Infrared thermal imaging-based hot spot early warning system for lining refractory in suspension kiln
By constructing a thermal conductivity model of the lining of a suspension kiln through infrared thermal imaging technology and data analysis, abnormal hot spots can be identified and graded for early warning. This solves the problems of low efficiency and poor accuracy of traditional manual inspection and realizes intelligent monitoring of refractory materials in the lining of suspension kilns.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANXI FUYUANTONG MINING CO LTD
- Filing Date
- 2026-04-02
- Publication Date
- 2026-05-29
AI Technical Summary
The identification of abnormal hot spots in the refractory lining of traditional suspension kilns relies on manual inspection, which is inefficient and easily affected by human factors, making it impossible to detect potential problems in a timely manner.
An infrared thermal imaging-based early warning system for hot spots in the refractory lining of a suspension kiln is adopted. By acquiring thermal images and operating parameter data through infrared thermal imaging equipment, a thermal conductivity model of the lining is constructed to identify and classify abnormal hot spots.
It enables intelligent identification and efficient early warning of abnormal hot spots in the refractory lining of suspension kilns, improving identification accuracy and efficiency while reducing the impact of human factors.
Smart Images
Figure CN122107799A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of monitoring technology for lining materials in suspension kilns, and particularly to a hot spot early warning system for refractory lining materials in suspension kilns based on infrared thermal imaging. Background Technology
[0002] Suspension kilns are a new type of high-efficiency heat treatment equipment, mainly used for calcining lime powder. The refractory lining is a crucial component of the suspension kiln, responsible for protecting the kiln body from high temperatures and corrosive gases, ensuring thermal efficiency and material quality within the kiln. However, under long-term high-temperature operation, the refractory lining may age, be damaged, or melt, leading to abnormally high localized temperatures and the formation of hot spots.
[0003] Traditionally, the identification and early warning of abnormal hot spots in the refractory lining of suspension kilns mainly rely on manual inspection and regular maintenance. This method is not only inefficient, but also fails to detect potential abnormal hot spots in a timely manner. In addition, manual inspection is easily affected by human factors and has a certain degree of subjectivity. Summary of the Invention
[0004] To address the aforementioned technical problems, this invention provides a hotspot early warning system for refractory lining materials in suspension kilns based on infrared thermal imaging. The technical solution of this invention is as follows: A hotspot early warning system for refractory lining materials in suspension kilns based on infrared thermal imaging includes: The data acquisition module is used to perform non-contact temperature measurement inside the suspension kiln using infrared thermal imaging equipment, and to acquire thermal imaging image sets of the refractory lining material in the suspension kiln during the current acquisition period. The kiln condition synchronization module is used to acquire various working parameter data of the suspension kiln in the current acquisition cycle, and determine the kiln condition of the suspension kiln in the current acquisition cycle based on the various working parameter data of the current acquisition cycle. The temperature field construction module is used to generate the lining thermal conductivity model for the current acquisition period based on the thermal imaging image group and the kiln condition for the current acquisition period, and to determine the internal surface temperature field of the suspension kiln for the current acquisition period based on the lining thermal conductivity model for the current acquisition period. The hotspot identification module is used to identify all abnormal hotspots in the refractory lining of the suspension kiln based on the internal surface temperature field of the current acquisition period. The early warning module is used to generate graded early warning signals for each abnormal hot spot of the refractory lining material based on the internal surface temperature field of the current acquisition period.
[0005] Preferably, the kiln condition synchronization module includes: The standardization unit is used to acquire and standardize various working parameter data of the suspension kiln in the current acquisition cycle to obtain various standard working parameter data of the current acquisition cycle. The motion state determination unit is used to determine the motion state trajectory of the solid mixed particles in the current acquisition cycle of the suspension kiln based on various standard working parameter data of the current acquisition cycle. The combustion state determination unit is used to generate a set of combustion thermal maps for the current acquisition cycle of the suspension kiln based on various standard operating parameter data of the current acquisition cycle. The synchronization unit is used to perform time alignment processing on the motion trajectory of solid mixed particles and the combustion heat map group of the current collection cycle of the suspension kiln to obtain the aligned motion trajectory and the aligned combustion heat map group. The kiln condition of the current collection cycle of the suspension kiln is obtained by combining the aligned motion trajectory and the aligned combustion heat map group.
[0006] Preferably, the motion state determination unit includes: The vector construction subunit is used to construct the working vector for each time node based on multiple standard working parameter data for each time node in the current acquisition cycle. The feature extraction subunit is used to extract features from the working vector at each time point using the intrinsic orthogonal decomposition method, thus obtaining the feature vector at each time point. The delayed construction subunit is used to map the feature vector of each time node into the phase space, calculate the delay time, add the delay time to the feature vector of each time node, and obtain the delay vector of each time node. The state transition statistics subunit is used to calculate the state transition coefficient between two adjacent time nodes based on the delay vector between each two adjacent time nodes. The trajectory construction sub-unit is used to simulate the motion of the solid mixed particles in the current collection period of the suspension kiln through stochastic differential equations based on the state transition coefficient between every two adjacent time nodes, and to generate the motion trajectory of the solid mixed particles in the current collection period of the suspension kiln based on the simulated motion of the solid mixed particles in the current collection period of the suspension kiln.
