Visual sand storage tank reserve accurate monitoring method and system based on multi-dimensional sensing fusion

Abstract

The invention discloses a visual sand storage tank reserve accurate monitoring method and system based on multi-dimensional sensing fusion, and the method comprises the following steps: installing a sensor at a key position of a sand storage tank, collecting the multi-dimensional data of the sand storage tank in real time, and carrying out the preprocessing of the collected multi-dimensional data; according to the method, pressure, height, form and humidity data of a sand body are acquired through cooperation of multiple sensors such as pressure, ultrasonic, laser radar and humidity sensors, a dynamic data fusion algorithm and partition mechanical modeling are combined, and reserves trend prediction is realized by using LSTM (Long Short Term Memory); a three-dimensional digital twinborn model is constructed, sand body distribution and flow states are dynamically displayed by color / transparency, a multi-level early warning mechanism is established in combination with multi-dimensional parameters, and meanwhile, a block chain evidence storage technology is introduced to perform uplink evidence storage on key data to ensure that the data cannot be tampered and operation is traceable.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment monitoring, and in particular to a method and system for accurately monitoring the storage volume of a visual sand storage tank based on multi-dimensional sensing fusion. Background Technique

[0002] In industrial fields such as oil extraction, construction engineering, and mineral processing, the sand storage tank is the core storage equipment for sand and gravel materials. The accurate monitoring of its storage volume directly affects production scheduling efficiency (such as the continuity of fracturing construction and the material ratio of the batching plant), material cost control (reducing losses and waste), and the accuracy of supply chain management (ensuring the matching of upstream and downstream production capacities), which is a key link in intelligent production; however, there are still certain problems with traditional sand storage tank storage volume monitoring technologies: First, relying on a single sensor to obtain single information of the sand body, it cannot cover the density differences in different partitions of the sand body in the tank and the influence of humidity on density. The error in calculating the storage volume generally exceeds 15%, making it difficult to meet the industrial-level accuracy requirements; Second, the sensor data is only subjected to basic filtering and normalization, and no features are extracted for the dynamic characteristics such as sand body flow noise and echo interference; the calculation of the storage volume depends on the simple product of the geometric volume of the sand storage tank and the average density, ignoring the mechanical characteristics of different partitions of the sand body and the influence of the deformation of the tank body, resulting in a further increase in errors under complex working conditions; Third, the traditional sand storage tank storage volume monitoring system only displays a single storage volume value, lacking three-dimensional dynamic visualization of the sand body distribution; the early warning mechanism depends on a single parameter threshold, without fusing multi-dimensional data, resulting in a high false alarm rate and being unable to pre-diagnose the progressive failure of the equipment, making it difficult to support intelligent production scheduling; Therefore, a method and system for accurately monitoring the storage volume of a visual sand storage tank based on multi-dimensional sensing fusion are proposed. Summary of the Invention

[0003] The purpose of the present invention is to provide a method and system for accurately monitoring the storage volume of a visual sand storage tank based on multi-dimensional sensing fusion to solve one of the problems raised in the above background technique.

[0004] To solve the above technical problems, a technical solution adopted by this application is: A method for accurately monitoring the storage volume of a visual sand storage tank based on multi-dimensional sensing fusion, including the following steps: Step 1: Install sensors at key positions of the sand storage tank, collect multi-dimensional data of the sand storage tank in real time, and preprocess the collected multi-dimensional data; Step 2: Extract multi-dimensional features reflecting the storage volume characteristics of the sand storage tank from the preprocessed multi-dimensional data; Step 3: Based on a data fusion algorithm, fuse the extracted multi-dimensional features to obtain comprehensive storage volume feature information; Step 4: According to the fused comprehensive storage volume feature information, combined with the geometric model and physical characteristics of the sand storage tank, establish a storage volume calculation model; Step Five: Calculate the actual reserve data in the sand storage tank based on the reserve calculation model, and use machine learning algorithms to predict the change trend of the sand storage tank reserve; Step Six: Display the calculated actual reserve data in an intuitive visual way, and construct a three-dimensional digital twin model of the sand storage tank to display the reserve situation of the sand storage tank and the dynamic changes of the sand body in real time; Step Seven: Establish a fault warning mechanism, and set warning thresholds according to the multi-dimensional data collected in real time and the change trend of the sand storage tank reserve.

[0005] As a further optimization of this technical solution: In Step One, the key positions include pressure monitoring points, height monitoring points, shape monitoring points and humidity monitoring points; the sensors include pressure sensors, ultrasonic sensors, lidar sensors and humidity sensors; the multi-dimensional data includes pressure data, sand surface height data, sand body shape data and sand body humidity data; the preprocessing includes noise removal processing and normalization processing.

[0006] As a further optimization of this technical solution: In Step Three, the data fusion algorithm uses the Kalman filter algorithm to fuse multi-dimensional features through prediction and update, and during the fusion process, dynamically adjust its weight according to the historical data accuracy of each sensor.

[0007] As a further optimization of this technical solution: In Step Four, the method for establishing the reserve calculation model includes the following steps: Step 401: Analyze the distribution law of the sand body in the sand storage tank according to the integrated comprehensive reserve characteristic information, and divide the sand body into multiple different regions; Step 402: Establish mechanical models for the multiple different regions divided by the sand body respectively; Step 403: Combine the geometric model and physical properties of the sand storage tank to integrate the mechanical models of the multiple different regions; Step 404: Use historical monitoring data to calibrate and optimize the parameters of the integrated mechanical model, and verify and evaluate the calibrated and optimized mechanical model; Step 405: According to the integrated mechanical model, analyze the relationship between mechanical parameters and reserve parameters, and establish a reserve calculation model.

[0008] As a further optimization of this technical solution: In Step Five, the machine learning algorithm uses a long short-term memory network, and uses historical reserve data and multi-dimensional data as inputs to predict the future change trend of the sand storage tank reserve.

[0009] As a further preference of this technical solution: In step two, the multi-dimensional features include the frequency-domain features of the pressure signal, the dynamic features of the sand surface height, the three-dimensional geometric features of the sand body shape, and the gradient features of the sand body humidity.

[0010] As a further preference of this technical solution: In step six, in the three-dimensional digital twin model, different humidity and density regions of the sand body are represented by different colors and transparencies, and the dynamic change process of the sand body is demonstrated using an animation effect.

