Material level detection method, electronic equipment and computer program product
By setting up multiple strain sensors at different heights of the hopper, different interferences of coal density are eliminated, and accurate level detection is performed using the level prediction model, which solves the deviation problem of the fiber grating strain monitoring system, and improves detection accuracy and system stability.
Patent Information
- Application Number
- CN202510408917.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-02
- Publication Date
- 2025-07-18
AI Technical Summary
The existing fiber grating strain monitoring system has significant deviations in the material level height conversion, resulting in spillage or air shock accidents, threatening production safety and equipment operation efficiency.
Multiple strain sensors are set up at different heights of the hopper. By removing the influence of coal density differences, the level prediction regression model and fitting coefficient are used to predict the material level to improve the measurement accuracy.
It realizes high-precision level detection on different types of hoppers and coal types, and has strong adaptability and ensures the safety and stability of the monitoring system.
Smart Images
Figure CN120333582A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of material level detection, and in particular, to a material level detection method, an electronic device, and a computer program product. Background Art
[0002] During the actual operation of a car dumper, when a coal flow of several tons rushes into the coal unloading hopper at high speed, a large amount of dust is generated along with violent air flow disturbance, resulting in a poor working environment. Among the existing material level monitoring technologies, although the gamma-ray level gauge has the characteristics of strong penetration and large detection thickness, it has potential radiation safety hazards and high equipment maintenance and management costs, which is contrary to the concept of building a green port; although the capacitive level gauge is highly economical, it is easily affected by changes in the dielectric constant of the material, and the measurement stability is poor; although the heavy hammer type detection device has strong environmental adaptability, it has inherent defects such as serious mechanical wear and inability to monitor in real time; although ultrasonic and radar gauges have high measurement accuracy, they are sensitive to dust concentration, medium temperature, and material surface characteristics, and require frequent maintenance and calibration.
[0003] Compared with traditional detection methods, fiber Bragg grating strain sensors exhibit significant technical advantages: they have high sensitivity, intrinsic safety and explosion protection, strong anti-electromagnetic interference characteristics, as well as excellent long-term stability and environmental tolerance. By innovatively adopting an external installation scheme and arranging the sensors on the surface of the hopper support beam, it is possible to avoid the influence of the harsh working conditions inside the hopper on the detection elements, and at the same time, accurately invert the material level change through the structural strain, greatly reducing the maintenance difficulty while ensuring the measurement accuracy, providing an innovative solution for intelligent material level monitoring.
[0004] The existing fiber Bragg grating strain monitoring system calculates the material level height based on the strain-weight relationship. However, it is found in the use process that there is still a serious distortion in the monitoring system, which may induce overfilling or empty shock accidents, seriously threatening production safety and equipment operation efficiency. Summary of the Invention
[0005] The purpose of the embodiments of the present application is to provide a material level detection method, an electronic device, and a computer program product, so as to alleviate the problem that the material level height calculated by the existing fiber Bragg grating strain monitoring system based on the strain-weight relationship will have significant deviations, which may lead to serious distortion of the monitoring system, and then induce overfilling or empty shock accidents, seriously threatening production safety and equipment operation efficiency.
[0006] A material level detection method provided by the embodiments of the present application includes:
[0007] Obtain the current strain vector of a plurality of strain sensors set at different heights of the hopper;
[0008] For the current strain vector of each strain sensor, calculate the first strain vector after removing the influence of density;
[0009] Normalize the first strain vector of each strain sensor to obtain the second strain vector of each strain sensor;
[0010] For the second strain vectors of multiple strain sensors, use the pre-obtained material level prediction regression model and fitting coefficients to make a prediction to obtain the current material level of the hopper.
[0011] In the above technical solution, multiple strain sensors are arranged at different heights of the hopper. Obtain the current strain vectors of these strain sensors, and these strain vectors reflect the strain states of the hopper at different heights. Remove the density influence on the current strain vector of each strain sensor to obtain the first strain vector, so as to eliminate the interference of coal density difference on strain measurement and improve the measurement accuracy. Normalize the first strain vector of each strain sensor to obtain the second strain vector, so that the current material level can be predicted through the material level prediction model and fitting coefficients. The material level detection method of this embodiment is applicable to different types of hoppers and coal types by arranging multiple strain sensors at different heights of the hopper and removing the influence of coal density difference on strain measurement, and has strong adaptability and generality.
