083] This invention also discloses a system for performing the above-described method. The system includes a terminal sensor network, a central processing unit, an environmental monitoring interface, an early warning output module, and a historical database. The terminal sensor network consists of no fewer than 36 high-precision dual-axis tilt sensors, evenly distributed on the surface of the floating dock. Each sensor integrates temperature sensing and time synchronization functions. The central processing unit is an industrial-grade embedded computer equipped with a multi-core processor and large-capacity memory, running a real-time operating system to perform data reception, preprocessing, model building, fusion calculation, dynamic modeling, and early warning judgment. The environmental monitoring interface connects to an anemometer, level gauge, and temperature sensor network via standard industrial protocols to acquire external disturbance data. The early warning output module includes a graphical user interface, an audible and visual alarm, and a remote communication interface for human-computer interaction and information push. The historical database adopts a relational database management system to store all raw data, intermediate results, and early warning events, supporting queries and backtracking by time, location, and early warning level. The system, through a hard real-time architecture, ensures that the end-to-end latency from data acquisition to early warning output does not exceed 200 milliseconds, meeting the timeliness requirements for large floating dock safety monitoring.

[0084] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0085] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating roofs, characterized in that, include: The surface of a large floating table is divided into spatial grids and a multi-sensor array is deployed. The multi-sensor array includes multiple high-precision dual-axis tilt sensors, each of which corresponds to a monitoring unit with defined geometric coordinates. All tilt sensors are synchronized at the microsecond level using a unified clock source, and raw tilt data is collected in real time, including tilt values ​​in two orthogonal directions. The original tilt angle data is preprocessed to remove outlier values ​​in the range, compensate for zero-point offset caused by temperature drift, and time-align the asynchronous data based on the assumption of rigid body motion of the floating board. Based on the physical installation positions of each sensor in the three-dimensional coordinate system of the floating table and the pre-processed tilt angle data, an initial model of the floating table tilt angle field is constructed. Spatial consistency verification is performed on the initial model of the floating disk tilt field. The tilt gradient between adjacent monitoring units is calculated. If the gradient exceeds the maximum allowable threshold, it is determined that there is a local anomaly. The spatial interpolation method based on radial basis function is used to repair the data of the abnormal unit. Based on the structural mechanical properties of the floating roof, adaptive spatial weight coefficients are assigned to each monitoring unit, and the tilt angle data of all monitoring units are weighted and fused according to the weight coefficients to calculate the overall equivalent tilt angle and the maximum local tilt angle of the floating roof. A dynamic evolution model of the floating roof tilt angle is constructed. The overall equivalent tilt angle and the maximum local tilt angle are sampled by a sliding window to form a time series. A disturbance-response relationship model is established in combination with environmental disturbance factors to distinguish between normal fluctuations and structural anomalies. A multi-level dynamic early warning threshold system is set up, and the early warning thresholds at each level are dynamically adjusted according to the intensity of the current environmental disturbance. The corresponding level of early warning response process is triggered based on the maximum local tilt angle and its rate of change.

2. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks according to claim 1, characterized in that, Spatial grid division and deployment of multi-sensor arrays on the surface of large floating docks, including: The surface of the floating table is divided into regular hexagonal or square monitoring units; At least one high-precision dual-axis tilt sensor is installed in each monitoring unit; Record the position coordinates of each sensor in a three-dimensional coordinate system with the geometric center of the floating disk as the origin, the X-axis along the east-west direction, the Y-axis along the north-south direction, and the Z-axis vertically upward. , For the first The X-axis coordinates of each sensor. For the first The Y-axis coordinate values ​​of each sensor. The design height for the floating roof is considered a constant.

3. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks according to claim 2, characterized in that, All tilt sensors are synchronized at the microsecond level using a unified clock source, and raw tilt data is acquired in real time, including: A high-stability crystal oscillator is provided by the central processing unit as a unified clock source, and synchronization signals are distributed through wired or low-latency wireless protocols; Raw tilt angle data is acquired at a fixed sampling frequency. The raw tilt angle data includes a unique sensor identifier, a timestamp, and the tilt angle in the X direction. Y-direction tilt angle and verification code; Data is transmitted to the central processing unit via industrial Ethernet or an anti-interference wireless communication module.

4. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks according to claim 3, characterized in that, The raw tilt angle data is preprocessed to remove outliers in the range, compensate for zero-point offset caused by temperature drift, and perform time alignment of asynchronous data based on the assumption of rigid body motion of the floating board, including: when or If so, the original tilt angle data is deemed invalid and discarded; Based on the ambient temperature measured by the built-in temperature sensor The tilt angle value is compensated using the factory-calibrated temperature-zero offset curve. The compensation formula is as follows: , For the first Tilt angle data in the X-axis direction of each sensor after temperature drift compensation For the first Tilt angle data in the Y-axis direction of each sensor after temperature drift compensation , This is the temperature drift coefficient. For calibration reference temperature; Using the sensor at the geometric center of the floating disk as a reference, time alignment is performed on the data from other sensors using linear interpolation. The interpolation formula is as follows: , For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. For the first The tilt angle values ​​of each sensor along the X-axis after time alignment. , This represents the rate of change of the tilt angle in the previous cycle. This is due to time deviation.

5. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks according to claim 4, characterized in that, Spatial consistency verification is performed on the initial model of the floating disk tilt field. The tilt gradient between adjacent monitoring units is calculated. If the gradient exceeds the maximum allowable threshold, a local anomaly is determined to exist. A spatial interpolation method based on radial basis functions is used to repair the data of the anomaly units, including: Calculate any adjacent monitoring units and The gradient of the inclination between , For the first The tilt vector of each sensor, No. The tilt vector of each sensor, Describes the Euclidean norm. The Euclidean distance between the two sensors; Set the maximum allowed gradient threshold ,like This identifies the unit most likely to be abnormal. ; Gaussian radial basis functions are used to pair Perform the repair; the repair formula is: , The reasonable tilt angle value of the abnormal element k after radial basis function interpolation is given. for The normal neighborhood set, For smoothness parameters, For normalized weights.

6. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating roofs according to claim 5, characterized in that, Based on the structural mechanical characteristics of the floating roof, adaptive spatial weight coefficients are assigned to each monitoring unit. Then, the tilt angle data of all monitoring units are weighted and fused according to these weight coefficients to calculate the overall equivalent tilt angle and the maximum local tilt angle of the floating roof, including: Calculate the first Weighting coefficients of each monitoring unit , The standard deviation of its historical dip data, The structural stiffness coefficient, The normalized distance to the center of the floating platform. , , To meet The positive real number adjustment parameter; For all Normalization processing ; Calculate the overall equivalent tilt angle of the floating roof And record the maximum local tilt angle. .

7. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating roofs according to claim 6, characterized in that, A dynamic evolution model of the floating roof tilt angle is constructed. The overall equivalent tilt angle and the maximum local tilt angle are sampled using a sliding window to form a time series. A disturbance-response relationship model is established by combining environmental disturbance factors to distinguish between normal fluctuations and structural anomalies, including: Using a 300-second sliding time window and Perform continuous sampling; calculate and instantaneous rate of change and : ； For time, The sampling interval; Access wind speed Liquid level change rate and temperature gradient Through multivariate linear regression model Calculate the expected dip angle. , , , For regression coefficients, if the actual and A deviation exceeding three times the standard deviation is considered abnormal.

8. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating roofs according to claim 7, characterized in that, A multi-level dynamic early warning threshold system is established, which dynamically adjusts the early warning thresholds at each level according to the current environmental disturbance intensity, and triggers the corresponding level of early warning response process based on the maximum local tilt angle and its rate of change, including: Set a Level 1 warning threshold The level is two, which is the second-level warning threshold. For the last 30 days 99%, Level 3 warning threshold for The speed is 10 degrees per second and the duration exceeds 10 seconds; when meters per second, and Increase by 15%; when meters per minute Increase by 10%; when A level one warning is triggered at this time. A level-two warning is triggered at this time. The conditions are met and A level-three warning is triggered at this time; The early warning response includes highlighting over-limit units with different colors on the monitoring interface, generating an early warning report containing a tilt field heat map and trend analysis, triggering audible and visual alarms, and recording the event to a historical database.

9. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating roofs according to claim 8, characterized in that, The structural stiffness coefficient The normalized distance is predetermined by finite element analysis of the floating roof. , To monitor the actual distance from the monitoring unit to the center of the floating table, Where is the radius of the floating platform.

10. The method for multi-sensor tilt angle fusion monitoring and dynamic early warning of large floating docks according to claim 9, characterized in that, The scaling parameter in the Gaussian radial basis function The stiffness distribution of the floating roof structure is dynamically adjusted, with smaller values ​​used in the central region to enhance the preservation of local details and larger values ​​used in the edge region to improve smoothness.

Images