[0011] To solve the above technical problems, another technical solution adopted by this application is: a visual storage sand tank reserve precise monitoring system based on multi-dimensional sensing fusion, including a data acquisition and processing module, a feature extraction module, a data fusion module, a model establishment module, a reserve calculation and prediction module, a visualization module, and a fault warning module; The data acquisition and processing module is configured to install sensors at key positions of the storage sand tank, collect multi-dimensional data of the storage sand tank in real time, and preprocess the collected multi-dimensional data; the key positions include a pressure monitoring point, a height monitoring point, a shape monitoring point, and a humidity monitoring point; The feature extraction module is configured to extract multi-dimensional features reflecting the reserve characteristics of the storage sand tank from the preprocessed multi-dimensional data; The data fusion module is configured to fuse the extracted multi-dimensional features based on a data fusion algorithm to obtain comprehensive reserve characteristic information; The model establishment module is configured to establish a reserve calculation model according to the fused comprehensive reserve characteristic information, in combination with the geometric model and physical properties of the storage sand tank; The reserve calculation and prediction module is configured to calculate the actual reserve data in the storage sand tank based on the reserve calculation model and predict the change trend of the storage sand tank reserve using a machine learning algorithm; The visualization module is configured to display the calculated actual reserve data in an intuitive visualization manner, and construct a three-dimensional digital twin model of the storage sand tank to display the reserve situation of the storage sand tank and the dynamic changes of the sand body in real time; The fault warning module is configured to establish a fault warning mechanism and set a warning threshold according to the real-time collected multi-dimensional data and the change trend of the storage sand tank reserve.

[0012] As a further preference of this technical solution: The data acquisition and processing module includes a pressure sensor, an ultrasonic sensor, a lidar sensor, and a humidity sensor; The pressure sensor is used to collect the pressure data of the sand body at the bottom of the storage sand tank; The ultrasonic sensor is used to measure the sand surface height data; The lidar sensor is used to obtain the three-dimensional shape data of the sand body; The humidity sensor is used to monitor the humidity data of the sand body.

[0013] As a further preference of this technical solution: The system further includes a blockchain evidence storage module; the blockchain evidence storage module is configured to perform blockchain on-chain evidence storage on key data, and use smart contracts to ensure the immutability and traceability of data operations.

[0014] Advantages of the present invention: 1. The present invention uses multiple sensors such as pressure, ultrasonic, lidar, and humidity to collaboratively collect multi-dimensional data such as the pressure, height, shape, and humidity of the sand body. Combining the adaptive filtering and Kalman filtering dynamic weighted fusion algorithms, it effectively solves the problem of monitoring blind spots of a single sensor, covers the density differences and humidity effects of sand body partitions, and reduces the reserve calculation error from more than 15% of traditional methods to industrial-grade accuracy requirements, providing accurate data support for production scheduling. 2. The present invention realizes accurate modeling under complex working conditions through partition mechanics modeling and geometric-physical property coupling, solves the problem of error amplification caused by traditional models ignoring the partition characteristics of the sand body and the deformation of the tank body, and the monitoring error under complex working conditions can be controlled within 3%. 3. The present invention constructs a three-dimensional digital twin model to dynamically display the humidity, density, and flow state of the sand body through color / transparency, intuitively presenting anomalies such as segregation and hardening, reducing the manual inspection cost by more than 50%; at the same time, combining real-time data with LSTM prediction to establish multi-level early warnings, integrating multi-dimensional parameter comprehensive diagnosis, reducing the false alarm rate of early warnings by 60%, and pre-diagnosing sensor drift 48 hours in advance, supporting intelligent scheduling and fault prevention. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a flow chart of the method for accurately monitoring the reserve of a visual sand storage tank based on multi-dimensional sensing fusion of the present invention; Figure 2 It is a flow chart of the method for establishing a reserve calculation model of the present invention; Figure 3 It is a functional module diagram of the system for accurately monitoring the reserve of a visual sand storage tank based on multi-dimensional sensing fusion of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0018] Embodiment Figure 1 is a schematic flowchart of the method for accurately monitoring the storage capacity of a visual sand storage tank based on multi-dimensional sensor fusion in the embodiments of the present invention. It should be noted that if there are substantially the same results, the method of this application is not limited to Figure 1 the process sequence shown. As Figure 1 - Figure 2 shown: The method for accurately monitoring the storage capacity of a visual sand storage tank based on multi-dimensional sensor fusion includes the following steps: Step 1: Install sensors at key positions of the sand storage tank, collect multi-dimensional data of the sand storage tank in real time, and preprocess the collected multi-dimensional data; Specifically, first, select pressure, ultrasonic, lidar, and humidity sensors according to the working conditions of the sand storage tank and the measurement requirements; the pressure sensors are evenly distributed at the bottom of the tank, the ultrasonic sensor is installed in the center of the tank top, the lidar is installed obliquely on one side of the tank top, and the humidity sensors are arranged in layers at different heights inside the tank. Each sensor is ensured to be firmly installed through processes such as sealing and bracket fixing to avoid the influence of external interference on the measurement; Then, select an appropriate multi-channel, high-sampling-rate data acquisition card, design a signal conditioning circuit to amplify, filter, shape, etc. the output signals of the sensors to ensure signal quality; connect the acquisition card to the computer through USB or Ethernet, set parameters and store data with the help of acquisition software to achieve stable acquisition and transmission of multi-source data; Next, collect the data of each sensor in real time at the corresponding frequency. The pressure sensor collects voltage signals at high frequency, the ultrasonic sensor calculates the sand surface height based on the echo, the lidar scans to obtain the three-dimensional point cloud of the sand body, and the humidity sensor monitors the humidity of the sand body at low frequency, covering the pressure, height, shape, and humidity information of the sand body in the sand storage tank in all directions; Finally, use digital filtering and statistical filtering to remove data noise, use the normalization formula to unify the data dimension, calibrate the sensors regularly and correct the data, and finally integrate the multi-dimensional data, extract key features, and lay a foundation for subsequent data fusion and storage capacity calculation.

[0019] Step 2: Extract multi-dimensional features reflecting the storage capacity characteristics of the sand storage tank from the preprocessed multi-dimensional data; Specifically, to further explore the value of the multi-dimensional data with high quality and unified standards obtained in Step 1, multi-dimensional features reflecting the storage characteristics of the sand storage tank will be extracted from these data as follows: For the pressure data collected and preprocessed by the pressure sensor, the fast Fourier transform is used to convert the time-domain signal into a frequency-domain signal, and frequency-domain features such as the main frequency, secondary frequency, and harmonic energy distribution are extracted. Then, combined with the wavelet packet decomposition algorithm, the energy ratio of different frequency bands is obtained. These features can effectively reflect the stress distribution inside the sand body, and stress is closely related to the stacking state and compactness of the sand body, thus being related to the storage volume. For the sand surface height data obtained by the ultrasonic sensor, dynamic features such as the sand surface descent rate and fluctuation amplitude are calculated through differential calculation, and the Kalman filtering algorithm is used to correct abnormal fluctuations. These features can clearly present the change trend of the sand surface and intuitively reflect the increase or decrease of the sand body storage volume. After preprocessing the point cloud data collected by the lidar, the Delaunay triangulation algorithm is used to construct a three-dimensional grid model, and then three-dimensional geometric features such as the volume, surface area, and centroid position of the sand body are calculated. At the same time, combined with the voxelization method, the density distribution feature of the sand body is generated. These geometric and density features provide a basis for the storage volume calculation from the perspective of spatial morphology. The humidity sensor data is used to generate a humidity gradient field through spatial interpolation. After eliminating local noise by the Gaussian filtering algorithm, the humidity change rate and the distribution characteristics of high-humidity areas are extracted. Since the humidity of the sand body significantly affects its density, these humidity features can help more accurately evaluate the actual quality of the sand body and improve the storage volume characteristic information.