[0012] In some alternative embodiments, the strain sensor includes: a fiber Bragg grating strain sensor.
[0013] In the above technical solution, the fiber Bragg grating strain sensor has extremely high sensitivity and can accurately sense the tiny strain changes of the hopper at different heights. This high sensitivity enables the sensor to more accurately reflect the material level situation of the hopper and improve the measurement accuracy. The fiber Bragg grating strain sensor uses optical fiber to transmit signals and no electric spark is generated, so it has the characteristic of intrinsic safety and explosion protection. In environments such as ports and mines where flammable and explosive gases may exist, this characteristic is particularly important and can ensure the safe operation of the monitoring system. The optical fiber transmission signal is not affected by electromagnetic interference, so the fiber Bragg grating strain sensor has strong anti-electromagnetic interference ability. In a complex industrial environment, this characteristic can ensure the stable operation of the sensor and improve the reliability of measurement data. The fiber Bragg grating strain sensor has excellent long-term stability and environmental tolerance and can work stably for a long time under harsh working conditions. This characteristic enables the sensor to adapt to the harsh conditions of environments such as ports and mines and ensure the long-term stable operation of the monitoring system.
[0014] In some alternative embodiments, for the current strain vector of each strain sensor, calculating the first strain vector after removing the density influence includes:
[0015] Among multiple strain sensors, determine the reference strain sensor;
[0016] The ratio of the current strain vector of each strain sensor to the current strain vector of the reference strain sensor is taken as the first strain vector of each strain sensor.
[0017] The derivation process of the density correction principle involved in the above technical solution is as follows:
[0018] The basic structure of the dumper hopper is a steel structure fixed upside down on the concrete ring beam of the pit. The main load it bears is the live load generated by the loaded coal on the hopper body. When coal of different weights acts on the hopper, the stress distribution of the hopper structure changes with the change of the load. The finite element software is used to analyze the stress state of the hopper structure body to master the stress and strain distribution laws of the components at different hopper levels.
[0019] According to the Rankine and Coulomb earth pressure theories, the horizontal and vertical components of the compressive stress of the coal on the steel plate inside the hopper are respectively:
[0020]
[0021] Where: γ is the unit weight, and the unit weight = density × gravitational acceleration, that is, γ = ρg; is the internal friction angle between the material and the steel plate; θ is the angle between each side of the coal bunker hopper body and the vertical plane; h is the depth of the material contained in the hopper body. Then the resultant force of the pressure acting on the steel plate inside the hopper is:
[0022]
[0023] The direction of this resultant force acts perpendicular to the steel plate.
[0024] From the geometric and statics relationships, the cross-sectional strain is:
[0025]
[0026] Where: M is the bending moment load borne by the beam, y is the distance from the strain sensor to the neutral axis, and EI is the flexural rigidity of the cross-section. The bending moment M is a function related to the resultant force Fa and the height and position of the calculated cross-section.
[0027] It can be seen from the above formula that the strain has a non-linear relationship with the hopper height and a linear relationship with the density, that is:
[0028] ε = ρ∞h 2 ∞M(h)
[0029] When a certain kind of coal is poured into the hopper, the load and density in this state are both determined. The strain values collected by each sensor (S1, S2,..., Sm) at different height positions of the hopper structure are ε1, ε2,..., ε m , corresponding to the heights h1, h2,..., h mTaking a certain strain sensor as a reference (taking S1 as an example), project the strain values of sensors at different positions onto this sensor respectively. The strain ratio of sensors at different height positions is different, that is:
[0030]
[0031] Therefore, taking the ratio of the current strain vector of each strain sensor to the current strain vector of the reference strain sensor as the first strain vector of each strain sensor can eliminate the density parameter while retaining the information of the coal material height, so as to calculate the material level at this moment and eliminate the systematic error of the material level caused by density difference to the greatest extent.
[0032] Among them, the selection method of the reference strain sensor is as follows: install multiple fiber Bragg grating sensors at different height positions on the outer wall of the hopper. According to the results of static simulation analysis, the strain changes more with the load at positions closer to the bottom, and the strain is greater at the same height where the structural width is wider. Select the sensor with a larger strain as the reference sensor S_base.
