Silo safety monitoring method and system

Abstract

The invention relates to the technical field of safety monitoring and industrial automation, and discloses a silo safety monitoring method and system.The method comprises the steps that a laser scanning unit, an annular pressure sensing belt, a weighing sensor and a temperature and humidity node are arranged, and coal pile surface point cloud, lateral pressure, vertical load and environment data are synchronously obtained; a high-precision three-dimensional form is reconstructed based on an adaptive mesh refinement algorithm, and mechanical data are fused to construct a joint representation model; establishing a stability discrimination function in combination with coal physical parameters, and identifying an instability critical region; through mode matching with a historical stable form library, graded early warning and blockage clearing strategy generation of abnormal working conditions such as suspension arching, bonding and blockage are realized. According to the method, the laser point cloud, the pressure distribution and the load sensing data are fused, a combined characterization model of the three-dimensional form and the mechanical state of the coal pile is constructed, the limitation that traditional single-point material level measurement cannot reflect the whole coal pile structure is broken through, and high-precision measurement with the reserve calculation error smaller than 2% is achieved.

Description

Technical Field

[0001] This invention belongs to the field of safety monitoring and industrial automation, specifically relating to a method and system for silo safety monitoring. Background Technology

[0002] With the increasing demands for silo storage efficiency and operational safety in industries such as thermal power generation, coal storage and transportation, and chemical production, accurate sensing of the material status inside coal silos has become a crucial link in ensuring continuous material supply and preventing safety accidents. As large-scale bulk material storage facilities, the coal piles inside silos are prone to forming complex stacking patterns under the influence of multiple factors such as gravity, humidity, particle size distribution, and feeding methods, which may induce risks such as local blockage, arching, uneven loading, or even collapse. Traditional monitoring methods mainly rely on fixed-point silo level gauges (such as radar, ultrasonic, or plumb bob level gauges), which can only obtain material level height data at a single location within the silo. They cannot reflect the overall three-dimensional contour and spatial distribution characteristics of the coal pile, resulting in significant deviations in reserve estimation and making it difficult to identify structural hazards hidden inside the silo.

[0003] Mobile platform-based 3D scanning technology offers a new solution for sensing the internal conditions of coal bunkers. This technology deploys sensors that can autonomously move along tracks or bunker walls to perform omnidirectional scanning of the coal pile surface, aiming to construct a high-precision spatial model. However, existing solutions mostly focus on static scanning or limited-view reconstruction, lacking adaptability to complex bunker environments (such as high dust levels, strong electromagnetic interference, and metallic reflective surfaces), and failing to establish an effective analysis link from raw point cloud data to safety warning indicators, making it difficult to support real-time, reliable anomaly identification and decision-making responses.

[0004] Fixed sensor networks have limited coverage and cannot dynamically track the evolution of coal pile morphology. While introducing drones or robots for internal inspections is feasible, their navigation and positioning are easily interfered with by the enclosed metal structure within the silo, and their scanning path planning lacks a coordination mechanism with coal flow conditions, resulting in low data acquisition efficiency and numerous blind spots. More importantly, current systems generally lack the ability to intelligently identify typical precursory features of silo blockage, such as "arching," "mouse holes," and "wall-attached caking," failing to provide effective early warnings of potential hazards. Therefore, under the extreme conditions of high dust and strong interference in silos, there is an urgent need for an intelligent safety monitoring method and system that integrates mobile scanning, 3D modeling, morphological analysis, and risk warning to achieve comprehensive, real-time, and accurate perception and proactive protection of the coal pile's condition within the silo. Summary of the Invention

[0005] To address the aforementioned technical issues, embodiments of the present invention provide a silo safety monitoring method and system. This method involves deploying a multi-source heterogeneous sensor array inside the silo to simultaneously collect point cloud data of the coal pile surface, internal stress distribution data, and environmental temperature and humidity field data, constructing a joint characterization model of the coal pile's three-dimensional morphology and mechanical state. Based on this joint characterization model, an adaptive mesh refinement algorithm is used to reconstruct the coal pile surface with high precision, and a coal pile stability discrimination function is established by combining physical parameters such as coal friction angle and cohesion. Furthermore, by comparing the current coal pile morphology with benchmark patterns in a historical stable morphology database in real time, abnormal operating conditions such as localized arching, sidewall adhesion, and bottom blockage are identified, and graded early warning signals and blockage clearing strategy suggestions are generated. This enables high-precision measurement of silo storage capacity and early proactive identification of safety hazards.