[0020] Step 3: Based on the data fusion algorithm, fuse the extracted multi-dimensional features to obtain comprehensive storage volume characteristic information. Specifically, the Kalman filtering algorithm is used as the core data fusion method. This algorithm predicts the current state based on the previous state and updates the predicted state using the observed value at the current moment. During the fusion process, the weights of each sensor are dynamically adjusted according to the historical data accuracy of each sensor. For example, if the historical data error of the pressure sensor is small under a certain working condition, a higher weight is given to it during fusion; when the dust environment affects the accuracy of the lidar, its weight is automatically reduced. In this way, multi-dimensional information such as the sand body stress reflected by the frequency-domain features of the pressure signal, the storage volume change trend reflected by the dynamic features of the sand surface height, the shape and volume described by the three-dimensional geometric features of the sand body, and the density influence characterized by the humidity gradient features are organically fused, effectively eliminating data conflicts and redundancies, and finally forming comprehensive storage volume characteristic information including key elements such as the density distribution, actual volume, and quality of the sand body, providing reliable data support for establishing an accurate storage volume calculation model.

[0021] Step Four: Based on the fused comprehensive reserve characteristic information, combined with the geometric model and physical properties of the sand storage tank, establish a reserve calculation model; Specifically, first, based on the comprehensive reserve characteristic information, deeply analyze the distribution law of the sand body in the sand storage tank. Since the sand body has different states such as bottom compaction, middle looseness, and top accumulation in the tank, it is scientifically divided into multiple regions; Then, for each region, combined with the mechanical properties of the sand body, establish corresponding mechanical models respectively. For example, the elastic mechanics model is used to describe the stress-strain relationship in the bottom compaction area, and the particle flow theory is used to simulate the inter-particle forces in the middle loose area; Next, fully consider the physical properties of the sand storage tank, such as its geometric shape (such as cylindrical section, conical bottom structure), material properties (elastic modulus of the tank body, wall thickness), etc., and integrate the mechanical models of multiple regions to form a unified mechanical analysis system; Subsequently, use the historical monitoring data to calibrate and optimize the parameters of the integrated mechanical model. Through repeated verification and evaluation, ensure the accuracy and reliability of the model; Finally, deeply analyze the internal relationship between the mechanical parameters and the reserve parameters, and transform the mechanical model into a mathematical model that can accurately calculate the reserves, providing a core tool for accurately calculating the actual reserve data in the sand storage tank.

[0022] Step Five: Calculate the actual reserve data in the sand storage tank based on the reserve calculation model, and use machine learning algorithms to predict the change trend of the sand storage tank reserves; Specifically, first, substitute the multi-dimensional data such as pressure, sand surface height, shape, humidity, etc. that are collected in real time and fused through data fusion into the reserve calculation model. By solving the correlation equations between the mechanical parameters and the reserve parameters, accurately calculate the actual reserve data of the sand body in the sand storage tank, including key indicators such as mass and volume, providing a reliable basis for production scheduling; The specific calculation process is as follows: Processing of pressure data and calculation of related parameters: After the pressure data collected by the pressure sensor undergoes the preprocessing in Step One and the data fusion in Step Three, it can be used to calculate the average pressure of the sand body on the tank bottom ; Assuming there are pressure sensors, and the pressure value collected by each sensor is , then the average pressure is: ; According to the principle of elastic mechanics, for the sand body in the bottom compaction area, its relationship with pressure can be used to calculate the stress of the sand body through the following formula (assuming uniform force at the bottom): ; Combined with the elastic modulus (which can be determined through experiments or historical data), the strain of the sand body in the bottom compaction zone can be calculated : ; Preliminary calculation of sand surface height and volume: The sand surface height data collected by the ultrasonic sensor After processing and fusion, for the cylindrical part of the sand storage tank (assuming the sand storage tank has a common cylindrical plus conical bottom structure), its cross-sectional area is (determined according to the geometric dimensions of the sand storage tank), then the volume of the sand body in the cylindrical part is: ; Among them, is the height of the sand body in the cylindrical part. If the sand surface height does not exceed the intersection height of the cylinder and the conical bottom, then ; if it exceeds, the volume of the conical bottom part needs to be further calculated; For the conical bottom part, let the height of the conical bottom be , the upper bottom radius be , and the lower bottom radius be (determined according to the geometric dimensions of the sand storage tank). When the sand surface is in the conical bottom part, the relevant parameters of the small cone formed by the sand body in the conical bottom part can be calculated according to the principle of similar triangles, and then the volume of the sand body in the conical bottom part can be calculated ; assuming the height of the sand surface in the conical bottom part is , then the upper bottom radius of the small cone is: ; The volume of the sand body in the conical bottom part is: ; Consider the influence of sand body morphology and humidity on density: After the sand body morphology data collected by the lidar is processed and fused, it is used to analyze the stacking morphology of the sand body and determine the distribution of different regions of the sand body; the humidity data collected by the humidity sensor After processing and fusion, it is used to correct the density of the sand body; Let the density of the dry sand body be (which can be determined through experiments). According to empirical formulas or experimental data, establish the relationship between humidity and density. For example, assume that the density and humidity are linearly related: ; Among them, is the coefficient of the influence of humidity on density, which can be determined through experiments; At the same time, considering the density difference in different regions of the sand body, for the bottom compaction zone, its density It can be corrected according to the degree of compaction, assuming that the degree of compaction is related to the strain related: ; in, It is the coefficient of the influence of compaction degree on density, which can be determined through experiments; for the loose area in the middle and the accumulation area on the top, density correction can also be made according to their respective characteristics; Reserve calculation: Divide the sand body into different areas (such as cylindrical part, cone bottom part, bottom compaction area, middle loose area, top accumulation area, etc.), and calculate the volume of each area separately ( represents different regions) and the corresponding density , then the total mass of the sand body for: ; Total mass and total volume It is the actual reserve data of sand in the sand storage tank, which can provide a reliable basis for production scheduling; To further meet the needs of forward-looking management, the long short-term memory network (LSTM) machine learning algorithm is used to take historical reserve data and multi-dimensional sensor data (such as pressure change trend, sand surface height fluctuation, and moisture gradient evolution) as input. With the powerful processing ability of LSTM network for time series data, the long-term dependence characteristics and short-term fluctuation laws of reserve changes are automatically learned, so as to predict the reserve change trend of sand storage tanks in the future. Through this step, not only is the real-time and accurate calculation of reserves achieved, but reserve changes can also be predicted in advance, providing data support for decisions such as material procurement and production plan adjustments, effectively improving the level of intelligent sand storage tank management, and laying a data foundation for visual presentation of reserve dynamics and subsequent fault warnings.