[0007] Preferably, the combustion state determination unit includes: The combustion matrix construction sub-unit is used to calculate the combustion flow rate and exhaust gas volume at each time node based on the working vector at each time node, and to calculate the covariance coefficient of the combustion flow rate and exhaust gas volume at each time node. The covariance coefficient, combustion flow rate and exhaust gas volume at each time node are used as diagonal elements to construct the combustion matrix at each time node. The clustering subunit is used to calculate the matrix distance between the combustion matrices of every two adjacent time nodes. Based on the matrix distance and the relationship between adjacent time nodes, the combustion matrices of all time nodes are clustered to obtain multiple sets of combustion states. The combustion heat map group construction sub-unit is used to obtain the representative matrix of each combustion state set, construct the combustion heat map of each combustion state set based on the representative matrix of each combustion state set, and combine the combustion heat maps of all combustion state sets to obtain the combustion heat map group of the current collection cycle of the suspension kiln.
[0008] Preferably, the temperature field construction module includes: The internal 3D model building unit is used to build the internal 3D model of the suspension kiln based on the factory parameters of the suspension kiln. The mapping unit is used to synchronously map the thermal imaging image group of the current acquisition cycle and the kiln condition time of the current acquisition cycle into the internal three-dimensional model of the suspension kiln to obtain the internal three-dimensional model of the suspension kiln. The heat source determination unit is used to determine the coordinates of the heat source location at each time point based on the kiln condition of the current acquisition cycle. The lining thermal conductivity model construction unit is used to determine the thermal conductivity domain of the suspension kiln in the current acquisition cycle based on the three-dimensional model of the internal state of the suspension kiln, and to construct the lining thermal conductivity model for the current acquisition cycle based on the coordinates of the heat source point at each time node and the thermal conductivity domain of the current acquisition cycle. The temperature field generation unit is used to determine the temperature of all mesh nodes on the inner lining surface of the internal three-dimensional model based on the lining thermal conductivity model of the current acquisition cycle. The temperature of all mesh nodes on the inner lining surface constitutes the inner surface temperature field of the suspension kiln in the current acquisition cycle.
[0009] Preferably, the mapping unit includes: The temperature mapping subunit is used to construct a projection matrix based on the equipment parameters of the infrared thermal imaging device, and to map each thermal imaging image in the thermal imaging image group of the current acquisition cycle to the internal three-dimensional model of the suspension kiln based on the projection matrix to obtain the initial internal three-dimensional model. The kiln condition mapping sub-unit is used to map the position of the solid mixed particles at each time node in the aligned motion state trajectory to the initial internal three-dimensional model according to the coordinate system relationship, and to map each aligned combustion heat map in the aligned combustion heat map group to the initial internal three-dimensional model, so as to obtain the internal state three-dimensional model of the suspension kiln.
[0010] Preferably, the hotspot identification module includes: The splitting unit is used to divide the inner surface temperature field of the current acquisition cycle into multiple local temperature fields based on the temperature change range threshold and the inner surface temperature field of the current acquisition cycle. A temperature gradient calculation unit is used to calculate the temperature gradient vector of each grid node on the lining surface in each local temperature field, wherein the temperature gradient vector contains the temperature change amplitude. The hotspot determination unit is used to determine the hotspots of each local temperature field based on the temperature gradient vector of each mesh node on the inner lining surface in each local temperature field. An anomaly detection unit is used to obtain the temperature gradient vector corresponding to the hot spot of each local temperature field. If the temperature change amplitude in the temperature gradient vector corresponding to the hot spot of any local temperature field is abnormal, and the area of the local temperature field is greater than the preset abnormal area, then the hot spot of the local temperature field is determined to be an abnormal hot spot, and all abnormal hot spots of the refractory lining in the suspension kiln are obtained.
[0011] Preferably, the hotspot determination unit includes: A matrix construction sub-unit is used to construct the temperature gradient matrix of each mesh node on the inner lining surface based on the temperature gradient vector of each mesh node on the inner lining surface in the target local temperature field, wherein the target local temperature field is any local temperature field. The scaling function construction sub-unit is used to construct the scaling function of each inner lining surface grid node based on the preset Gaussian kernel function, preset scaling parameters, and the temperature gradient matrix of each inner lining surface grid node. The operator computation sub-unit is used to calculate the Laplacian operator of each inner lining surface mesh node based on the scaling function of each inner lining surface mesh node, and selects the inner lining surface mesh node with the largest Laplacian operator in the target local temperature field as the hot spot of the target local temperature field.
[0012] Preferably, the early warning module includes: An initialization unit is used to configure an initial warning level for each abnormal hotspot; The first-level anomaly calculation unit is used to generate the first-level anomaly value for each anomaly hotspot based on the temperature change rate of each anomaly hotspot in the inner surface temperature field during the current acquisition cycle. The secondary anomaly calculation unit is used to generate secondary anomaly values for each anomaly hotspot based on its temperature. The early warning signal generation unit is used to superimpose the initial early warning level of each abnormal hotspot with the first-level and second-level abnormal values to obtain the final abnormal level of each abnormal hotspot, and generate a graded early warning signal for each abnormal hotspot based on the final abnormal level of each abnormal hotspot.
[0013] All of the above-mentioned optional technical solutions can be combined arbitrarily, and the present invention will not provide a detailed description of the structure after each combination.
[0014] By means of the above solution, the beneficial effects of the present invention are as follows: By acquiring a set of thermal imaging images of the refractory lining material in the suspension kiln during the current acquisition period, and obtaining various operating parameter data for the current acquisition period, the kiln condition for the current acquisition period is calculated based on these parameters. Furthermore, the internal surface temperature field of the suspension kiln during the current acquisition period is obtained based on the kiln condition and the thermal imaging images. Abnormal hot spots in the refractory lining material of the suspension kiln are identified through this internal surface temperature field, and graded warnings are issued for these abnormal hot spots. Compared to traditional methods, this invention provides an intelligent method for identifying abnormal hot spots by combining infrared thermal imaging with the operating parameter data of the suspension kiln. This method not only accurately identifies abnormal hot spots in the refractory lining material of the suspension kiln during the current acquisition period, but also provides more accurate and efficient identification.