[0006] This invention provides a silo safety monitoring method, comprising: uniformly arranging no fewer than eight laser scanning units along the circumference below the dome structure inside the silo; each laser scanning unit emitting a fan-shaped laser beam downwards at a fixed pitch angle to scan the surface of the coal pile and acquire raw point cloud data of the coal pile surface; embedding no fewer than three layers of annular pressure sensing strips along the height direction on the inner wall of the silo, each layer of pressure sensing strip consisting of no fewer than 16 piezoresistive thin-film sensors for real-time monitoring of the lateral pressure distribution of the coal on the inner wall of the silo; and installing no fewer than four weighing sensors above the conical discharge port at the bottom of the silo. This system is used to monitor the vertical load of the coal mass on the unloading port area; at least three temperature and humidity sensing nodes are deployed inside the silo to collect ambient temperature and humidity data; the original point cloud data is denoised, filtered, and processed using a coordinate system to generate a standardized point cloud dataset; based on the standardized point cloud dataset, an initial voxel mesh is constructed using an octree spatial segmentation method, and the voxel resolution is dynamically adjusted according to the point cloud density gradient; local mesh refinement is implemented at the coal pile edge and slope area to generate a high-precision 3D model of the coal pile surface; the high-precision 3D model of the coal pile surface is then compared with... The lateral pressure distribution data and vertical load data are spatiotemporally aligned to construct a joint characterization model of the coal pile's three-dimensional morphology and mechanical state. Based on this joint characterization model, the local slope angle, rate of curvature change, and stress concentration coefficient of each region of the coal pile are calculated. According to the physical parameters of the coal body, including friction angle, cohesion, and bulk density, a stability discrimination function for the coal pile is established. The stability discrimination function is defined as follows: when the local slope angle is greater than the friction angle and the stress concentration coefficient exceeds a preset threshold, the region is determined to be in a critical state of instability. The current three-dimensional morphology of the coal pile is matched with a pre-stored historical stable morphology library, which includes three benchmark morphologies: normal flow, complete emptying, and uniform accumulation. If the Euclidean distance between the current morphology and any benchmark morphology exceeds a first preset threshold, a level one warning is triggered. If the conditions for determining the critical state of instability are met simultaneously, a level two warning is triggered. If the bottom vertical load is continuously higher than a second preset threshold during the unloading process and there is no significant drop in the coal pile surface, it is determined to be bottom blockage, triggering a level three warning. Based on the warning level, corresponding unblocking strategy suggestions are generated, including vibration frequency, air cannon injection angle, and unloading port opening adjustment parameters.

[0007] This invention also provides a silo safety monitoring system, comprising: a multi-source sensor data acquisition module, used to uniformly deploy no fewer than 8 laser scanning units along the circumference below the dome structure inside the silo; to embed no fewer than three layers of annular pressure sensing strips along the height direction on the inner wall of the silo, each layer of pressure sensing strips consisting of no fewer than 16 piezoresistive thin-film sensors; to set no fewer than 4 weighing sensors above the conical discharge port at the bottom of the silo; and to deploy no fewer than 3 temperature and humidity sensing nodes in the internal space of the silo, respectively collecting point cloud data of the coal pile surface, lateral pressure distribution data, vertical load data, and environmental temperature and humidity data; a point cloud preprocessing and 3D reconstruction module, used to perform noise reduction, filtering, and coordinate system-unified processing on the original point cloud data to generate a standardized point cloud dataset, and to generate a high-precision 3D model of the coal pile surface using an octree spatial segmentation and adaptive mesh refinement algorithm; and a joint characterization modeling module, used to spatiotemporally align the high-precision 3D model of the coal pile surface with the lateral pressure distribution data and vertical load data to construct a joint characterization model of the coal pile's 3D morphology and mechanical state. The stability analysis module is used to calculate the local slope angle, rate of change of curvature, and stress concentration coefficient based on the joint characterization model, and to establish a stability discrimination function based on the physical parameters of the coal body to perform instability critical state determination; the anomaly identification and early warning module is used to perform pattern matching between the current three-dimensional shape of the coal pile and the historical stable shape library, and to generate first-level, second-level, or third-level early warning signals based on the matching deviation and stability determination results; the unblocking strategy generation module is used to output the corresponding vibration frequency, air cannon injection angle, and unloading port opening adjustment parameters according to the early warning level.