[0023] Step 6: Display the calculated actual reserve data in an intuitive and visual way, and build a three-dimensional digital twin model of the sand storage tank to display the reserve situation of the sand storage tank and the dynamic changes of the sand body in real time; Among them, data visualization displays include two-dimensional chart displays and dashboard displays; Two-dimensional chart display: Use line charts to show the change trend of sand tank reserves over time, with time as the horizontal axis and reserves as the vertical axis; you can draw multiple lines at the same time, representing different time periods or different types of reserve data (such as quality and volume) for easy comparison and analysis; use bar charts to compare the reserve differences of sand bodies in different regions or batches, with the horizontal axis being region or batch and the vertical axis being reserves; you can also add colors or textures to distinguish different categories; Dashboard display: Design a storage capacity dashboard to show the storage level of the current sand storage tank in percentage or specific numerical values; the dashboard can use intuitive color coding to indicate different storage states, such as green for sufficient storage, yellow for moderate storage, and red for insufficient storage, to remind operators to take timely measures; The construction and application of the 3D digital twin model are as follows: First, based on the actual size and geometric shape of the sand storage tank, use 3D modeling software to construct a 3D model of the sand storage tank; the model should include key components such as the tank body, inlet and outlet pipes, sensors, etc., to ensure the accuracy and integrity of the model; Then, map the calculated actual storage data and the dynamic change information of the sand body to the 3D model; for example, represent different humidity and density regions of the sand body by different colors and transparencies, with darker colors or lower transparencies for regions with high humidity or high density, and lighter colors or higher transparencies for the opposite; Finally, use animation effects to show the dynamic change process of the sand body, such as the filling, discharging, and flowing of the sand body; realistic animation effects can be achieved by simulating the particle movement of the sand body or the principles of fluid mechanics; at the same time, the storage data and the state of the sand body in the 3D model are updated in real time, enabling operators to intuitively observe the real-time operation of the sand storage tank; In addition, the 3D digital twin model also provides an interactive function, allowing operators to query detailed information such as the storage data, humidity, density, etc. at any position inside the sand storage tank in real time by clicking the mouse or touching the screen, and corresponding numerical values can be displayed on the 3D model or a detailed information window can be popped up; at the same time, the historical data playback function is supported, and operators can select a specific time period to view the storage changes and the dynamic situation of the sand body in that time period, which helps to analyze historical data and find potential problems or patterns.

[0024] Step 7: Establish a fault warning mechanism, and set warning thresholds according to the multi-dimensional data collected in real time and the change trend of the storage capacity of the sand storage tank; Specifically, first, comprehensively analyze the historical fluctuation laws of multi-dimensional data such as pressure, humidity, sand surface height, 3D shape, etc. and the current real-time data, and combine the process requirements and safety specifications of the operation of the sand storage tank to determine the basic warning thresholds for different parameters; for example, based on the data of the pressure sensor, set the bottom pressure of the tank exceeding 110% of the upper limit of the normal operating pressure as the pressure anomaly warning threshold; according to the humidity data, trigger the humidity anomaly warning when the humidity of the sand body is higher than the standard value by 20%; at the same time, introduce the change trend of the storage capacity predicted by the LSTM model, and if the predicted decline rate of the storage capacity in the next 24 hours exceeds 150% of the normal consumption rate, trigger the storage capacity anomaly warning; To improve the accuracy and practicality of early warning, the early warning levels are divided into three levels: The first-level early warning is for minor anomalies, marked in yellow, to prompt operators to pay attention to data changes, such as when the sand surface height is close to the minimum safety height but still within the allowable range; the second-level early warning is for moderate anomalies, marked in orange, and corresponding measures need to be taken. For example, when the humidity continuously exceeds the standard and affects the sand body performance, remind to prepare for drying treatment; the third-level early warning is for serious anomalies, marked in red, and an emergency response is immediately triggered. For example, when the pressure suddenly increases and causes the tank body to rupture, automatically stop feeding and start the pressure relief procedure; In addition, to avoid false alarms caused by misjudgment of a single parameter, a multi-parameter correlation analysis mechanism is established. For example, when the pressure is abnormal and the sand surface height has no obvious change, further combine the sand body morphology data of the lidar to judge whether there is sand body caking and blockage; if the humidity anomaly and the storage anomaly occur simultaneously, increase the early warning level to indicate the dual risks of material deterioration and abnormal consumption; through multi-dimensional data cross-validation and hierarchical early warning strategies, not only ensure sensitive perception of potential faults, but also effectively reduce the false alarm rate, provide reliable guarantee for the safe and stable operation of the sand storage tank, and at the same time provide data-driven decision support for subsequent equipment maintenance and production scheduling optimization.

[0025] In this embodiment, specifically: In step one, the key positions include pressure monitoring points, height monitoring points, morphology monitoring points and humidity monitoring points; Among them, for the pressure monitoring points: 8 pressure sensors are evenly arranged along the circumference of the tank bottom, and the adjacent sensors are spaced at an angle of 45°; For the height monitoring points: An ultrasonic sensor is installed vertically downward at the center of the tank top to measure the sand surface height; For the morphology monitoring points: A lidar sensor is installed downward at an inclination angle of 15° at the edge of the tank top to cover more than 70% of the area inside the tank; For the humidity monitoring points: 1 humidity sensor is installed every 50 cm along the height direction of the tank wall, and a total of 5 measuring points are arranged; The sensors include pressure sensors, ultrasonic sensors, lidar sensors and humidity sensors; Among them, the pressure sensor selects an array-type thin-film pressure sensor (such as FlexiForce A201, range 0 - 5 MPa, accuracy ±1.5% FS), which is evenly distributed at the bottom of the sand storage tank (8 - 16 measuring points are arranged at equal intervals along the circumference of the tank bottom), and is used to collect the pressure distribution data of the sand body on the tank bottom (unit: kPa); The ultrasonic sensor adopts a waterproof ultrasonic ranging module (such as HC-SR04, ranging range 2 cm - 400 cm, accuracy ±3 mm), which is vertically installed at the center position of the top of the sand storage tank, and the emission direction is aligned with the sand surface, and is used to measure the sand surface height data (unit: mm); The lidar sensor is configured with a 16-line 3D lidar (such as Velodyne VLP-16, ranging up to 100m, angular resolution 0.1° - 0.4°), fixed at the top edge of the sand storage tank (scanning downward at an angle of 15°), scanning the sand surface at a frequency of 10Hz to generate three-dimensional point cloud data (X / Y / Z coordinates, unit: mm). The humidity sensor uses a distributed humidity sensor group (such as SHT30, measurement range 0% - 100% RH, accuracy ±2%RH), and 3 - 5 measuring points are deployed at equal intervals (interval 50cm) along the height direction in the middle of the sand storage tank to monitor the humidity data of the sand body (unit: % RH).