[0015] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description
[0016] Figure 1 This is a schematic diagram of the hot spot early warning system for refractory lining of a suspension kiln based on infrared thermal imaging provided by the present invention. Detailed Implementation
[0017] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.
[0018] like Figure 1 As shown in the embodiment of the present invention, the hot spot early warning system for refractory lining of a suspension kiln based on infrared thermal imaging includes: The data acquisition module is used to perform non-contact temperature measurement inside the suspension kiln using infrared thermal imaging equipment, and to acquire thermal imaging image sets of the refractory lining material in the suspension kiln during the current acquisition period. The kiln condition synchronization module is used to acquire various working parameter data of the suspension kiln in the current acquisition cycle, and determine the kiln condition of the suspension kiln in the current acquisition cycle based on the various working parameter data of the current acquisition cycle. The temperature field construction module is used to generate the lining thermal conductivity model for the current acquisition period based on the thermal imaging image group and the kiln condition for the current acquisition period, and to determine the internal surface temperature field of the suspension kiln for the current acquisition period based on the lining thermal conductivity model for the current acquisition period. The hotspot identification module is used to identify all abnormal hotspots in the refractory lining of the suspension kiln based on the internal surface temperature field of the current acquisition period. The early warning module is used to generate graded early warning signals for each abnormal hot spot of the refractory lining material based on the internal surface temperature field of the current acquisition period.
[0019] Specifically, in the data acquisition module, the infrared thermal imaging device is an instrument that uses the principle of infrared radiation for temperature measurement and image processing. In this embodiment of the invention, the infrared thermal imaging device is installed on the side of the suspension kiln, and the acquisition frequency is one time node (1 second). The current acquisition cycle refers to the time cycle of one working cycle of the suspension kiln from feeding to discharging. The thermal imaging image group is a collection composed of thermal imaging images at each time node of the current acquisition cycle.
[0020] The kiln condition synchronization module stores various operating parameters, including pressure, exhaust gas emission, combustion flow rate, and the position of solid mixed particles. In this embodiment of the invention, kiln condition data includes the motion trajectory of the solid mixed particles and a combustion thermal map. Pressure, exhaust gas emission, and combustion flow rate are measured by pressure sensors, exhaust gas emission sensors, and combustion flow rate sensors, respectively. The position of the solid mixed particles is acquired by a high-speed industrial camera, and the position is represented by the camera coordinate system. The pressure sensor, exhaust gas emission sensor, combustion flow rate sensor, and high-speed industrial camera all acquire data at the same frequency and synchronously.
[0021] In the temperature field construction module, the lining thermal conductivity model is a function used to describe and analyze the temperature distribution of the refractory lining material during heat conduction in a suspension kiln. The inner surface temperature field refers to the temperature distribution of each node in the lining surface mesh on the surface of the refractory lining material in the suspension kiln.
[0022] In the hotspot identification module, abnormal hotspots refer to locations on the refractory lining material where there are abnormal temperatures.
[0023] In the early warning module, the graded early warning signal in this embodiment of the invention is an early warning signal generated according to the anomaly level of different abnormal hot spots, including normal, low-level, medium-level, and critical warnings. For example, when the anomaly level of a certain abnormal hot spot is level 5, the graded early warning signal is a medium-level warning, requiring personnel to check the location of the abnormal hot spot in the refractory lining material.
[0024] In one specific embodiment, the kiln condition synchronization module includes: The standardization unit is used to acquire and standardize various working parameter data of the suspension kiln in the current acquisition cycle to obtain various standard working parameter data of the current acquisition cycle. The motion state determination unit is used to determine the motion state trajectory of the solid mixed particles in the current acquisition cycle of the suspension kiln based on various standard working parameter data of the current acquisition cycle. The combustion state determination unit is used to generate a set of combustion thermal maps for the current acquisition cycle of the suspension kiln based on various standard operating parameter data of the current acquisition cycle. The synchronization unit is used to perform time alignment processing on the motion trajectory of solid mixed particles and the combustion heat map group of the current collection cycle of the suspension kiln to obtain the aligned motion trajectory and the aligned combustion heat map group. The kiln condition of the current collection cycle of the suspension kiln is obtained by combining the aligned motion trajectory and the aligned combustion heat map group.
[0025] Specifically, in the standardization unit, when standardizing the various working parameter data of the current acquisition cycle, all working parameter data are time-aligned to obtain various initial working parameter data. Subsequently, the various initial working parameter data are processed by extremum standardization to obtain various standard working parameter data of the current acquisition cycle.
[0026] In the motion state determination unit, the motion state trajectory of the solid mixed particles in the current collection cycle of the suspension kiln is, in this embodiment of the invention, a set of discrete points formed by the combination of each time node and the position of the corresponding solid mixed particles.
[0027] In the combustion state determination unit, the combustion thermal map group consists of multiple combustion thermal maps. The combustion thermal maps are used to describe the heat distribution during the combustion process of solid mixed particles inside the suspension kiln.
[0028] In the synchronization unit, when aligning the motion trajectory of the solid mixed particles in the current collection cycle of the suspension kiln with the combustion thermal map group, the time alignment is achieved based on the center time node of each combustion thermal map in the combustion thermal map group. The specific time alignment method will be described in the following embodiments.