[0008] In one embodiment of the present invention, the laser scanning unit uses a pulsed laser with a wavelength of 905 nanometers, a scanning frequency of 20 Hz, a horizontal field of view of 360 degrees, a vertical field of view of 60 degrees, and a ranging accuracy of ±2 mm.

[0009] In one embodiment of the present invention, the piezoresistive thin-film sensor has a range of 0 to 50 kPa, a sensitivity of 0.1 kPa, and a response time of 5 milliseconds. It is installed between the stainless steel liner plate and the concrete wall of the silo and is fixed by epoxy resin adhesive.

[0010] In one embodiment of the present invention, the weighing sensor is a shear beam structure with a rated load of 10 tons and a nonlinear error of less than 0.02%. It is installed at four symmetrical support points of the conical discharge port support beam and connected by flange bolts.

[0011] As one embodiment of the present invention, the temperature and humidity sensing node includes a capacitive humidity sensor and a platinum resistance temperature sensor. The humidity measurement range is 0 to 100% relative humidity with an accuracy of ±2% relative humidity, and the temperature measurement range is -20 to 80 degrees Celsius with an accuracy of ±0.3 degrees Celsius.

[0012] As one embodiment of the present invention, the historical stable morphology library is constructed through long-term operational data accumulation. Each benchmark morphology contains no less than 100 sets of stable state samples collected under different charging rates, coal particle sizes, and moisture contents. Each set of samples contains complete three-dimensional morphological data, pressure distribution maps, and environmental parameters.

[0013] As one embodiment of the present invention, the pattern matching uses a dynamic time warping algorithm to calculate the similarity between the current form and the reference form, and the Euclidean distance is the square root of the sum of squares of the positional deviations of the corresponding voxel points in the normalized coordinate system.

[0014] In one embodiment of the present invention, the first preset threshold is set to 0.15 meters, and the second preset threshold is set to 80% of the rated unloading load.

[0015] As one embodiment of the present invention, the vibration frequency range of the unblocking strategy generation module is 5 to 30 Hz, the air cannon spray angle range is 30 to 75 degrees, and the discharge port opening adjustment step is 5.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention constructs a joint characterization model of the three-dimensional morphology and mechanical state of a coal pile by integrating laser point cloud, pressure distribution and load sensing data. It breaks through the limitation of traditional single-point material level measurement that cannot reflect the overall coal pile structure and achieves high-precision measurement with a reserve calculation error of less than 2%.

[0017] 2. This invention introduces a stability discrimination function based on physical parameters, which can quantitatively identify local slope over-limit and stress concentration areas, and effectively warn of the risk of cantilever formation.

[0018] 3. This invention can distinguish between normal fluctuations and abnormal blockage conditions by matching patterns with a historical stable pattern library, thus avoiding false alarms. The response time for identifying bottom blockage is less than 3 minutes.

[0019] 4. The blockage clearing strategy generated by this invention is directly linked to the early warning level and equipment control parameters, providing executable instructions for the automated blockage clearing system and significantly improving the safety and reliability of silo operation. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention.

[0021] Figure 2 This is a schematic diagram of the core principle framework of the three-dimensional morphology-mechanical state joint characterization model of coal pile in this invention.

[0022] Figure 3This is a logical flowchart of the multi-source heterogeneous sensor data acquisition and preprocessing in this invention.

[0023] Figure 4 This is a logical flowchart of the high-precision three-dimensional reconstruction of the coal pile surface based on adaptive mesh refinement in this invention.

[0024] Figure 5 This is a logical flowchart of the abnormal working condition identification and hierarchical early warning generation in this invention.

[0025] Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the internal sensing unit of the silo and the external control actuator in this invention. Detailed Implementation

[0026] Please refer to the attached document. Figure 1 To be continued Figure 6 This invention provides a method and system for silo safety monitoring, aiming to solve the problem that traditional silo level gauges can only obtain single-point material level information and cannot reflect the overall three-dimensional shape and mechanical state of the coal pile. In the prior art, the storage measurement of silos relies on ultrasonic or radar level gauges, whose measurement results are significantly affected by coal surface undulations, dust interference, and local arching, resulting in calculation errors often exceeding 10%. At the same time, due to the lack of coordinated perception of the internal stress distribution and surface geometric features of the coal body, it is difficult to identify potential safety hazards such as sidewall adhesion, bottom blockage, and local arching. To address the above deficiencies, this invention deploys a multi-source heterogeneous sensor array inside the silo to simultaneously collect point cloud data of the coal pile surface, lateral pressure distribution of the inner wall, vertical load of the unloading port, and environmental temperature and humidity field data. It constructs a joint characterization model of the three-dimensional shape and mechanical state of the coal pile, and based on this model, achieves high-precision storage measurement, stability discrimination, and graded early warning of abnormal operating conditions.