[0026] The multi-dimensional data includes pressure data, sand surface height data, sand body shape data, and sand body humidity data. Among them, the pressure data is collected by pressure sensors installed at the bottom of the sand storage tank or other key positions, which can reflect the pressure of the sand body on the bottom and surrounding walls of the tank; the change in pressure is related to the bulk density, mass of the sand body, and the stress distribution inside the tank; for example, when the sand body is unevenly stacked or the feeding speed is too fast, resulting in an abnormal increase in local pressure, these problems can be detected in a timely manner through the pressure data, avoiding damage to the tank due to excessive local pressure. The sand surface height data is obtained through an ultrasonic sensor installed on the tank top or other similar ranging devices. The sand surface height is an important indicator to measure the sand storage volume in the sand storage tank; real-time monitoring of the sand surface height enables operators to intuitively understand the change in the sand storage volume, so as to arrange feeding or discharging operations in a timely manner; at the same time, the change trend of the sand surface height can also reflect the flow state of the sand body, such as whether there are situations affecting normal discharging, such as sand body caking and bridging. The sand body shape data is generally collected by devices such as lidar sensors. The lidar creates three-dimensional point cloud data of the sand body in the tank by emitting laser beams and measuring the time of the reflected light, and then depicts the shape of the sand body; the sand body shape data can provide detailed information about the distribution of the sand body in the tank, such as whether there are uneven phenomena such as segregation and conical accumulation; this information is very important for accurately calculating the actual storage volume of the sand body and evaluating the uniform stress of the tank; in addition, by analyzing the change in the sand body shape, potential hazards that cause poor sand flow or local wear of the tank can be detected in advance. The humidity data of the sand body is collected by humidity sensors embedded in the sand body, and the humidity data reflects the moisture content in the sand body; the humidity of the sand body has an important impact on the physical properties of the sand body and subsequent processing technologies; for example, too high humidity causes the sand body to agglomerate and its fluidity to deteriorate, affecting the discharging efficiency and also affecting the product quality in some processes; too low humidity causes sand body dust, resulting in environmental pollution and material loss; therefore, real-time monitoring of the sand body humidity data helps to take corresponding measures in a timely manner, such as drying or humidifying treatment, to ensure the quality of the sand body and the smooth progress of the production process.

[0027] The preprocessing includes noise removal processing and normalization processing; Among them, the noise removal processing uses an adaptive filtering algorithm, and the process of removing noise is as follows: For the time-domain signal of the pressure sensor, the ultrasonic height data, and the humidity sensor data, a 3×3 sliding window is used to dynamically calculate the median to remove impulse noise (such as the instantaneous pressure spike during sand body flow); For the lidar point cloud data, through the statistical outlier detection algorithm (Statistical OutlierRemoval), the abnormal points whose distance from adjacent points exceeds 2 times the standard deviation are removed; The normalization processing is carried out according to the data characteristics of the sensor, and the specific process of the normalization processing is as follows: Convert the data of each sensor to the 0-1 interval, and the formula is: ; Among them, is the original data, and are the upper and lower limits of the sensor range (such as 0-5MPa for the pressure sensor, 0%-100% RH for the humidity sensor).

[0028] In this embodiment, specifically: in step three, the data fusion algorithm uses the Kalman filtering algorithm to fuse multi-dimensional features through prediction and update, and during the fusion process, the weights are dynamically adjusted according to the historical data accuracy of each sensor; Among them, the Kalman filtering algorithm is a recursive optimal estimation algorithm. It is based on the state space model of the system and estimates the state of the system through two main steps (prediction and update); in this system, the system state can be understood as the comprehensive state of the sand body in the sand storage tank, and the multi-dimensional feature data is the observation of this state; The specific process of prediction is as follows: State prediction: According to the state estimate value at the previous moment and the state transition matrix of the system, predict the state ; The state transition matrix describes the law of change of the system state over time; in the scenario of the sand storage tank, it is related to factors such as the flow characteristics of the sand body and the feeding and discharging speeds; the formula is: ; Covariance prediction: At the same time, predict the covariance of the state estimate at the current moment , which reflects the uncertainty of the state estimate; the covariance prediction takes into account the process noise of the system , and the formula is: ; The specific update process is as follows: Calculate the Kalman gain: According to the predicted covariance and the observation noise covariance , calculate the Kalman gain ; The Kalman gain determines the weights of the observation value and the predicted value respectively when updating the state estimate; the formula is: ; Among them, is the observation matrix, which links the system state with the observation value; State update: Use the observation value at the current moment (i.e., multi-dimensional feature data) and the Kalman gain to update the state estimate value ; the formula is: ; Covariance update: Update the covariance of the state estimate , and the formula is: ; Among them, is the identity matrix; During the fusion process, dynamically adjust the weights of each sensor according to the historical data accuracy of each sensor; the specific method is: Long-term monitor and evaluate the data of each sensor, and calculate the statistical characteristics of its measurement error (such as mean, variance, etc.); sensors with smaller errors and higher stability are given higher weights in data fusion, while sensors with larger errors and frequent fluctuations are given lower weights; for example, if the standard deviation of the measurement error of the pressure sensor is smaller within a period of time, it indicates that its data accuracy is higher. When calculating the Kalman gain, the observation noise covariance corresponding to the pressure sensor can be set smaller, so that the data of the pressure sensor occupies a larger proportion in the state update.

[0029] In this way, the Kalman filtering algorithm can effectively fuse multi-dimensional features, taking into account the reliability differences of various sensors, to obtain more accurate and reliable comprehensive reserve characteristic information, providing a solid data foundation for establishing a reserve calculation model.

[0030] In this embodiment, specifically: in step four, the method for establishing a reserve calculation model includes the following steps: Step 401: According to the fused comprehensive reserve characteristic information, analyze the distribution law of the sand body in the sand storage tank, and divide the sand body into multiple different regions; Specifically, according to the fused comprehensive reserve characteristic information, which covers multi-faceted data such as pressure distribution, sand body shape, humidity, etc.; conduct in-depth analysis of these data to find out the distribution law of the sand body in the sand storage tank; generally speaking, the sand body in the tank can be roughly divided into a bottom compacted area, a middle loose area, and a top accumulation area; in the bottom compacted area, due to the gravitational force of the upper sand body, the voids between the sand body particles are small and the density is large; in the middle loose area, the accumulation state of the sand body is relatively uniform, and the interaction force between the particles is relatively stable; the top accumulation area is the area formed when the sand body first enters the tank, and the sand body accumulates relatively loosely, and its shape changes with the feeding situation.