[0033] Specifically, since each side of the hopper body is an irregular geometric body, at the same material level height of the same type of coal material, the strain ratio of sensors at different positions to the strain of the reference sensor is different. The algorithm establishes a material level monitoring model based on the ratio of sensors at multiple different positions to the reference sensor, and combines the ratio labels recorded in the calibration test to calculate the material level value after eliminating the density influence. Note that the more sensors are installed, the richer the collected information, and the more accurate the material level calculated by the material level monitoring model. Connect multiple fiber Bragg grating sensors in series to the demodulator in turn, connect the communication optical cable into the equipment room, and form an optical path to access the monitoring host.
[0034] In some optional embodiments, the number of strain sensors is m;
[0035] The method for obtaining the material level prediction regression model and the fitting coefficient includes:
[0036] After the m strain sensors synchronously collect n times, strain vectors S_1, S_2,..., S_i,..., S_n are obtained, where S_i = (S1, S2, S3,…, Sm), 1 ≤ i ≤ n, and the material level information corresponding to S_i is Y_i; the sensing data matrix of the m strain sensors is M0 = [S_1, S_2,..., S_n];
[0037] For the sensing data matrix M0, calculate the first matrix M1 after removing the density influence;
[0038] Normalize the first matrix M1 to obtain the second matrix M2;
[0039] The material level is fitted using the material level prediction regression model model, and the fitting coefficient is vec_coe. The fitting process minimizes the second norm of model(vec_coe, M2)-Y to obtain the value of the fitting coefficient.
[0040] In some alternative embodiments, the material level prediction regression model includes: a statistical regression model, a machine learning regression model, or a deep learning regression model.
[0041] In the above technical solution, the statistical regression model includes, for example, linear regression and ridge regression. Linear regression: Assume that there is a linear relationship between the independent variable and the dependent variable, and the model parameters are fitted by the least squares method; when there is a linear relationship between the material level change and the strain sensor data, linear regression is a simple and effective choice. Ridge regression: Based on linear regression, the L2 regularization term is introduced to constrain the model complexity and alleviate the problem of multicollinearity; when there is a high correlation between the strain sensor data, ridge regression can improve the stability and prediction accuracy of the model.
[0042] The machine learning regression model includes, for example, the multi-layer perceptron (MLP): It is a feedforward neural network that fits complex non-linear relationships through multi-layer non-linear transformations; when there is a complex non-linear relationship between the material level change and the strain sensor data, the multi-layer perceptron can capture these relationships and make accurate predictions; the multi-layer perceptron has strong learning and generalization abilities and can handle high-dimensional data and non-linear relationships.
[0043] The deep learning regression model includes, for example, the deep neural network (DNN): It consists of multiple hidden layers and fits complex functional relationships through multi-layer non-linear transformations; when the relationship between the material level change and the strain sensor data is very complex and the data volume is large, the deep neural network can learn the deep features in the data and make accurate predictions; the deep neural network has powerful representation and learning abilities and can handle various types of data and tasks.
[0044] Specifically, regarding the selection of the model: If the data volume is small, statistical regression models such as linear regression or ridge regression can be selected. If the data volume is large, machine learning or deep learning models such as the multi-layer perceptron or the deep neural network can be considered. If the relationship between the material level change and the strain sensor data is relatively simple (such as a linear relationship), linear regression can be selected. If the relationship is complex and difficult to capture by traditional methods, the multi-layer perceptron or the deep neural network can be considered. If the reason for the model prediction result needs to be explained, statistical regression models such as linear regression or ridge regression can be selected. If more attention is paid to the prediction accuracy rather than the model interpretability, the multi-layer perceptron or the deep neural network can be considered.
[0045] In some alternative embodiments, for the sensing data matrix M0, calculating a first matrix M1 that removes the influence of density includes:
[0046] In each acquisition, one of the m strain sensors is used as a reference strain sensor, and the strain vector of the reference strain sensor is used as a reference value. The ratio of the strain vectors of each strain sensor to the reference value is calculated to obtain the first matrix M1.
[0047] In some alternative embodiments, normalizing the first matrix M1 to obtain a second matrix M2 includes:
[0048] Obtaining the maximum value VMAXj and the minimum value VMINj of each column of the first matrix;
[0049] According to the maximum value VMAXj and the minimum value VMINj of each column of the first matrix, the strain vectors of the corresponding columns of the first matrix are normalized to obtain the second matrix.