[0027] The silo safety monitoring method includes the following steps: S1, at least 8 laser scanning units are evenly distributed circumferentially below the dome structure inside the silo.

[0028] S2, embed at least three layers of annular pressure sensing strips in layers along the height direction on the inner wall of the silo.

[0029] S3, No fewer than 4 weighing sensors are installed above the conical discharge port at the bottom of the silo.

[0030] S4, at least three temperature and humidity sensing nodes are installed inside the silo.

[0031] S5 performs denoising, filtering, and coordinate system unification processing on the original point cloud data to generate a standardized point cloud dataset.

[0032] S6, based on a standardized point cloud dataset, uses an octree spatial segmentation method to construct an initial voxel mesh and dynamically adjusts the voxel resolution according to the point cloud density gradient. Local mesh refinement is implemented at the edge of the coal pile and on the slope to generate a high-precision 3D model of the coal pile surface.

[0033] S7 aligns the high-precision three-dimensional model of the coal pile surface with the lateral pressure distribution data and vertical load data in time and space to construct a joint characterization model of the three-dimensional morphology and mechanical state of the coal pile.

[0034] S8, based on the joint characterization model, calculates the local slope angle, curvature change rate, and stress concentration factor of each region of the coal pile.

[0035] S9. Based on the physical parameters of the coal body, including friction angle, cohesion, and bulk density, a coal pile stability discrimination function is established.

[0036] S10 performs pattern matching between the current three-dimensional shape of the coal pile and the pre-stored historical stable shape library.

[0037] S11 generates a level 1, level 2, or level 3 early warning signal based on the matching deviation and stability determination results; S12 generates corresponding congestion clearing strategy suggestions based on the warning level.

[0038] In step S1, the laser scanning units are installed on a steel structure support below the silo dome, evenly distributed at 45-degree intervals along the circumference, totaling 8 units. Each laser scanning unit uses a pulsed laser with a wavelength of 905 nanometers, emitting a fan-shaped laser beam with a horizontal field of view of 360 degrees and a vertical field of view of 60 degrees. The scanning frequency is 20 Hz, and the ranging accuracy is ±2 mm. The laser beam is projected downwards onto the surface of the coal pile, and the return time of the reflected light is measured using the time-of-flight method to obtain the spatial coordinates of each point on the coal pile surface, forming the raw point cloud data. All laser scanning units share the same time reference and are triggered by a central synchronization controller to ensure that each unit completes scanning at the same time, avoiding data misalignment caused by dynamic changes in the coal pile. The raw point cloud data includes three-dimensional coordinates (x, y, z) and corresponding timestamps, in an uncompressed floating-point sequence format, with a sampling period of 50 milliseconds.

[0039] In step S2, the annular pressure sensing strip is arranged in three layers along the height of the silo's inner wall, located at 25%, 50%, and 75% of the total silo height, respectively. Each layer of the pressure sensing strip consists of 16 piezoresistive thin-film sensors, uniformly embedded at 22.5-degree intervals along the circumference. The piezoresistive thin-film sensors have a range of 0 to 50 kPa, a sensitivity of 0.1000 Pa, a response time of 5 milliseconds, and are encapsulated in a flexible polyimide substrate with a thickness not exceeding 0.5 mm. The sensors are installed in the interlayer between the stainless steel lining plate and the concrete wall of the silo's inner wall, fixed with epoxy resin adhesive to ensure direct contact with the coal and deformation in response to coal pressure changes. Each sensor outputs an analog voltage signal, which is converted from analog to digital and uploaded to the data acquisition module in digital form, with a sampling frequency of 100 Hz. The three layers of pressure sensing strips together constitute a two-dimensional pressure distribution map in both the circumferential and axial directions, used to characterize the static and dynamic pressure distribution characteristics of the coal on the silo's sidewall.

[0040] In step S3, the load cell is a shear beam structure with a rated load of 10 tons and a nonlinear error of less than 0.02%. It is installed at four symmetrical support points of the conical discharge port support beam and rigidly connected by flange bolts. The load cell monitors the vertical load exerted by the coal on the discharge port area in real time. This load reflects the compaction degree and flow resistance of the bottom coal. When the coal flows normally, the vertical load decreases monotonically with the discharge process; if bottom blockage occurs, the vertical load remains high or even fluctuates slightly, while the coal pile surface does not show a significant drop. The load cell output signal, after temperature compensation and zero-point calibration, is uploaded to the central processing unit at a frequency of 20 times per second.