[0029] In one specific embodiment, the motion state determination unit includes: The vector construction subunit is used to construct the working vector for each time node based on multiple standard working parameter data for each time node in the current acquisition cycle. The feature extraction subunit is used to extract features from the working vector at each time point using the intrinsic orthogonal decomposition method, thus obtaining the feature vector at each time point. The delayed construction subunit is used to map the feature vector of each time node into the phase space, calculate the delay time, add the delay time to the feature vector of each time node, and obtain the delay vector of each time node. The state transition statistics subunit is used to calculate the state transition coefficient between two adjacent time nodes based on the delay vector between each two adjacent time nodes. The trajectory construction sub-unit is used to simulate the motion of the solid mixed particles in the current collection period of the suspension kiln through stochastic differential equations based on the state transition coefficient between every two adjacent time nodes, and to generate the motion trajectory of the solid mixed particles in the current collection period of the suspension kiln based on the simulated motion of the solid mixed particles in the current collection period of the suspension kiln.
[0030] Specifically, in the vector construction sub-unit, the working vector at a certain time node can be represented as: [pressure, exhaust gas emission, combustion flow rate, position of solid mixed particles].
[0031] In the feature extraction subunit, when extracting features from the working vector at a certain time node using the intrinsic orthogonal decomposition method, the mean vector of the working vectors at all time nodes is first calculated, and the mean vector is subtracted from the working vector at each time node to obtain the center vector of each time node. Then, a center matrix is constructed from the center vectors of all time nodes (one row represents the center vector of a time node, and one column represents an element of the center vector). The product of the center matrix and the transpose of the center matrix is calculated to obtain the correlation matrix. The correlation matrix is subjected to eigenvalue eigenvalue decomposition to obtain intrinsic eigenvectors. An intrinsic feature space is constructed from the intrinsic eigenvectors. The working vector of the time node is projected onto the intrinsic feature space to obtain the feature vector of the time node.
[0032] In the delayed construction subunit, after mapping the feature vector of each time node to the phase space, the delay time parameter is first constructed when calculating the delay time. With autocorrelation coefficient The functional relationship is expressed as: ; in, This represents the feature vector at the k-th time point. This represents the mean eigenvector of all eigenvectors, where n represents the total number of time points. The autocorrelation coefficient is selected. Minimum delay time parameter As a delay time, the feature vectors of all time nodes are shifted backward by the delay time to obtain the delay vector for each time node. For example, if the feature vector of the first time node is 'a' and the delay time is 3 time nodes, then after shifting, 'a' becomes the delay vector of the fourth time node. It should be noted here that after shifting, the delay vectors of the first 3 time nodes are all defined as... That is, the feature vector of the first time node.
[0033] In the state transition statistics subunit, when calculating the state transition coefficient between two adjacent time nodes based on the delay vectors of two adjacent time nodes, the cosine similarity between the delay vectors of the two adjacent time nodes is calculated, and the cosine similarity is used as the state transition coefficient between the two adjacent time nodes.
[0034] In the trajectory construction subunit, when simulating the motion of solid mixed particles in the current collection cycle of the suspension kiln using stochastic differential equations, the state transition coefficients between all adjacent time nodes are first processed using the least squares estimation method to obtain the transition fitting function. All transition coefficients within the transition fitting function are then obtained, forming a one-row, multi-column estimation matrix. Subsequently, stochastic differential equations are generated based on the estimation matrix. Where A represents the estimation matrix, t represents the time parameter, and L(t) represents the simulated motion trajectory function of the solid mixed particles in the current collection cycle of the suspension kiln. Each time node of the current collection cycle is input into the simulated motion trajectory function, and the position of the solid mixed particles at each time node is output by the simulated motion trajectory function. The positions of the solid mixed particles at all time nodes of the current collection cycle are sorted by time to generate a sequence, thus obtaining the motion state trajectory of the solid mixed particles in the current collection cycle of the suspension kiln.
[0035] In this embodiment of the invention, the various standard operating parameter data at each time point in the current acquisition cycle of the suspension kiln are mixed with a large amount of measurement noise and instantaneous disturbances. Moreover, the measurement noise and instantaneous disturbances generally appear in the form of deviation. Therefore, by constructing a delay time in the phase space and generating a delay vector based on the delay time, the delay vector can effectively suppress a large amount of measurement noise and instantaneous disturbances appearing in the form of deviation. The state transition coefficient is accurately obtained based on the delay vector, thereby accurately constructing the motion trajectory of the solid mixed particles in the current acquisition cycle of the suspension kiln.
[0036] In one specific embodiment, the combustion state determination unit includes: The combustion matrix construction sub-unit is used to calculate the combustion flow rate and exhaust gas volume at each time node based on the working vector at each time node, and to calculate the covariance coefficient of the combustion flow rate and exhaust gas volume at each time node. The covariance coefficient, combustion flow rate and exhaust gas volume at each time node are used as diagonal elements to construct the combustion matrix at each time node. The clustering subunit is used to calculate the matrix distance between the combustion matrices of every two adjacent time nodes. Based on the matrix distance and the relationship between adjacent time nodes, the combustion matrices of all time nodes are clustered to obtain multiple sets of combustion states. The combustion heat map group construction sub-unit is used to obtain the representative matrix of each combustion state set, construct the combustion heat map of each combustion state set based on the representative matrix of each combustion state set, and combine the combustion heat maps of all combustion state sets to obtain the combustion heat map group of the current collection cycle of the suspension kiln.
[0037] Specifically, in the combustion matrix construction sub-unit, when calculating the combustion flow rate and exhaust gas emission at a certain time point, the difference between the exhaust gas emission at that time point and the exhaust gas emission at the previous time point is calculated as the exhaust gas emission at that time point. Similarly, the combustion flow rate at that time point can be obtained. The combustion matrix at a certain time point can be represented as follows: .