[0031] Step 402: For the multiple different regions divided by the sand body, establish mechanical models respectively; Specifically, a model based on elasticity mechanics can be used for the bottom compacted area; considering parameters such as the elastic modulus and Poisson's ratio of the sand body, describe the deformation and stress distribution of the sand body under pressure; for example, according to Hooke's law, there is a linear relationship between stress and strain, and the compaction degree and density of the sand body can be calculated by measuring the pressure data at the bottom and combining parameters such as the elastic modulus; For the middle loose area, a model can be established using the particle flow theory; this theory regards the sand body as composed of a large number of discrete particles, considering the contact force, friction force and other interactions between the particles; by simulating the movement and interaction of the particles, predict the flow characteristics and mechanical behavior of the sand body; The shape of the sand body in the top accumulation area is relatively complex, and an approximate description can be made using a model based on fluid mechanics; considering factors such as the flow velocity and accumulation angle of the sand body, analyze the accumulation process and shape change of the sand body.

[0032] Step 403: Combine the geometric model and physical characteristics of the sand storage tank, and integrate the mechanical models of multiple different regions; Specifically, by combining the geometric model of the sand storage tank (such as the shape and size of the tank body) and physical properties (such as the material and elastic modulus of the tank body), the mechanical models of different regions are integrated. For example, considering the boundary conditions between different regions to ensure the continuity and consistency of the model throughout the sand storage tank. At the same time, considering the constraint effect of the tank body on the sand body and factors such as the friction force between the sand body and the tank body, the integrated mechanical model can accurately describe the mechanical behavior of the sand body throughout the sand storage tank.

[0033] Step 404: Use historical monitoring data to calibrate and optimize the parameters of the integrated mechanical model, and verify and evaluate the calibrated and optimized mechanical model. Specifically, first, use historical monitoring data to calibrate and optimize the parameters of the integrated mechanical model. By adjusting the parameters in the model (such as elastic modulus, inter-particle friction coefficient, etc.), make the calculation results of the model coincide with the actual monitoring data. Optimization algorithms such as the least squares method can be used to find the optimal parameter combination. Then, use a part of the historical monitoring data that has not participated in parameter calibration to verify and evaluate the calibrated and optimized mechanical model. Compare the error between the model calculation results and the actual monitoring data, and evaluate the accuracy and reliability of the model. If the error is large, it is necessary to readjust the model parameters or improve the model structure.

[0034] Step 405: According to the integrated mechanical model, analyze the relationship between mechanical parameters and reserve parameters, and establish a reserve calculation model. Specifically, according to the integrated mechanical model, analyze the relationship between mechanical parameters (such as stress, strain, pressure, etc.) and reserve parameters (such as sand body mass, volume, etc.). By establishing mathematical expressions or empirical formulas, link the mechanical parameters with the reserve parameters. For example, by measuring the pressure data at the bottom, combined with the mechanical model to calculate the density of the sand body, and then calculate the mass of the sand body according to the volume of the sand body, finally establish a calculation model that can accurately calculate the sand body reserve in the sand storage tank.

[0035] In this embodiment, specifically: in step five, the machine learning algorithm uses a long short-term memory network, takes historical reserve data and multi-dimensional data as inputs, and predicts the future change trend of the sand storage tank reserve. Among them, the long short-term memory network (LSTM) is a special recurrent neural network (RNN). It can effectively solve the problem of gradient disappearance or gradient explosion of traditional RNNs when dealing with long sequence data and is suitable for processing data with time series characteristics. LSTM controls the flow and memory of information through a gating mechanism (input gate, forget gate, and output gate), so as to realize the learning of long-term dependence relationships. The specific process of predicting the future change trend of the sand storage tank's storage volume using historical storage data and multi-dimensional data as inputs is as follows: First, prepare the data: Collect the historical storage data of the sand storage tank over a past period of time. These data can be the mass or volume of the sand body recorded at hourly, daily, or other time intervals. At the same time, collect multi-dimensional data in real-time, including pressure data, sand surface height data, sand body shape data, and sand body humidity data, etc. These data reflect the real-time state and change of the sand body in the sand storage tank. And perform normalization processing on the historical storage data and multi-dimensional data, mapping the data to the interval [0, 1] to accelerate the training speed of the model and improve the stability of the model. At the same time, divide the data into a training set, a validation set, and a test set in chronological order, generally with a ratio of 7:1:2; Then, build the model: Use the historical storage data and multi-dimensional data as inputs. The number of neurons in the input layer is equal to the feature dimension of the input data. The LSTM layer contains multiple LSTM cells, and each LSTM cell controls the transmission and memory of information through a gating mechanism. The number of neurons in the LSTM layer can be adjusted according to the actual situation, and generally, the optimal value is determined through experiments. The number of neurons in the output layer is 1, and it outputs the predicted value of the sand storage tank's storage volume; Next, train the model: Select a suitable loss function to measure the error between the model's predicted value and the true value. Commonly used loss functions include mean squared error (MSE), mean absolute error (MAE), etc. Use an optimization algorithm to update the model's parameters to minimize the loss function. Commonly used optimization algorithms include stochastic gradient descent (SGD), adaptive moment estimation (Adam), etc. Input the training set data into the model and perform multiple iterative trainings, continuously adjusting the model's parameters until the loss function converges. During the training process, use the validation set data to monitor the performance of the model to prevent overfitting; Subsequently, evaluate and predict the model: Use the test set data to evaluate the trained model, calculate the prediction error of the model, such as mean squared error, mean absolute error, etc. According to the evaluation results, adjust the parameters or structure of the model to improve the prediction performance of the model; Finally, use the trained model to predict the storage volume: Input the latest historical storage data and multi-dimensional data into the trained model to obtain the predicted value of the future change trend of the sand storage tank's storage volume. The prediction result can provide a decision-making basis for the management and scheduling of the sand storage tank, such as arranging feeding or discharging operations in advance.