[0041] In step S4, the temperature and humidity sensing nodes are deployed at three typical locations inside the silo: upper, middle, and lower, near the dome, the surface of the middle coal pile, and above the unloading port, respectively. Each temperature and humidity sensing node includes a capacitive humidity sensor and a platinum resistance temperature sensor. The humidity measurement range is 0 to 100% relative humidity with an accuracy of ±2% relative humidity, and the temperature measurement range is -20 to 80 degrees Celsius with an accuracy of ±0.3 degrees Celsius. The temperature and humidity data are used to correct for air refractive index deviations in laser ranging and serve as auxiliary parameters for assessing coal moisture content and flowability. All sensing nodes communicate via an industrial-grade RS485 bus, with a data update cycle of 1 second.

[0042] In step S5, the raw point cloud data is preprocessed. First, outlier removal is performed using a statistical filtering algorithm to calculate the average distance between each point and its 30 nearest neighbors. If this distance is greater than the mean plus twice the standard deviation, it is identified as a noise point and removed. Next, moving least squares smoothing filtering is applied to locally fit the point cloud surface to a spherical neighborhood with a radius of 50 mm, suppressing high-frequency jitter. Finally, coordinate system one is implemented, transforming the point cloud data collected by the eight laser scanning units from the local coordinate system to a global coordinate system with the silo's central axis as the Z-axis and the ground as the XY plane using a pre-calibrated extrinsic parameter matrix. The calibration process uses a checkerboard target and an iterative nearest-neighbor algorithm to ensure that the coordinate transformation error between units is less than 0.5 mm. After the above processing, a standardized point cloud dataset is generated, containing no fewer than 500,000 valid points with an average point spacing of 20 mm.

[0043] In step S6, a high-precision 3D model of the coal pile surface is generated based on a standardized point cloud dataset. First, an initial voxel mesh is constructed using an octree spatial segmentation method, dividing the internal space of the silo into cubic voxels with a side length of 200 mm. All point cloud data is traversed, and the number of points within each voxel is counted. If the number of points exceeds a threshold of 5, it is marked as an occupied voxel. Then, adaptive mesh refinement is performed: the point cloud density gradient between each occupied voxel and its neighboring voxels is calculated. If the absolute value of the gradient exceeds a preset threshold of 0.3, the voxel is subdivided into 8 equal parts, generating sub-voxels with a side length of 100 mm. At the edge region of the coal pile, i.e., at the interface where the voxel occupancy status abruptly changes from "occupied" to "empty," further subdivision to a resolution of 50 mm is performed. This process is recursively executed until the maximum subdivision level reaches level 4 or the voxel side length is less than 25 mm. Finally, isosurfaces are extracted using the moving cube algorithm to generate a triangular mesh model. The number of vertices is controlled to within 200,000, and the facet normals are continuously differentiable to meet the requirements of subsequent curvature calculations.

[0044] In step S7, a joint characterization model of the coal pile's three-dimensional morphology and mechanical state is constructed. First, the high-precision three-dimensional model of the coal pile surface is voxelized to generate a three-dimensional voxel field corresponding to the spatial location of the pressure sensing strips. The measured values ​​of the three-layer annular pressure sensing strips are extended to the entire inner wall surface of the silo through radial basis function interpolation, forming a continuous lateral pressure distribution field. Simultaneously, the vertical load data from the four weighing sensors are mapped to the bottom voxels of the unloading port area to construct the bottom pressure boundary conditions. Then, spatiotemporal alignment is performed: using the laser scanning timestamp as a reference, linear interpolation is performed on the pressure and load data to place them in the same time slice as the point cloud data. Finally, the joint characterization model consists of three parts: the geometric part is a triangular mesh surface model, and the mechanical part consists of the lateral pressure field and the bottom load field. The two are coupled through a shared voxel space, with each voxel storing its surface normal vector, local curvature, lateral pressure value, and bottom support state.

[0045] In step S8, key stability indices are calculated based on the joint characterization model. The local slope angle is determined by the angle between the normal vector of the triangular mesh patch and the vertical direction, and the calculation formula is as follows: ,in The unit normal vector of the surface. The vector is a vertical unit vector. The rate of change of curvature is calculated using discrete differential geometry, calculating the Gaussian curvature and mean curvature of each vertex, and determining the magnitude of the curvature gradient between adjacent surfaces. The stress concentration factor is defined as... ,in This represents the measured lateral pressure at a specific location on a voxel. This represents the average circumferential pressure at the same height. This coefficient reflects whether the local coal seam is under abnormally high pressure; the stress concentration factor in a typical cantilever region can reach over 1.5.