[0038] In the clustering subunit, when calculating the matrix distance between the combustion matrices of every two adjacent time nodes, this embodiment of the invention uses Euclidean distance. When clustering the combustion matrices of all time nodes based on the relationship between matrix distance and adjacent time nodes, the combustion matrices of all time nodes are first clustered according to the matrix distance to obtain multiple initial clusters. The time nodes in each initial cluster that are not adjacent to other time nodes are taken as discrete time nodes. The combustion matrices of all discrete time nodes are removed from each initial cluster to obtain multiple sets of combustion states.
[0039] In the combustion heatmap group construction sub-unit, the representative matrix of a certain combustion state set refers to the combustion matrix with the smallest modulus among the combustion matrices. When constructing the combustion heatmap of a combustion state set based on its representative matrix, the matrix distances between all other combustion matrices and the representative matrix are first obtained. The maximum and minimum distances between these distances are then divided into three equal categories (Category 1, Category 2, and Category 3, where Category 1 is closest to the minimum distance and Category 3 is closest to the maximum distance). Blue labels are added to other combustion matrices whose distances to the representative matrix belong to Category 1, orange labels to those belonging to Category 2, and red labels to those belonging to Category 3. The color labels for each time point are arranged sequentially (with the representative matrix's color label being white), resulting in the combustion heatmap for that combustion state set. The combustion heatmap includes multiple time points and corresponding color labels for each time point.
[0040] Based on the above, if the delay time is 3 time nodes, when the synchronization unit performs time alignment processing on the motion trajectory of solid mixed particles and the combustion heat map group of the current collection cycle of the suspension kiln, it first aligns the time nodes in the motion trajectory with the time nodes in all the combustion heat maps one by one, and then removes the last 3 time nodes of the motion trajectory and the corresponding positions of the solid mixed particles to obtain a fully time-aligned motion trajectory and an aligned combustion heat map group.
[0041] In one specific embodiment, the temperature field construction module includes: The internal 3D model building unit is used to build the internal 3D model of the suspension kiln based on the factory parameters of the suspension kiln. The mapping unit is used to synchronously map the thermal imaging image group of the current acquisition cycle and the kiln condition time of the current acquisition cycle into the internal three-dimensional model of the suspension kiln to obtain the internal three-dimensional model of the suspension kiln. The heat source determination unit is used to determine the coordinates of the heat source location at each time point based on the kiln condition of the current acquisition cycle. The lining thermal conductivity model construction unit is used to determine the thermal conductivity domain of the suspension kiln in the current acquisition cycle based on the three-dimensional model of the internal state of the suspension kiln, and to construct the lining thermal conductivity model for the current acquisition cycle based on the coordinates of the heat source point at each time node and the thermal conductivity domain of the current acquisition cycle. The temperature field generation unit is used to determine the temperature of all mesh nodes on the inner lining surface of the internal three-dimensional model based on the lining thermal conductivity model of the current acquisition cycle. The temperature of all mesh nodes on the inner lining surface constitutes the inner surface temperature field of the suspension kiln in the current acquisition cycle.
[0042] Specifically, in the internal 3D model construction unit, the length, width and height dimensions of the internal part of the suspension kiln are obtained according to the factory parameters of the suspension kiln. The internal 3D model of the suspension kiln is constructed using modeling software. The internal coordinate system of the suspension kiln is constructed with the center of the internal 3D model as the origin, the horizontal axis as parallel to the horizontal plane, the vertical axis as perpendicular to the horizontal plane, and the vertical axis and the other two axes as a right-handed coordinate system.
[0043] In the heat source determination unit, when determining the heat source location coordinates of a certain time node based on the kiln condition of the current acquisition cycle, all time nodes with red color labels in the aligned combustion heat map where the time node is located are taken as multiple heat source time nodes. The positions of the solid mixed particles corresponding to the working vectors of all heat source time nodes are calibrated into the internal three-dimensional model through the internal coordinate system of the suspension kiln to obtain multiple heat source points corresponding to the aligned combustion heat map. The coordinates of the centroid of all heat source points are obtained as the heat source location coordinates of the time node.
[0044] In the internal lining thermal conductivity model construction unit, when determining the thermal conductivity domain for the current sampling cycle of the suspension kiln based on the internal state 3D model, the portion of the internal state 3D model surrounded by multiple regions with thermal imaging colors of dark red, red, and light red is taken as the influence region, and this influence region is used as the thermal conductivity domain for the current sampling cycle of the suspension kiln. The coordinates of the heat source point location at each time node are input into a pre-trained Gaussian kernel model to obtain the heat source distribution function. Using the thermal conductivity domain as the boundary, the heat conduction equation is solved based on the heat source distribution function. The thermal conductivity model of the lining for the current acquisition cycle is obtained. Where k represents the thermal conductivity of the refractory lining material. Let (x, y, z) represent the heat source distribution function, and (x, y, z) represent the location. This represents the heat flux value at location (x, y, z) in the three-dimensional model of the internal state.
[0045] In the temperature field generation unit, the internal region of the suspension kiln is extracted from the internal 3D model (directly extracting the boundary of the internal 3D model). The internal region is divided into multiple unit meshes, where each unit mesh contains four lining surface mesh nodes. When determining a specific lining surface mesh node... temperature First, the heat flux value of the mesh node on the lining surface is obtained from the lining thermal conductivity model. Then, the mesh node on the lining surface is calculated based on the heat flux value. temperature The calculation formula is: ;in, Indicates ambient temperature. This represents the heat flux value of the grid nodes on the lining surface. This indicates the thickness of the refractory lining. The internal surface temperature field of the suspension kiln during the current sampling period is composed of the temperatures of all grid nodes on the lining surface.