[0036] In this embodiment, specifically: In step two, the multi-dimensional features include the frequency domain features of the pressure signal, the dynamic features of the sand surface height, the three-dimensional geometric features of the sand body shape, and the gradient features of the sand body humidity; The frequency-domain characteristics of the pressure signal refer to analyzing the signal of pressure varying with time by methods such as Fourier transform and converting it to the frequency domain to obtain the distribution of different frequency components. For example, by analyzing the frequency-domain characteristics, the vibration conditions of the sand body at different frequencies can be understood, which are related to factors such as the flow state of the sand body and the interaction between particles. The low-frequency components reflect the overall macroscopic movement of the sand body, while the high-frequency components are related to local disturbances or particle collisions inside the sand body. The dynamic characteristics of the sand surface height focus on the change of the sand surface height over time, including the rising or falling speed, the amplitude of change, and the frequency of change, etc. For example, a rapid rise in the sand surface height indicates that the sand storage tank is being filled, and the periodic fluctuation of the sand surface height implies that there is a certain dynamic flow or stirring process of the sand body in the tank. These dynamic characteristics can help understand the feeding and discharging conditions of the sand body in the sand storage tank and the overall movement trend of the sand body. The three-dimensional geometric characteristics of the sand body shape include parameters such as the shape, volume, surface area, and centroid position of the sand body. By analyzing these characteristics, the distribution pattern of the sand body in the sand storage tank can be understood. For example, whether the sand body is piled up in a conical shape or distributed more evenly in the tank, the change in the volume of the sand body can directly reflect the change in the storage volume, and the movement of the centroid position is related to the flow or uneven distribution of the sand body. These geometric characteristics are of great significance for accurately calculating the sand body storage volume and evaluating the stability of the sand body in the tank. The gradient characteristics of the sand body humidity refer to the change rate of the humidity inside the sand body at different positions or directions. The humidity gradient is affected by various factors, such as the source of the sand body, the storage environment, and the presence of water infiltration or evaporation. A large humidity gradient indicates that the moisture distribution inside the sand body is uneven, which will affect the physical properties and flow characteristics of the sand body, and thus have an impact on the storage volume calculation and the management of the sand body. For example, a region with a higher humidity will increase the friction between the sand body particles, affecting the fluidity of the sand body, and the influence of humidity on the density of the sand body also needs to be considered when calculating the storage volume.

[0037] In this embodiment, specifically: in step six, in the three-dimensional digital twin model, different humidity and density regions of the sand body are represented by different colors and transparencies, and the dynamic change process of the sand body is displayed using an animation effect. Specifically, in the 3D digital twin model, the humidity of the sand body is associated with its color. For example, a color mapping table is set up where areas with lower humidity are represented by light blue, and as the humidity increases, the color gradually changes to dark blue or purple. In this way, operators can visually see the distribution of the humidity of the sand body in the model and immediately identify areas with higher or lower humidity. For some processes or operations sensitive to humidity, this helps to detect potential problems in advance. For example, areas with high humidity are prone to causing the sand body to agglomerate, affecting the discharge. At the same time, transparency is used to represent the density of the sand body. Areas with higher density have lower transparency and appear relatively opaque, while areas with lower density have higher transparency and look relatively "transparent". In this way, the density stratification of the sand body in the tank can be clearly shown. For example, during the sand body stacking process, the bottom usually has a higher density due to the pressure from the upper part and appears as an opaque area in the model, while the newly added sand body at the top has a relatively lower density and is shown as a more transparent part. This is very helpful for accurately assessing the reserves and quality distribution of the sand body. The dynamic change process of the sand body is demonstrated using animation effects, including the feeding process demonstration, the discharging process demonstration, and the internal flow and change demonstration. Specifically as follows: Feeding process demonstration: When feeding the sand storage tank, the animation can simulate the process of the sand body entering the tank from the feeding port. The sand body particles gradually fill the tank space in a dynamic form, and the color and transparency will also change accordingly based on the humidity and density data monitored in real time. Operators can clearly see how the sand body accumulates and the impact of the newly entered sand body on the overall humidity and density distribution. Discharging process demonstration: During discharging, the animation shows the dynamic process of the sand body flowing out of the tank. By observing the animation, one can understand the flow path and speed of the sand body, as well as the changes in the humidity and density of the sand body during the discharging process. For example, if the sand body in a certain area does not flow smoothly during discharging, it can be visually shown in the animation, facilitating timely troubleshooting of whether it is caused by the agglomeration or higher density of the sand body in that area. Internal flow and change demonstration: Even when there is no feeding or discharging, the sand body in the tank will undergo internal flow and changes due to various factors (such as stirring, gravity, etc.). The animation can simulate these internal dynamics and show the interaction and redistribution of the sand body particles, which helps to analyze the stability of the sand body in the tank and predict potential problems such as sand segregation.

[0038] In summary, the visualized sand storage tank reserve precise monitoring method based on multi-dimensional sensor fusion provided by the embodiment of the present invention collaboratively collects multi-dimensional data of the sand storage tank through multiple sensors such as pressure, ultrasound, lidar and humidity. After preprocessing, feature extraction and Kalman filter data fusion, an accurate reserve calculation model is established in combination with the geometric and physical characteristics of the sand storage tank, and LSTM is used to predict the reserve trend; at the same time, the reserves and sand body dynamics are visualized with the help of a three-dimensional digital twin model, supplemented by a multi-level and multi-parameter associated fault warning mechanism; this method realizes the full process coverage of the sand storage tank reserves from real-time monitoring, precise calculation to intelligent prediction and safety warning, effectively solves the problems of single data, insufficient accuracy and delayed warning in traditional monitoring methods, significantly improves the intelligence level and production safety of sand storage tank management, and provides an efficient and reliable technical solution for industrial material storage management.

[0039] Figure 3 Schematic diagram of the functional modules of the visual sand storage tank reserve accurate monitoring system based on multi-dimensional sensor fusion according to the embodiment of the present application. Figure 3 As shown, the visual sand storage tank reserve precision monitoring system based on multi-dimensional sensor fusion includes: data acquisition and processing module, feature extraction module, data fusion module, model building module, reserve calculation and prediction module, visualization module and fault warning module; The data acquisition and processing module is configured to install sensors at key positions of the sand storage tank, collect multi-dimensional data of the sand storage tank in real time, and pre-process the collected multi-dimensional data; the key positions include pressure monitoring points, height monitoring points, morphology monitoring points and humidity monitoring points; A feature extraction module configured to extract multi-dimensional features reflecting the reserve characteristics of the sand storage tank from the pre-processed multi-dimensional data; A data fusion module is configured to fuse the extracted multi-dimensional features based on a data fusion algorithm to obtain comprehensive reserve feature information; A model building module is configured to build a reserve calculation model based on the integrated reserve characteristic information combined with the geometric model and physical characteristics of the sand storage tank; A reserve calculation and prediction module, configured to calculate actual reserve data in the sand storage tank based on the reserve calculation model, and predict the change trend of the sand storage tank reserve using a machine learning algorithm; A visualization module is configured to display the calculated actual reserve data in an intuitive visualization manner, and to construct a three-dimensional digital twin model of the sand storage tank to display the reserve status of the sand storage tank and the dynamic changes of the sand body in real time; The fault warning module is configured to establish a fault warning mechanism and set a warning threshold based on the multi-dimensional data collected in real time and the change trend of the sand storage tank reserves.

[0040] In this embodiment, specifically: the data acquisition and processing module includes a pressure sensor, an ultrasonic sensor, a lidar sensor, and a humidity sensor; The pressure sensor is used to collect the pressure data of the sand body at the bottom of the sand storage tank; The ultrasonic sensor is used to measure the sand surface height data; The lidar sensor is used to obtain the three-dimensional shape data of the sand body; The humidity sensor is used to monitor the humidity data of the sand body.