[0046] In step S9, a stability discrimination function for the coal pile is established. This function is based on the Mohr-Coulomb strength theory, combined with the physical parameters of the coal body. Let the friction angle of the coal body be... Cohesion is The bulk density is For any area on the surface of a coal pile, if the local slope angle... And stress concentration factor If the value is greater than 1.2, the region is determined to be in a critical state of instability. This criterion comprehensively considers the dual risks of geometric overload and mechanical overload, avoiding misjudgment based on a single indicator. Friction angle With cohesion Through laboratory triaxial shear tests, a parameter database was established for different coal types, and the corresponding parameters were automatically called up according to the current coal type during operation.

[0047] In step S10, the current three-dimensional morphology of the coal pile is pattern-matched with a historical stable morphology database. The historical stable morphology database contains three baseline morphologies: normal flow morphology, fully emptied morphology, and uniform packing morphology. Each morphology consists of no fewer than 100 samples, each recording the stable state under specific charging rates, coal particle sizes, and moisture content conditions, including a complete three-dimensional surface model, pressure distribution map, and environmental parameters. Pattern matching employs a dynamic time warping algorithm, aligning the voxel occupancy sequence of the current morphology with the sequences of each baseline morphology, and calculating the square root of the sum of squares of the positional deviations of the corresponding voxel points in the normalized coordinate system, i.e., the Euclidean distance D. The coordinate system normalization process scales the silo diameter to a unit length and the height to 1, eliminating scale effects.

[0048] In step S11, a graded early warning signal is generated. If the Euclidean distance between the current shape and any reference shape... If the vertical load at the bottom is consistently above 80% of the rated unloading load, and the coal pile surface height decreases by less than 5 millimeters within 3 consecutive minutes, it is considered a bottom blockage, triggering a Level 3 warning. All warning signals are transmitted to the central monitoring platform via industrial Ethernet and trigger audible and visual alarms.

[0049] In step S12, a blockage-clearing strategy recommendation is generated based on the warning level. A Level 1 warning recommends initiating low-frequency vibration at a frequency of 5 Hz, applied to the vibrator on the outer wall of the silo. A Level 2 warning recommends activating the air cannon system, setting the air cannon's spray angle to 415 degrees, targeting the stress concentration area, and applying a spray pressure of 0.6 MPa. A Level 3 warning recommends simultaneously executing vibration and air cannon operations, increasing the vibration frequency to 20 Hz, adjusting the air cannon's spray angle to 60 degrees, and increasing the discharge port opening by 10%. All strategy parameters are output to the actuator controller in digital command form, achieving closed-loop linkage.

[0050] The silo safety monitoring system includes a multi-source sensor data acquisition module, a point cloud preprocessing and 3D reconstruction module, a joint characterization modeling module, a stability analysis module, an anomaly identification and early warning module, and a blockage clearing strategy generation module. The multi-source sensor data acquisition module integrates a laser scanning unit, pressure sensing belt, weighing sensor, and temperature and humidity sensing nodes, and is connected to the central processing unit via an industrial fieldbus. The point cloud preprocessing and 3D reconstruction module is deployed on an embedded graphics processing unit, possessing parallel computing capabilities and capable of completing point cloud filtering and mesh generation within two seconds. The joint characterization modeling module runs on a real-time operating system, ensuring data alignment latency is less than 100 milliseconds. The stability analysis module has a built-in coal body parameter database that supports online updates. The anomaly identification and early warning module adopts a state machine architecture, executing judgments according to S10 to S11 logic. The blockage clearing strategy generation module outputs standardized control commands, compatible with mainstream PLC protocols.

[0051] This embodiment achieves a silo storage measurement error of less than 2%, an arch identification accuracy of over 95%, and a bottom blockage response time of less than 3 minutes through the above-described method and system, which is significantly better than the traditional single-point material level monitoring scheme.