[0046] In one specific embodiment, the mapping unit includes: The temperature mapping subunit is used to construct a projection matrix based on the equipment parameters of the infrared thermal imaging device, and to map each thermal imaging image in the thermal imaging image group of the current acquisition cycle to the internal three-dimensional model of the suspension kiln based on the projection matrix to obtain the initial internal three-dimensional model. The kiln condition mapping sub-unit is used to map the position of the solid mixed particles at each time node in the aligned motion state trajectory to the initial internal three-dimensional model according to the coordinate system relationship, and to map each aligned combustion heat map in the aligned combustion heat map group to the initial internal three-dimensional model, so as to obtain the internal state three-dimensional model of the suspension kiln.
[0047] Specifically, in the temperature mapping subunit, the device parameters of the infrared thermal imaging equipment include focal length, the pose (rotation matrix and translation vector) of the infrared thermal imaging equipment and the suspension kiln, and intrinsic parameter matrices. The extrinsic parameter matrix is constructed based on the pose of the infrared thermal imaging equipment and the suspension kiln as [rotation matrix | translation vector], and the projection matrix is the matrix product of the intrinsic and extrinsic parameter matrices. Multiplying a pixel in a thermal imaging image by the projection matrix yields the pixel's position coordinates in the internal 3D model. Combining the position coordinates of all pixels in the thermal imaging image yields the initial internal 3D model.
[0048] In the kiln condition mapping subunit, when mapping the position of the solid mixed particles at each time node in the aligned motion trajectory to the initial internal 3D model according to the coordinate system relationship, the equipment parameters of the high-speed industrial camera (including focal length, intrinsic parameter matrix, and the pose relationship between the camera coordinate system and the suspension kiln) are obtained. The subsequent reference temperature mapping subunit implements the method of constructing a projection matrix and mapping the thermal imaging image. When mapping each aligned combustion thermal image in the aligned combustion thermal image group to the initial internal 3D model, the color label of each time node in each aligned combustion thermal image is superimposed onto the initial internal 3D model to obtain the internal state 3D model.
[0049] In one specific embodiment, the hotspot identification module includes: The splitting unit is used to divide the inner surface temperature field of the current acquisition cycle into multiple local temperature fields based on the temperature change range threshold and the inner surface temperature field of the current acquisition cycle. A temperature gradient calculation unit is used to calculate the temperature gradient vector of each grid node on the lining surface in each local temperature field, wherein the temperature gradient vector contains the temperature change amplitude. The hotspot determination unit is used to determine the hotspots of each local temperature field based on the temperature gradient vector of each mesh node on the inner lining surface in each local temperature field. An anomaly detection unit is used to obtain the temperature gradient vector corresponding to the hot spot of each local temperature field. If the temperature change amplitude in the temperature gradient vector corresponding to the hot spot of any local temperature field is abnormal, and the area of the local temperature field is greater than the preset abnormal area, then the hot spot of the local temperature field is determined to be an abnormal hot spot, and all abnormal hot spots of the refractory lining in the suspension kiln are obtained.
[0050] Specifically, in the splitting unit, all inner lining surface grid nodes are clustered based on their temperatures to obtain multiple groups of candidate nodes. For any group of candidate nodes, if the minimum Euclidean distance between one candidate node and the other candidate nodes in that group is greater than a preset region distance, then that candidate node is considered a deviation node, and all deviation nodes in that group are removed, resulting in a local temperature field composed of multiple candidate nodes that can form a region. In this embodiment of the invention, the preset region distance is preferably 5m.
[0051] In the temperature gradient calculation unit, when calculating the temperature gradient vector of a certain lining surface mesh node in a local temperature field, the lining surface mesh node closest to that node is selected as the reference point. The gradient of the reference point relative to the horizontal axis of the lining surface mesh node is then calculated. for , This represents the x-axis coordinate of the mesh node on the lining surface. The x-axis coordinates of the reference point are represented. This indicates the temperature of the grid nodes on the lining surface. This indicates the temperature at the control point; the gradient between the control point and the grid node on the lining surface along the longitudinal axis. , This represents the ordinate of the grid node on the lining surface. The vertical coordinate of the control point; the gradient of the control point relative to the vertical axis of the mesh node on the lining surface. , This represents the vertical axis coordinate of the mesh node on the lining surface. The vertical axis coordinates represent the reference point. Based on this, the temperature gradient vector of the mesh nodes on the lining surface is expressed as: . This indicates the temperature variation amplitude of the grid nodes on the lining surface (through calculation). (The model is obtained).
[0052] In the anomaly detection unit, when determining whether the temperature change amplitude in the temperature gradient vector corresponding to a hot spot in a local temperature field is abnormal, it checks whether the temperature change amplitude in the temperature gradient vector corresponding to the hot spot in the local temperature field is greater than a preset amplitude threshold (the preset amplitude threshold is obtained through historical experience in this embodiment of the invention). If the temperature change amplitude in the temperature gradient vector corresponding to the hot spot in the local temperature field is greater than the preset amplitude threshold, then the temperature change amplitude in the temperature gradient vector corresponding to the hot spot in the local temperature field is determined to be abnormal. The preset abnormal area is assessed by experts based on historical data.
[0053] In one specific embodiment, the hotspot determination unit includes: A matrix construction sub-unit is used to construct the temperature gradient matrix of each mesh node on the inner lining surface based on the temperature gradient vector of each mesh node on the inner lining surface in the target local temperature field, wherein the target local temperature field is any local temperature field. The scaling function construction sub-unit is used to construct the scaling function of each inner lining surface grid node based on the preset Gaussian kernel function, preset scaling parameters, and the temperature gradient matrix of each inner lining surface grid node. The operator computation sub-unit is used to calculate the Laplacian operator of each inner lining surface mesh node based on the scaling function of each inner lining surface mesh node, and selects the inner lining surface mesh node with the largest Laplacian operator in the target local temperature field as the hot spot of the target local temperature field.