[0050] In some alternative embodiments, normalizing the first strain vector of each strain sensor to obtain the second strain vector of each strain sensor includes:
[0051] According to the maximum value VMAXj and the minimum value VMINj, the corresponding first strain vector is normalized to obtain the second strain vector.
[0052] An electronic device provided by an embodiment of the present application includes: a processor and a memory. The memory stores machine-readable instructions executable by the processor. When the machine-readable instructions are executed by the processor, the method described in any one of the above is executed.
[0053] A computer program product provided by an embodiment of the present application includes computer programs / instructions. When the computer programs / instructions are executed by a processor, the steps of the method described in any one of the above are implemented. Description of the Drawings
[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.
[0055] Figure 1 It is a flowchart of the steps of a material level detection method provided by an embodiment of the present application;
[0056] Figure 2 It is a possible structural schematic diagram of an electronic device provided by an embodiment of the present application;
[0057] Figure 3 This is a schematic diagram of the strain sensor layout provided by the embodiments of the present application.
[0058] Icons: 1 - Processor, 2 - Memory, 3 - Communication interface, 4 - Communication bus. Detailed implementation manners
[0059] Next, the technical solutions in the embodiments of the present application will be described with reference to the accompanying drawings in the embodiments of the present application. The following embodiments are only used to illustrate the technical solutions of the present application more clearly, so they are only examples and cannot be used to limit the protection scope of the present application.
[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion.
[0061] In the description of the embodiments of the present application, technical terms such as "first" and "second" are only used to distinguish different objects and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity, specific order or primary-secondary relationship of the indicated technical features. In the description of the embodiments of the present application, "a plurality" means more than two unless otherwise specifically defined.
[0062] Referring to "embodiments" herein means that the specific features, structures or characteristics described in connection with the embodiments may be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.
[0063] The existing fiber Bragg grating strain monitoring system calculates the material level height according to the strain-weight relationship. However, it is found in the use process that there is still serious distortion in the monitoring system, which further induces overfilling or empty shock accidents, seriously threatening production safety and equipment operation efficiency.
[0064] There are various reasons for distortion, and it is difficult to identify the dominant cause of the current distortion in a specific scenario during actual use. After further research and analysis, the applicant found that although the fiber optic grating strain monitoring system can effectively sense the strain response of the support beam, the difference in coal density poses a significant challenge to the measurement accuracy. Since the bulk density of coal from different origins is affected by factors such as the type of origin, degree of metamorphism, and water content, its value range usually lies between 0.5 g / cm 3 and 1.70 g / cm 3 . When hoppers of the same volume are loaded with coal of different densities, the level height converted by the system based on the strain-weight relationship will have a significant deviation, which will further lead to serious distortion of the monitoring system, trigger overfilling or empty shock accidents, and seriously threaten production safety and the operating efficiency of equipment.
[0065] One or more embodiments of the present application aim to provide a level detection method, an electronic device, and a computer program product. By setting multiple strain sensors at different heights of the hopper and removing the density influence from the current strain vector of each strain sensor, the interference of coal density difference on strain measurement can be eliminated, and the measurement accuracy can be improved. The technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings in the embodiments of the present application.
[0066] Figure 1 The following is a flowchart of the steps of a level detection method provided for an embodiment of the present application, including:
[0067] Step 100: Obtain the current strain vectors of multiple strain sensors set at different heights of the hopper;
[0068] Among them, the multiple strain sensors need to be arranged to cover different heights, and some of the strain sensors can be arranged on different sides at the same height of the hopper. The arrangement directions of the multiple strain sensors can be the same or different. The strain sensors can be fiber optic grating strain sensors, capacitive strain sensors, piezoelectric strain sensors, resistive strain sensors, etc. The strain vector of the strain sensor includes a sensed value and a strain direction, and this strain direction is related to the arrangement direction of the strain sensor.
[0069] Step 200: For the current strain vector of each strain sensor, calculate the first strain vector after removing the density influence;
[0070] Among them, taking the ratio between two strain sensors as the first strain vector can remove the density influence of the coal in the hopper.
[0071] Step 300: Normalize the first strain vector of each strain sensor to obtain the second strain vector of each strain sensor;
[0072] Step 400: Predict the current material level of the hopper by using the pre-acquired material level prediction regression model and fitting coefficients for the second strain vectors of multiple strain sensors.
