On-line monitoring system and method for steel slag pressurized hot and stuffy gas components

Abstract

The invention relates to the technical field of gas component monitoring, in particular to a steel slag pressure hot disintegration gas component online monitoring system and method.The time difference of CO and H2 signals received by adjacent sensors is measured based on three-dimensional coordinates of gas sensors of a steel slag tank, concentration amplitude vectors are extracted after waveforms are corrected, and the Euclidean distance ratio is calculated; matching working condition threshold values, screening standard nodes, monitoring a concentration change rate curve, collecting temperature and pressure data, and comparing with a safe concentration threshold value to finish analysis. According to the method, a sensor is dynamically calibrated through coordinate and signal time difference, time sequence misalignment is corrected to guarantee data synchronism, a signal model is constructed to suppress interference to improve the anti-noise capability, credible node extreme value matching calibration error recognition stable platform is screened, and the concentration and temperature and pressure parameter ratio is synchronously analyzed to eliminate environmental interference. And gradient interpolation compensation data and threshold verification are carried out, so that risk early warning and monitoring robustness improvement are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of gas composition monitoring, and in particular to an online monitoring system and method for pressurized, hot and stuffy gas composition in steel slag. Background Art

[0002] The field of gas composition monitoring encompasses a collection of technologies for real-time detection and analysis of gas compositions generated during industrial production. Its core focus is on obtaining gas samples and identifying their chemical composition through specific means, ensuring accurate and reliable monitoring, particularly in high-temperature, high-pressure, or corrosive environments. These technologies involve high-temperature design of gas sampling devices, optimized sensor array layout, and simultaneous acquisition and transmission of multiple parameters. These technologies primarily serve process control and emission management in industries such as metallurgy and chemical engineering, requiring systems that can adapt to complex operating conditions and guarantee long-term stability.

[0003] Among them, an online monitoring system for the composition of pressurized hot-stifling gases in steel slag refers to a technology for online, real-time monitoring of the gas composition within a sealed container during the pressurized hot-stifling treatment of steel slag. A high-temperature sampling probe is directly inserted into the processing vessel, and a multi-component gas sensor array is used to simultaneously detect key components such as carbon monoxide, hydrogen, and oxygen. A pressure-sealed structure ensures that the sampling process is isolated from the external environment. The system uses corrosion-resistant alloy materials to construct the gas flow path, corrects sensor output data through temperature and pressure compensation mechanisms, and integrates a data acquisition and transmission unit to provide real-time feedback of gas concentration information to the control terminal, providing a basis for adjusting process parameters.

[0004] Existing technologies rely on fixed compensation mechanisms and physical structures to passively adapt to the environment, and no dynamic calibration mechanism has been established. When high-temperature deformation causes spatial offset of the sensor array, the signal propagation path error increases. The synchronous acquisition of multiple parameters uses a preset compensation coefficient, which cannot respond to the instantaneous temperature and pressure fluctuations caused by turbulence in the gas flow channel. The data correction lag affects the timeliness of control. There is a lack of a quantitative assessment layer for signal credibility. Signal attenuation caused by particle adhesion or electromagnetic interference can easily be misjudged as real concentration changes, leading to process adjustment errors. The lack of concentration stabilization platform identification logic makes it impossible to distinguish between equipment seal failure and the flat concentration curve in the process stabilization stage, which may delay fault handling. The monitoring system has insufficient data repair capabilities for abnormal nodes in the closed flow field and has low robustness under highly dynamic conditions. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an online monitoring system and method for the pressurized hot and stuffy gas composition of steel slag.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: an online monitoring system for pressurized hot and stuffy gas composition in steel slag, the system comprising:

[0007] The sensor calibration module obtains the three-dimensional coordinate parameters of each sensor, measures the arrival time difference of gas signals from adjacent sensors, and calculates the ratio of the time difference to the sensor spacing to generate the sensor delay calibration coefficient.

[0008] a signal credibility module, which extracts the gas concentration amplitude of the calibration node based on the sensor delay calibration coefficient, calculates the ratio of the amplitude vectors of adjacent cycles, and compares it with the dynamic calibration threshold to generate a gas signal credibility coefficient;

[0009] The concentration stabilization module calls the gas signal credibility coefficient to screen the qualified nodes, extracts the CO and H2 concentration period extreme values, calculates the concentration difference and compares it with the calibration error range to generate a gas concentration stabilization platform;

[0010] The anomaly detection module extracts the CO and H2 concentration change rates in the first cycle after the gas concentration stabilization platform is terminated, simultaneously obtains the temperature and pressure change rate ratio in the current cycle, compares the temperature and pressure ratio with the set ratio range, and generates a gas concentration mutation event;

[0011] The real-time monitoring module combines the gas concentration mutation event with the gas signal credibility coefficient, performs gradient interpolation compensation on the abnormal node concentration data, determines whether the compensation value exceeds the safe concentration threshold, and generates real-time monitoring results of the gas composition.

[0012] As a further solution of the present invention, the sensor delay calibration coefficient includes three-dimensional coordinate calibration parameters and delay scaling factors; the gas signal credibility coefficient specifically includes credibility threshold range, dynamic amplitude deviation ratio, and node credibility weight; the gas concentration stabilization platform includes a stable concentration baseline value, a qualified node screening identifier, and a platform duration cycle number; the gas concentration mutation event specifically includes a mutation type classification identifier, a temperature-pressure correlation anomaly coefficient, and a trigger judgment threshold; the real-time monitoring results of the gas composition include an interpolated compensation concentration value, an exceeding-standard node positioning index, and a real-time safety assessment level.