[0054] Specifically, in the matrix construction sub-unit, the rows of the temperature gradient matrix represent the temperature gradient vectors of the mesh nodes on the inner lining surface, and the columns represent the elements of the temperature gradient vectors (a total of four columns).
[0055] In the sub-element constructed by the scaling function, a certain mesh node B on the inner lining surface ( The scaling function S can be expressed as: ;in This represents the Gaussian kernel function of the mesh nodes on the lining surface. This indicates the preset scale parameters. This represents the temperature change amplitude in the temperature gradient matrix of the grid nodes on the lining surface.
[0056] In the operator computation sub-element, the Laplacian operator is used to compute the mesh nodes of the lining surface. When, the formula is: .
[0057] In one specific embodiment, the early warning module includes: An initialization unit is used to configure an initial warning level for each abnormal hotspot; The first-level anomaly calculation unit is used to generate the first-level anomaly value for each anomaly hotspot based on the temperature change rate of each anomaly hotspot in the inner surface temperature field during the current acquisition cycle. The secondary anomaly calculation unit is used to generate secondary anomaly values for each anomaly hotspot based on its temperature. The early warning signal generation unit is used to superimpose the initial early warning level of each abnormal hotspot with the first-level and second-level abnormal values to obtain the final abnormal level of each abnormal hotspot, and generate a graded early warning signal for each abnormal hotspot based on the final abnormal level of each abnormal hotspot.
[0058] Specifically, in the initialization unit, if an abnormal hotspot has never appeared at a certain location in all previous sampling periods, the initial warning level of the abnormal hotspot at that location in the current sampling period is configured to 0. Conversely, if an abnormal hotspot has never appeared at a certain location in the previous sampling period, the initial warning level of the abnormal hotspot at that location in the current sampling period is configured to the final abnormality level when the abnormal hotspot last appeared at that location.
[0059] In the first-level anomaly calculation unit, when the temperature change rate of an abnormal hot spot is greater than the preset temperature change rate threshold, the first-level anomaly value of the abnormal hot spot is 1; otherwise, the first-level anomaly value of the abnormal hot spot is 0.
[0060] In the secondary anomaly calculation unit, for any anomalous hotspot, if the temperature of the anomalous hotspot is greater than the primary temperature anomaly value but less than the secondary temperature anomaly value, then the secondary anomaly value of the anomalous hotspot is 1; if the temperature of the anomalous hotspot is less than the primary temperature anomaly value, then the secondary anomaly value of the anomalous hotspot is 0; if the temperature of the anomalous hotspot is greater than the secondary temperature anomaly value, then the secondary anomaly value of the anomalous hotspot is 2. The preferred value for the primary temperature anomaly value is 1300℃; the preferred value for the secondary temperature anomaly value is 1500℃.
[0061] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A hotspot early warning system for refractory lining of a suspension kiln based on infrared thermal imaging, characterized in that, include: The data acquisition module is used to perform non-contact temperature measurement inside the suspension kiln using infrared thermal imaging equipment, and to acquire thermal imaging image sets of the refractory lining material in the suspension kiln during the current acquisition period. The kiln condition synchronization module is used to acquire various working parameter data of the suspension kiln in the current acquisition cycle, and determine the kiln condition of the suspension kiln in the current acquisition cycle based on the various working parameter data of the current acquisition cycle. The temperature field construction module is used to generate the lining thermal conductivity model for the current acquisition period based on the thermal imaging image group and the kiln condition for the current acquisition period, and to determine the internal surface temperature field of the suspension kiln for the current acquisition period based on the lining thermal conductivity model for the current acquisition period. The hotspot identification module is used to identify all abnormal hotspots in the refractory lining of the suspension kiln based on the internal surface temperature field of the current acquisition period. The early warning module is used to generate graded early warning signals for each abnormal hot spot of the refractory lining material based on the internal surface temperature field of the current acquisition period.
2. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 1, characterized in that, The kiln condition synchronization module includes: The standardization unit is used to acquire and standardize various working parameter data of the suspension kiln in the current acquisition cycle to obtain various standard working parameter data of the current acquisition cycle. The motion state determination unit is used to determine the motion state trajectory of the solid mixed particles in the current acquisition cycle of the suspension kiln based on various standard working parameter data of the current acquisition cycle. The combustion state determination unit is used to generate a set of combustion thermal maps for the current acquisition cycle of the suspension kiln based on various standard operating parameter data of the current acquisition cycle. The synchronization unit is used to perform time alignment processing on the motion trajectory of solid mixed particles and the combustion heat map group of the current collection cycle of the suspension kiln to obtain the aligned motion trajectory and the aligned combustion heat map group. The kiln condition of the current collection cycle of the suspension kiln is obtained by combining the aligned motion trajectory and the aligned combustion heat map group.
3. The infrared thermal imaging-based early warning system for hotspots in refractory linings of a suspension kiln according to claim 2, characterized in that, The motion state determination unit includes: The vector construction subunit is used to construct the working vector for each time node based on multiple standard working parameter data for each time node in the current acquisition cycle. The feature extraction subunit is used to extract features from the working vector at each time point using the intrinsic orthogonal decomposition method, thus obtaining the feature vector at each time point. The delayed construction subunit is used to map the feature vector of each time node into the phase space, calculate the delay time, add the delay time to the feature vector of each time node, and obtain the delay vector of each time node. The state transition statistics subunit is used to calculate the state transition coefficient between two adjacent time nodes based on the delay vector between each two adjacent time nodes. The trajectory construction sub-unit is used to simulate the motion of the solid mixed particles in the current collection period of the suspension kiln through stochastic differential equations based on the state transition coefficient between every two adjacent time nodes, and to generate the motion trajectory of the solid mixed particles in the current collection period of the suspension kiln based on the simulated motion of the solid mixed particles in the current collection period of the suspension kiln.
4. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 3, characterized in that, The combustion state determination unit includes: The combustion matrix construction sub-unit is used to calculate the combustion flow rate and exhaust gas volume at each time node based on the working vector at each time node, and to calculate the covariance coefficient of the combustion flow rate and exhaust gas volume at each time node. The covariance coefficient, combustion flow rate and exhaust gas volume at each time node are used as diagonal elements to construct the combustion matrix at each time node. The clustering subunit is used to calculate the matrix distance between the combustion matrices of every two adjacent time nodes. Based on the matrix distance and the relationship between adjacent time nodes, the combustion matrices of all time nodes are clustered to obtain multiple sets of combustion states. The combustion heat map group construction sub-unit is used to obtain the representative matrix of each combustion state set, construct the combustion heat map of each combustion state set based on the representative matrix of each combustion state set, and combine the combustion heat maps of all combustion state sets to obtain the combustion heat map group of the current collection cycle of the suspension kiln.
5. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 2, characterized in that, The temperature field construction module includes: The internal 3D model building unit is used to build the internal 3D model of the suspension kiln based on the factory parameters of the suspension kiln. The mapping unit is used to synchronously map the thermal imaging image group of the current acquisition cycle and the kiln condition time of the current acquisition cycle into the internal three-dimensional model of the suspension kiln to obtain the internal three-dimensional model of the suspension kiln. The heat source determination unit is used to determine the coordinates of the heat source location at each time point based on the kiln condition of the current acquisition cycle. The lining thermal conductivity model construction unit is used to determine the thermal conductivity domain of the suspension kiln in the current acquisition cycle based on the three-dimensional model of the internal state of the suspension kiln, and to construct the lining thermal conductivity model for the current acquisition cycle based on the coordinates of the heat source point at each time node and the thermal conductivity domain of the current acquisition cycle. The temperature field generation unit is used to determine the temperature of all mesh nodes on the inner lining surface of the internal three-dimensional model based on the lining thermal conductivity model of the current acquisition cycle. The temperature of all mesh nodes on the inner lining surface constitutes the inner surface temperature field of the suspension kiln in the current acquisition cycle.
6. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 5, characterized in that, The mapping unit includes: The temperature mapping subunit is used to construct a projection matrix based on the equipment parameters of the infrared thermal imaging device, and to map each thermal imaging image in the thermal imaging image group of the current acquisition cycle to the internal three-dimensional model of the suspension kiln based on the projection matrix to obtain the initial internal three-dimensional model. The kiln condition mapping sub-unit is used to map the position of the solid mixed particles at each time node in the aligned motion state trajectory to the initial internal three-dimensional model according to the coordinate system relationship, and to map each aligned combustion heat map in the aligned combustion heat map group to the initial internal three-dimensional model, so as to obtain the internal state three-dimensional model of the suspension kiln.
7. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 5, characterized in that, The hotspot identification module includes: The splitting unit is used to divide the inner surface temperature field of the current acquisition cycle into multiple local temperature fields based on the temperature change range threshold and the inner surface temperature field of the current acquisition cycle. A temperature gradient calculation unit is used to calculate the temperature gradient vector of each grid node on the lining surface in each local temperature field, wherein the temperature gradient vector contains the temperature change amplitude. The hotspot determination unit is used to determine the hotspots of each local temperature field based on the temperature gradient vector of each mesh node on the inner lining surface in each local temperature field. An anomaly detection unit is used to obtain the temperature gradient vector corresponding to the hot spot of each local temperature field. If the temperature change amplitude in the temperature gradient vector corresponding to the hot spot of any local temperature field is abnormal, and the area of the local temperature field is greater than the preset abnormal area, then the hot spot of the local temperature field is determined to be an abnormal hot spot, and all abnormal hot spots of the refractory lining in the suspension kiln are obtained.
8. The infrared thermal imaging-based hotspot early warning system for refractory lining of a suspension kiln according to claim 7, characterized in that, The hotspot determination unit includes: A matrix construction sub-unit is used to construct the temperature gradient matrix of each mesh node on the inner lining surface based on the temperature gradient vector of each mesh node on the inner lining surface in the target local temperature field, wherein the target local temperature field is any local temperature field. The scaling function construction sub-unit is used to construct the scaling function of each inner lining surface grid node based on the preset Gaussian kernel function, preset scaling parameters, and the temperature gradient matrix of each inner lining surface grid node. The operator computation sub-unit is used to calculate the Laplacian operator of each inner lining surface mesh node based on the scaling function of each inner lining surface mesh node, and selects the inner lining surface mesh node with the largest Laplacian operator in the target local temperature field as the hot spot of the target local temperature field.
9. The infrared thermal imaging-based early warning system for hotspots in refractory linings of a suspension kiln according to claim 1, characterized in that, The early warning module includes: An initialization unit is used to configure an initial warning level for each abnormal hotspot; The first-level anomaly calculation unit is used to generate the first-level anomaly value for each anomaly hotspot based on the temperature change rate of each anomaly hotspot in the inner surface temperature field during the current acquisition cycle. The secondary anomaly calculation unit is used to generate secondary anomaly values for each anomaly hotspot based on its temperature. The early warning signal generation unit is used to superimpose the initial early warning level of each abnormal hotspot with the first-level and second-level abnormal values to obtain the final abnormal level of each abnormal hotspot, and generate a graded early warning signal for each abnormal hotspot based on the final abnormal level of each abnormal hotspot.