[0073] The strain sensors are arranged as Figure 3 S1 - S10 shown in the figure. Multiple strain sensors are installed on different support beams on the outer wall of the hopper or at different height positions on the same support beam.
[0074] Among them, the material level prediction regression model can adopt linear regression, ridge regression, multi - layer perceptron, deep neural network, etc. The training process of the material level prediction regression model is to adjust the model parameters to minimize the prediction error, so as to find the best fit for the data.
[0075] In the embodiments of the present application, multiple strain sensors are set at different heights of the hopper. The current strain vectors of these strain sensors are obtained, and these strain vectors reflect the strain states of the hopper at different heights. For the current strain vector of each strain sensor, the density influence is removed to obtain the first strain vector, so as to eliminate the interference of coal density difference on strain measurement and improve the measurement accuracy. The first strain vector of each strain sensor is normalized to obtain the second strain vector, so that the current material level can be predicted through the material level prediction model and fitting coefficients. The material level detection method of this embodiment is applicable to different types of hoppers and coal types by setting multiple strain sensors at different heights of the hopper and removing the influence of coal density difference on strain measurement, and has strong adaptability and versatility.
[0076] In some alternative embodiments, the strain sensor includes: a fiber Bragg grating strain sensor.
[0077] In the embodiments of the present application, the fiber Bragg grating strain sensor has extremely high sensitivity and can accurately sense the tiny strain changes of the hopper at different heights. This high sensitivity enables the sensor to more accurately reflect the material level situation of the hopper and improve the measurement accuracy. The fiber Bragg grating strain sensor uses optical fiber to transmit signals and has no electric spark generation, so it has the characteristic of intrinsically safe explosion protection. In environments such as ports and mines where flammable and explosive gases may exist, this characteristic is particularly important and can ensure the safe operation of the monitoring system. The optical fiber transmission signal is not affected by electromagnetic interference, so the fiber Bragg grating strain sensor has strong anti - electromagnetic interference ability. In a complex industrial environment, this characteristic can ensure the stable operation of the sensor and improve the reliability of the measurement data. The fiber Bragg grating strain sensor has excellent long - term stability and environmental tolerance and can work stably for a long time under harsh working conditions. This characteristic enables the sensor to adapt to the harsh conditions of environments such as ports and mines and ensure the long - term stable operation of the monitoring system.
[0078] In some alternative embodiments, the material level prediction regression model includes: a statistical regression model, a machine learning regression model, or a deep learning regression model.
[0079] In the embodiments of the present application, the statistical regression models include, for example, linear regression and ridge regression. Linear regression: Assume that there is a linear relationship between the independent variable and the dependent variable, and the model parameters are fitted by the least squares method; when there is a linear relationship between the material level change and the strain sensor data, linear regression is a simple and effective choice. Ridge regression: On the basis of linear regression, the L2 regularization term is introduced to constrain the model complexity and alleviate the problem of multicollinearity; when there is a high correlation between the strain sensor data, ridge regression can improve the stability and prediction accuracy of the model.
[0080] The machine learning regression model includes, for example, a multi-layer perceptron (MLP): It is a feedforward neural network that fits complex non-linear relationships through multi-layer non-linear transformations; when there is a complex non-linear relationship between the material level change and the strain sensor data, the multi-layer perceptron can capture these relationships and make accurate predictions; the multi-layer perceptron has strong learning and generalization abilities and can handle high-dimensional data and non-linear relationships.
[0081] The deep learning regression model includes, for example, a deep neural network (DNN): It consists of multiple hidden layers and fits complex functional relationships through multi-layer non-linear transformations; when the relationship between the material level change and the strain sensor data is very complex and the data volume is large, the deep neural network can learn the deep features in the data and make accurate predictions; the deep neural network has powerful representation and learning abilities and can handle various types of data and tasks.
[0082] Specifically, regarding the selection of the model: If the data volume is small, statistical regression models such as linear regression or ridge regression can be selected. If the data volume is large, machine learning or deep learning models such as multi-layer perceptrons or deep neural networks can be considered. If the relationship between the material level change and the strain sensor data is relatively simple (such as a linear relationship), linear regression can be selected. If the relationship is complex and difficult to capture by traditional methods, multi-layer perceptrons or deep neural networks can be considered. If the reason for the model prediction result needs to be explained, statistical regression models such as linear regression or ridge regression can be selected. If more attention is paid to the prediction accuracy and less to the model interpretability, multi-layer perceptrons or deep neural networks can be considered.