Claims

1. A method of silo safety monitoring, characterized in that The application relates to a coal pile stability monitoring method, which comprises the following steps: acquiring original point cloud data of a coal pile surface, monitoring lateral pressure distribution data of a coal body on an inner wall of a silo, monitoring vertical load of a coal body on an unloading port area, collecting environmental temperature and humidity data in the silo; performing denoising, filtering and coordinate system unification processing on the original point cloud data to generate a standardized point cloud data set; based on the standardized point cloud data set, an initial voxel grid is constructed by adopting an octree space segmentation method, and the voxel resolution is dynamically adjusted according to a point cloud density gradient, local grid refinement is implemented in the edge and slope surface areas of the coal pile, and a high-precision coal pile surface three-dimensional model is generated; the high-precision coal pile surface three-dimensional model is spatiotemporally aligned with the lateral pressure distribution data and the vertical load data to construct a coal pile three-dimensional morphology-mechanical state joint representation model; based on the joint representation model, the local slope angle, the curvature change rate and the stress concentration coefficient of each area of the coal pile are calculated; a coal pile stability discrimination function is established, and the stability discrimination function is defined as: when the local slope angle is greater than the friction angle and the stress concentration coefficient exceeds a preset threshold value, it is determined that the area is in a critical unstable state; the current coal pile three-dimensional morphology is matched with a pre-stored historical stable morphology library, and the historical stable morphology library comprises three reference morphologies, namely normal flow, complete emptying and uniform accumulation; if the Euclidean distance between the current morphology and any reference morphology exceeds a first preset threshold value, a first-level early warning is triggered; if the critical unstable state determination condition is met at the same time, a second-level early warning is triggered; if the vertical load at the bottom continuously exceeds a second preset threshold value and the coal pile surface does not significantly decrease during the unloading process, it is determined that the bottom is blocked, and a third-level early warning is triggered; according to the warning level, a corresponding unblocking strategy suggestion is generated.

2. The silo safety monitoring method of claim 1, wherein, The original point cloud data is denoised, filtered and processed in a coordinate system to generate a standardized point cloud data set, which comprises the following steps: a statistical filtering algorithm is adopted to remove outliers, the average distance of each point and 30 nearest neighbor points in the neighborhood is calculated, and if the distance is greater than the mean value plus twice the standard deviation, the point is removed; a moving least square smoothing filter is adopted to locally fit the point cloud surface and suppress high-frequency jitter; through a pre-calibrated external parameter matrix, the point cloud data collected by each laser scanning unit in a local coordinate system is converted into a global coordinate system with the silo center axis as the Z axis and the ground as the XY plane to generate the standardized point cloud data set.

3. The silo safety monitoring method of claim 2, wherein, Based on the standardized point cloud data set, an initial voxel grid is constructed by adopting an octree space segmentation method, and the voxel resolution is dynamically adjusted according to a point cloud density gradient, local grid refinement is implemented in the edge and slope surface areas of the coal pile, and a high-precision coal pile surface three-dimensional model is generated, which comprises the following steps: the internal space of the silo is divided into cubic voxels with a side length of 200 mm, the number of points in each voxel is counted, and if the number of points is greater than a threshold value 5, the voxel is marked as an occupied voxel; the point cloud density gradient of each occupied voxel and its adjacent voxels is calculated, and if the gradient absolute value exceeds a preset threshold value 0.3, the voxel is 8-subdivided; in the edge area of the coal pile, that is, at the interface where the voxel occupation state changes from "occupied" to "empty", further subdivision is performed to a 50 mm resolution; An isosurface is extracted by a marching cubes algorithm, and a triangular mesh model is generated as the high-precision coal pile surface three-dimensional model.

4. The silo safety monitoring method of claim 3, wherein, The high-precision coal pile surface three-dimensional model is spatiotemporally aligned with the lateral pressure distribution data and the vertical load data to construct a coal pile three-dimensional morphology-mechanical state joint representation model, including: The high-precision coal pile surface three-dimensional model is voxelized to generate a three-dimensional voxel field corresponding to the spatial position of the pressure sensing belt; The measurement values of the three-layer annular pressure sensing belt are extended to the entire silo inner wall surface through radial basis function interpolation to form a continuous lateral pressure distribution field; The vertical load data of the four load cells are mapped to the bottom surface voxels in the discharge port area to construct the bottom pressure boundary condition; The pressure and load data are linearly interpolated based on the timestamp of the laser scanning to make them in the same time slice as the point cloud data, and the joint representation model is constructed.