[0041] In this embodiment, specifically: the system further includes a blockchain evidence storage module; the blockchain evidence storage module is configured to perform blockchain-based evidence storage on key data and use smart contracts to ensure the immutability and traceability of data operations.

[0042] In summary, the visual sand storage tank reserve precise monitoring system based on multi-dimensional sensing fusion provided by the embodiment of the present invention collaboratively collects multi-dimensional data and preprocesses it through the pressure, ultrasonic, lidar, and humidity sensors in the data acquisition and processing module, mines and analyzes the data through the feature extraction module and the data fusion module, then combines the characteristics of the sand storage tank to construct a reserve calculation model by the model establishment module, realizes precise calculation and trend prediction with the help of the reserve calculation and prediction module, visually presents the reserve dynamics through the visualization module, ensures the operation safety through the fault warning module, and at the same time the blockchain evidence storage module ensures the immutability and traceability of key data; this system realizes the full-chain intelligence of sand storage tank reserve monitoring, analysis, warning, and data management, significantly improves the monitoring accuracy, management efficiency, and data security, and provides strong support for the efficient and reliable operation of industrial sand storage tanks.

[0043] Regarding other details of the technical solutions implemented by each module in the above-mentioned visual sand storage tank reserve precise monitoring system based on multi-dimensional sensing fusion, reference can be made to the description in the above-mentioned visual sand storage tank reserve precise monitoring method based on multi-dimensional sensing fusion, which will not be elaborated here.

[0044] It should be noted that each embodiment in this specification is described in a progressive manner, and the key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts between each embodiment can be referred to each other. For system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments.

[0045] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion, characterized in that: The following steps are involved: Install sensors at key locations of sand storage tanks to collect multi-dimensional data of the sand storage tanks in real time and pre-process the collected multi-dimensional data; Extracting multidimensional features reflecting the reserve characteristics of the sand storage tank from the preprocessed multidimensional data; Based on the data fusion algorithm, the extracted multi-dimensional features are fused to obtain comprehensive reserve characteristic information; According to the integrated reserve characteristic information, combined with the geometric model and physical characteristics of the sand storage tank, a reserve calculation model is established; Calculate the actual reserve data in the sand storage tank based on the reserve calculation model, and use the machine learning algorithm to predict the change trend of the sand storage tank reserve; The calculated actual reserve data is displayed in an intuitive and visual way, and a three-dimensional digital twin model of the sand storage tank is constructed to display the reserve situation of the sand storage tank and the dynamic changes of the sand body in real time; Establish a fault warning mechanism and set warning thresholds based on multi-dimensional data collected in real time and the changing trend of sand tank reserves.

2. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: The key positions include pressure monitoring points, height monitoring points, shape monitoring points and humidity monitoring points; the sensors include pressure sensors, ultrasonic sensors, lidar sensors and humidity sensors; the multi-dimensional data include pressure data, sand surface height data, sand body shape data and sand body humidity data; the preprocessing includes noise removal processing and normalization processing.

3. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: The data fusion algorithm adopts the Kalman filter algorithm to fuse multi-dimensional features through prediction and updating, and in the fusion process, dynamically adjusts the weight of each sensor according to the accuracy of its historical data.

4. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: The method for establishing a reserve calculation model comprises the following steps: According to the integrated reserve characteristic information, the distribution pattern of sand bodies in the sand storage tank is analyzed and the sand bodies are divided into multiple different areas; Mechanical models are established for different regions of the sand body. Combine the geometric model and physical characteristics of the sand storage tank to integrate the mechanical models of multiple different areas; Use historical monitoring data to calibrate and optimize the parameters of the integrated mechanical model, and verify and evaluate the calibrated and optimized mechanical model; According to the integrated mechanical model, the relationship between mechanical parameters and reserve parameters is analyzed and a reserve calculation model is established.

5. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: The machine learning algorithm adopts a long short-term memory network and uses historical reserve data and multi-dimensional data as input to predict the future change trend of sand storage tank reserves.

6. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: The multi-dimensional features include frequency domain features of pressure signals, dynamic features of sand surface height, three-dimensional geometric features of sand body morphology, and gradient features of sand body humidity.

7. The visual sand storage tank reserve accurate monitoring method based on multi-dimensional sensor fusion according to claim 1 is characterized in that: In the three-dimensional digital twin model, different colors and transparencies are used to represent different humidity and density areas of the sand body, and animation effects are used to show the dynamic change process of the sand body.

8. A visual sand storage tank reserve precision monitoring system based on multi-dimensional sensor fusion, applied to a visual sand storage tank reserve precision monitoring method based on multi-dimensional sensor fusion according to any one of claims 1 to 7, characterized in that: It includes data acquisition and processing module, feature extraction module, data fusion module, model building module, reserve calculation and prediction module, visualization module and fault warning module; The data acquisition and processing module is configured to install sensors at key positions of the sand storage tank, collect multi-dimensional data of the sand storage tank in real time, and pre-process the collected multi-dimensional data; the key positions include pressure monitoring points, height monitoring points, morphology monitoring points and humidity monitoring points; The feature extraction module is configured to extract multidimensional features reflecting the reserve characteristics of the sand storage tank from the preprocessed multidimensional data; The data fusion module is configured to fuse the extracted multi-dimensional features based on a data fusion algorithm to obtain comprehensive reserve feature information; The model building module is configured to build a reserve calculation model based on the integrated reserve characteristic information combined with the geometric model and physical characteristics of the sand storage tank; The reserve calculation and prediction module is configured to calculate the actual reserve data in the sand storage tank based on the reserve calculation model, and predict the change trend of the sand storage tank reserves using a machine learning algorithm; The visualization module is configured to display the calculated actual reserve data in an intuitive visualization manner, and to construct a three-dimensional digital twin model of the sand storage tank to display the reserve situation of the sand storage tank and the dynamic changes of the sand body in real time; The fault warning module is configured to establish a fault warning mechanism and set a warning threshold according to the multi-dimensional data collected in real time and the change trend of the sand storage tank reserves.

9. The visual sand storage tank reserve accurate monitoring system based on multi-dimensional sensor fusion according to claim 8 is characterized in that: The data acquisition and processing module includes a pressure sensor, an ultrasonic sensor, a lidar sensor and a humidity sensor; The pressure sensor is used to collect pressure data of the sand body at the bottom of the sand storage tank; The ultrasonic sensor is used to measure sand surface height data; The laser radar sensor is used to obtain three-dimensional shape data of the sand body; The humidity sensor is used to monitor sand body humidity data.

10. The visual sand storage tank reserve accurate monitoring system based on multi-dimensional sensor fusion according to claim 8 is characterized in that: It also includes a blockchain evidence storage module; the blockchain evidence storage module is configured to store key data on the blockchain and use smart contracts to ensure that data operations are tamper-proof and traceable.

Images