[0083] In some alternative embodiments, for the current strain vector of each strain sensor, calculating the first strain vector after removing the density influence includes:
[0084] Among the multiple strain sensors, determining a reference strain sensor;
[0085] The ratio of the current strain vector of each strain sensor to the current strain vector of the reference strain sensor is used as the first strain vector of each strain sensor.
[0086] The derivation process of the density correction principle involved in the above technical solution is as follows:
[0087] The basic structure of the tipper hopper is a steel structure that is inversely suspended and fixed on the concrete ring beam of the pit. The main load it bears is the live load generated by the loaded coal on the hopper body. When coal of different weights acts on the hopper, the stress distribution of the hopper structure changes with the change of the load. The finite element software is used to analyze the stress state of the hopper structure body to master the stress and strain distribution laws of the components at different hopper levels.
[0088] According to the Rankine and Coulomb earth pressure theories, the horizontal and vertical components of the compressive stress of the coal on the steel plate inside the hopper are respectively:
[0089]
[0090] Where: γ is the unit weight, and the unit weight = density × gravitational acceleration, that is, γ = ρg; is the internal friction angle between the material and the steel plate; θ is the angle between each side of the coal bunker hopper body and the vertical plane; h is the depth of the material contained in the hopper body. Then the resultant force of the pressure acting on the steel plate inside the hopper is:
[0091]
[0092] The direction of this resultant force acts perpendicular to the steel plate.
[0093] From the geometric and statics relationships, the cross-sectional strain is:
[0094]
[0095] Where: M is the bending moment load borne by the beam, y is the distance from the strain sensor to the neutral axis, and EI is the flexural rigidity of the cross-section. The bending moment M is a function related to the resultant force Fa and the height and position of the calculation cross-section.
[0096] It can be seen from the above formula that the strain has a non-linear relationship with the hopper height and a linear relationship with the density, that is:
[0097] ε = ρ∞h 2 ∞M(h)
[0098] When a certain kind of coal is poured into the hopper, the load and density are both determined in this state. The strain values collected by each sensor (S1, S2,..., Sm) at different height positions of the hopper structure are ε1, ε2,..., ε m , corresponding to the heights h1, h2,..., h mTaking a certain strain sensor as a reference (taking S1 as an example), project the strain values of sensors at different positions onto this sensor respectively. The strain ratio of sensors at different height positions is different, that is:
[0099]
[0100] Therefore, taking the ratio of the current strain vector of each strain sensor to the current strain vector of the reference strain sensor as the first strain vector of each strain sensor can eliminate the density parameter while retaining the information of the coal material height, so as to calculate the material level at this moment and eliminate the systematic error of the material level caused by density difference to the greatest extent.
[0101] Among them, the selection method of the reference strain sensor is as follows: install multiple fiber Bragg grating sensors at different height positions on the outer wall of the hopper. According to the results of static simulation analysis, the strain changes with the load more significantly at positions closer to the bottom, and the strain is larger at the same height where the structural width is wider. Select the sensor with a larger strain as the reference sensor S_base.
[0102] Specifically, since each side of the hopper body is an irregular geometric body, under the same coal material and the same material level height, the ratio of the strain of sensors at different positions to the strain of the reference sensor is different. The algorithm establishes a material level monitoring model based on the ratios of multiple sensors at different positions to the reference sensor, and combines the ratio tags recorded in the calibration test to calculate the material level value after eliminating the density influence. Note that the more sensors are installed, the richer the collected information is, and the more accurate the material level calculated by the material level monitoring model is. Connect multiple fiber Bragg grating sensors in series to the demodulator in turn, connect the communication optical cable into the equipment room, and form an optical path to access the monitoring host.
[0103] In the actual implementation process, first determine the material level prediction regression model and calculate the fitting coefficient, specifically including:
[0104] After the m strain sensors synchronously collect n times, the strain vectors S_1, S_2,..., S_i,..., S_n are obtained, where S_i = (S1, S2, S3,…, Sm), 1 ≤ i ≤ n, and the material level information corresponding to S_i is y_i; the sensing data matrix of the m strain sensors is M0 = [S_1, S_2,..., S_n];
[0105] Among them, the j-th column of the sensing data matrix M0 is the n times of data collected by the j-th strain sensor, and the i-th row of the sensing data matrix M0 is the data collected by the m strain sensors for the i-th time.