5. The silo safety monitoring method of claim 4, wherein, Based on the joint representation model, the local slope angle, curvature change rate and stress concentration coefficient of each region of the coal pile are calculated, including the following steps: The local slope angle is determined by the included angle between the normal vector of the triangular mesh patch and the vertical direction, and the calculation formula is wherein is the unit normal vector of the patch, is the unit vertical vector; The Gaussian curvature and mean curvature of each vertex are calculated using the discrete differential geometry method, and the curvature gradient amplitude between adjacent facets is taken as the curvature change rate; The stress concentration factor is defined as wherein is the measured lateral pressure at the position of the voxel, is the average circumferential pressure at the same height.

6. The silo safety monitoring method of claim 5, wherein, According to the coal body physical parameters including the friction angle, cohesion and bulk density, a coal pile stability discrimination function is established, including the following steps: A friction angle and cohesion parameter database is established for different coal types, and the corresponding parameters are automatically called according to the current coal type during operation; For any region on the surface of the coal pile, if the local slope angle and the stress concentration coefficient >1.2, then it is determined that the region is in a critical state of instability.

7. The silo safety monitoring method of claim 6, wherein, The current coal pile three-dimensional morphology is matched with the pre-stored historical stable morphology library, including: Each reference morphology in the historical stable morphology library contains not less than 100 groups of stable state samples collected under different charging rates, coal particle sizes and moisture contents; The voxel occupancy sequence of the current morphology is aligned with the sequences of each reference morphology using the dynamic time warping algorithm; The square root of the sum of the position deviations of the corresponding voxels in the normalized coordinate system is taken as the Euclidean distance, and the diameter of the silo is scaled to unit length and the height is scaled to 1 in the normalized coordinate system.

8. The silo safety monitoring method of claim 7, wherein, If the Euclidean distance between the current morphology and any reference morphology exceeds a first preset threshold, a first-level warning is triggered; if the instability critical state judgment condition is also met, a second-level warning is triggered; If the bottom vertical load is continuously higher than a second preset threshold and the coal pile surface does not significantly decrease during the discharging process, it is determined that the bottom is blocked, and a third-level warning is triggered, including: The first preset threshold is set to 0.15 meters; The second preset threshold is set to 80% of the rated discharge load; If the coal pile surface height decreases by less than 5 mm in 3 consecutive minutes and the bottom vertical load is continuously higher than the second preset threshold during the discharging process, it is determined that the bottom is blocked.

9. The silo safety monitoring method of claim 8, wherein, According to the warning level, corresponding unblocking strategy suggestions are generated, including: When the first-level warning is triggered, a low-frequency vibration instruction with a vibration frequency of 5 Hz is output; When the second-level warning is triggered, an air cannon operation instruction with an air cannon injection angle of 45 degrees and an injection pressure of 0.6 MPa is output; When the third-level warning is triggered, a composite unblocking instruction with a vibration frequency of 20 Hz, an air cannon injection angle of 60 degrees and an increase of 10% in the opening degree of the discharge port is output.

10. A silo safety monitoring system characterized by, including: A multi-source sensing data acquisition module is configured to uniformly arrange no less than eight laser scanning units circumferentially below the internal dome structure of the silo, embed no less than three layers of annular pressure sensing belts in the silo wall along the height direction, with each layer of the pressure sensing belt composed of no less than 16 piezoresistive film sensors, arrange no less than four weighing sensors above the conical discharge port at the bottom of the silo, and arrange no less than three temperature and humidity sensing nodes in the internal space of the silo to respectively collect point cloud data of the coal pile surface, lateral pressure distribution data, vertical load data, and environmental temperature and humidity data. A point cloud preprocessing and three-dimensional reconstruction module is configured to perform denoising, filtering, and coordinate system unification processing on the original point cloud data, generate a standardized point cloud dataset, and generate a high-precision coal pile surface three-dimensional model by using an octree space segmentation and adaptive grid refinement algorithm. A joint representation modeling module is configured to perform spatio-temporal alignment of the high-precision coal pile surface three-dimensional model with the lateral pressure distribution data and the vertical load data, and construct a coal pile three-dimensional morphology-mechanical state joint representation model. A stability analysis module is configured to calculate local slope angle, curvature change rate, and stress concentration coefficient based on the joint representation model, establish a stability discrimination function according to coal body physical parameters, and perform instability critical state determination. An abnormality identification and early warning module is configured to perform pattern matching of the current coal pile three-dimensional morphology with a historical stable morphology library, generate a first, second, or third early warning signal according to the matching deviation and the stability determination result. A blockage clearing strategy generation module is configured to output corresponding vibration frequency, air cannon injection angle, and discharge port opening adjustment parameters according to the early warning level.

Images