[0106] For the sensing data matrix M0, calculate the first matrix M1 after removing the density influence, including: in each acquisition, take one of the m strain sensors as the reference strain sensor, and the strain vector of the reference strain sensor as the reference value. Then the reference values for n acquisitions are S_base1, S_base2, …, S_basen. Calculate the ratio of the strain vector of each strain sensor to the reference value to obtain the first matrix M1:
[0107] M1 = M0 / S_base = [S_1 / S_base1, S_2 / S_base2, ..., S_n / S_basen] = (s_1, s_2, ..., s_n)
[0108] Normalize the first matrix M1 to obtain the second matrix M2, including: obtain the maximum value VMAXj and the minimum value VMINj of each column of the first matrix, where j represents the j-th column; normalize the strain vector of the corresponding column of the first matrix according to the maximum value VMAXj and the minimum value VMINj of each column of the first matrix to obtain the second matrix. For example, the first matrix Then v_max = (3, 6, 9), v_min = (1, 4, 7). At this time,
[0109] Use the level prediction regression model model to fit the level, and the fitting coefficient is vec_coe. The fitting process makes the second norm of model(vec_coe, M2) - Y the smallest to obtain the value of the fitting coefficient, where Y is the true label value of the level information, Y = [y_1, y_2, ..., y_n]. Among them, the second norm of model(vec_coe, M2) - Y being the smallest means making the following loss function the smallest: LOSS ||model(vec_coe, M2) - Y||2.
[0110] After obtaining the above level prediction model and fitting coefficient, perform the prediction of the actual level, specifically including:
[0111] During actual detection, the current strain vectors of the m strain sensors are S1’, S2’, S3’, …, Sm’. Take the ratio of the current strain vector of each strain sensor to the current strain vector Sbase’ of the reference strain sensor as the first strain vector of each strain sensor. The first strain vectors of the m strain sensors are S1’ / Sbase’, S2’ / Sbase’, …, Sm’ / Sbase’.
[0112] Normalize the first strain vector of each strain sensor to obtain the second strain vector of each strain sensor, including: according to the maximum value VMAXj and the minimum value VMINj obtained in the aforementioned process of calculating the fitting coefficient, normalize the corresponding first strain vector to obtain the second strain vector, that is:
[0113] The second strain vector sj’ of the jth strain sensor = (Sj’ / Sbase’ - VMINj) / (VMAXj - VMINj);
[0114] For the second strain vectors of m strain sensors, use the pre-acquired level prediction regression model and fitting coefficients for prediction to obtain the current level of the hopper, that is:
[0115] The current level Y’ = model(vec_coe, [s1’, s2’,.., si’,.., sm’]).
[0116] Figure 2 Shows a possible structure of the electronic device provided by the embodiments of the present application. Refer to Figure 2 , the electronic device includes: a processor 1, a memory 2, and a communication interface 3, and these components are interconnected and communicate with each other through a communication bus 4 and / or other forms of connection mechanisms (not shown).
[0117] Among them, the memory 2 includes one or more (only one is shown in the figure), which may be, but is not limited to, a random access memory (Random Access Memory, abbreviated as RAM), a read-only memory (Read Only Memory, abbreviated as ROM), a programmable read-only memory (Programmable Read-Only Memory, abbreviated as PROM), an erasable programmable read-only memory (Erasable Programmable Read-Only Memory, abbreviated as EPROM), an electrically erasable programmable read-only memory (Electric Erasable Programmable Read-Only Memory, abbreviated as EEPROM), etc. The processor 1 and other possible components can access the memory 2, read and / or write the data therein.
[0118] The processor 1 includes one or more (only one is shown in the figure), which can be an integrated circuit chip with the ability to process signals. The above-mentioned processor 1 can be a general-purpose processor, including a Central Processing Unit (CPU), a Micro Controller Unit (MCU), a Network Processor (NP), or other conventional processors; it can also be a dedicated processor, including a Neural-network Processing Unit (NPU), a Graphics Processing Unit (GPU), a Digital Signal Processor (DSP), an Application Specific Integrated Circuits (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Moreover, when there are multiple processors 1, a part of them can be general-purpose processors, and another part can be dedicated processors.
[0119] The communication interface 3 includes one or more (only one is shown in the figure), which can be used to communicate directly or indirectly with other devices for data interaction. The communication interface 3 can include interfaces for wired and / or wireless communication.
[0120] One or more computer program instructions can be stored in the memory 2, and the processor 1 can read and run these computer program instructions to implement the method provided by the embodiments of the present application.
[0121] It can be understood that Figure 2 The structure shown is only schematic, and the electronic device may also include more or fewer components than those shown Figure 2 in the figure, or have a structure different from that shown Figure 2 in the figure. Figure 2 Each component shown in the figure can be implemented by hardware, software, or a combination thereof. The electronic device may be a physical device, such as a PC, a laptop, a tablet, a mobile phone, a server, an embedded device, etc., or a virtual device, such as a virtual machine, a virtualization container, etc. Moreover, the electronic device is not limited to a single device, and can also be a combination of multiple devices or a cluster composed of a large number of devices.
[0122] A computer program product provided by an embodiment of the present application includes a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of any of the above-mentioned methods are implemented.
[0123] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some communication interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.
[0124] In addition, the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0125] Furthermore, in each embodiment of the present application, the various functional modules can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.
[0126] In this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.
[0127] The above is only the embodiments of the present application and is not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A material level detection method, characterized in that, Including: Obtain the current strain vectors of multiple strain sensors set at different heights of the hopper; For the current strain vector of each of the strain sensors, calculate a first strain vector after removing the influence of density; Normalize the first strain vector of each of the strain sensors to obtain a second strain vector for each of the strain sensors; For the second strain vectors of the multiple strain sensors, use a previously obtained material level prediction regression model and fitting coefficients to make a prediction to obtain the current material level of the hopper.
2. The method according to claim 1, characterized in that, The strain sensors include: fiber Bragg grating strain sensors.
3. The method according to claim 1, characterized in that The step of, for the current strain vector of each of the strain sensors, calculating a first strain vector after removing the influence of density includes: Among the multiple strain sensors, determine a reference strain sensor; Take the ratio of the current strain vector of each of the strain sensors to the current strain vector of the reference strain sensor as the first strain vector of each of the strain sensors.
4. The method according to claim 1, wherein The number of the strain sensors is m; The method for obtaining the material level prediction regression model and the fitting coefficients includes: After synchronously collecting n times for the m strain sensors, strain vectors S_1, S_2,..., S_i,..., S_n are obtained, where S_i = (S1, S2, S3,…, Sm), 1 ≤ i ≤ n, and the material level information corresponding to S_i is Y_i; the sensing data matrix of the m strain sensors is M0 = [S_1, S_2,..., S_n]; For the sensing data matrix M0, calculate a first matrix M1 after removing the influence of density; Normalize the first matrix M1 to obtain a second matrix M2; Use the material level prediction regression model model to fit the material level, and the fitting coefficient is vec_coe. The fitting process minimizes the two-norm of model(vec_coe, M2) - Y to obtain the value of the fitting coefficient.
5. The method according to claim 4, wherein The material level prediction regression model includes: a statistical regression model, a machine learning regression model, or a deep learning regression model.
6. The method according to claim 4, characterized in that, The step of, for the sensing data matrix M0, calculating a first matrix M1 after removing the influence of density includes: In each acquisition, take the strain vector of one of the m strain sensors as a reference value, and calculate the ratio of the strain vectors of each strain sensor to the reference value to obtain the first matrix M1.
7. The method according to claim 4, characterized in that The step of, for the first matrix M1, normalizing it to obtain a second matrix M2 includes: Obtain the maximum value VMAXj and the minimum value VMINj of each column of the first matrix; According to the maximum value VMAXj and the minimum value VMINj of each column of the first matrix, normalize the strain vectors of the corresponding columns of the first matrix to obtain the second matrix.
8. The method according to claim 7, characterized in that, The step of, for the first strain vector of each of the strain sensors, normalizing it to obtain a second strain vector for each of the strain sensors includes: According to the maximum value VMAXj and the minimum value VMINj, normalize the corresponding first strain vector to obtain a second strain vector.
9. An electronic device, characterized in that, Including: A processor and a memory, the memory storing machine-readable instructions executable by the processor, and when the machine-readable instructions are executed by the processor, the method according to any one of claims 1-8 is performed.
10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1-8 are implemented.