[0013] As a further solution of the present invention, the sensor calibration module includes:

[0014] The spatial parameter acquisition submodule obtains the three-dimensional coordinate parameters of each sensor, calculates the Euclidean distance between adjacent sensors, and generates a three-dimensional coordinate distance matrix containing the distances between all nodes;

[0015] The delay scale factor calculation submodule calls the three-dimensional coordinate spacing matrix, synchronously collects the arrival time difference of the gas signals of adjacent sensors, divides the time difference by the corresponding spacing value, and obtains the delay scale factor set of each node pair;

[0016] The abnormal node calibration submodule calls the delay scale factor set, compares each scale factor with the theoretical range of sound wave propagation delay one by one, filters out node pairs that exceed the upper or lower limit of the theoretical range, marks their status as abnormal, and generates a sensor delay calibration coefficient based on the scale factor deviation value of the abnormal node.

[0017] As a further solution of the present invention, the signal credibility module includes:

[0018] The amplitude vector construction submodule calls the sensor delay calibration coefficient, extracts the gas concentration amplitude data of the calibration node, integrates the three-dimensional spatial distribution characteristics of each node according to the time period, and generates a three-dimensional amplitude vector set containing a multi-period amplitude sequence;

[0019] A vector similarity analysis submodule calculates the ratio of the dot product of adjacent periodic vectors to the product of the modulus lengths of the two vectors based on the three-dimensional amplitude vector set, compares the ratio of each node with the dynamic calibration threshold on a period-by-period basis, and selects nodes whose ratios are lower than the threshold for three consecutive periods to generate a low-similarity node set;

[0020] The offset tolerance determination submodule calls the low-similarity node set, extracts the signal derivative trajectories of its adjacent nodes, calculates the absolute value of the phase offset angle, compares the offset angle with the correction tolerance threshold, and marks the node as an abnormal node if it exceeds the threshold. The gas signal credibility coefficient is generated based on the ratio of the number of abnormal nodes to the total number of nodes.

[0021] As a further solution of the present invention, the absolute value of the phase offset angle is calculated using the formula:

[0022]

[0023] Where ΔA j Represents the difference in derivative trajectories of adjacent nodes in the jth dimension (x / y / z), ΔB j Represents the difference in derivative trajectories of low similarity nodes in the jth dimension, is the Euclidean distance l from the jth adjacent node to the low similarity node j is the inverse weight of , n is the total number of adjacent nodes, and θ is the absolute value of the weighted phase offset angle.

[0024] As a further solution of the present invention, the concentration stabilization module includes:

[0025] The node screening submodule calls the gas signal credibility coefficient, compares each node coefficient with a set credibility threshold one by one, screens sensor nodes whose coefficients exceed the threshold, and generates a qualified node set containing valid monitoring points;

[0026] The extreme value difference calculation submodule extracts the maximum and minimum CO concentrations and the maximum and minimum H2 concentrations of each node within the period based on the qualified node set, calculates the absolute values ​​of the extreme value differences of the two types of gas concentrations, and generates an extreme value difference sequence for all nodes;

[0027] The stable window determination submodule calls the extreme value difference sequence, compares the difference of each node with the calibration error range of its sensor window by window, and counts the number of nodes whose difference values ​​of three consecutive windows are within the error range. If the proportion of the number of nodes exceeds the node number threshold, the current window end timestamp is recorded to generate a gas concentration stability platform.

[0028] As a further solution of the present invention, the anomaly detection module includes:

[0029] The change rate extraction submodule calls the end time point of the gas concentration stabilization platform, extracts the CO concentration change rate and the H2 concentration change rate in the first period thereafter, and generates a concentration change rate set containing the change rates of the two types of gases;

[0030] The parameter ratio calculation submodule synchronously obtains the temperature change rate and pressure change rate in the current cycle, divides the absolute value of the temperature change rate by the absolute value of the pressure change rate, and obtains the temperature-pressure ratio sequence at each time point;

[0031] The mutation event determination submodule calls the concentration change rate set and temperature-pressure ratio sequence, compares the CO change rate with the set change rate threshold, and the H2 change rate with the set change rate threshold, and determines whether the temperature-pressure ratio exceeds the preset ratio range. If the change rates of both types of gases exceed the threshold and the temperature-pressure ratio exceeds the range, a gas concentration mutation event is generated.

[0032] As a further solution of the present invention, the real-time monitoring module includes:

[0033] The abnormal node screening submodule calls the gas concentration mutation event and the gas signal credibility coefficient, screens the nodes whose credibility coefficient is lower than the credibility threshold and has mutation events, and generates an abnormal node set containing spatial position and concentration data;

[0034] The gradient interpolation compensation submodule extracts the CO and H2 concentration values ​​of its adjacent nodes in three-dimensional space based on the abnormal node set, calculates gradient interpolation using the inverse distance weighted method, overwrites the original data of the abnormal node with the compensation value, and generates a compensated concentration set containing the corrected concentration value;

[0035] The safety threshold determination submodule calls the compensation concentration set and compares the CO concentration and H2 concentration of each node with the corresponding concentration thresholds one by one. If any concentration value exceeds the corresponding threshold, it is marked as a risk node. The time period when the proportion of risk nodes exceeds the proportion threshold is counted to generate real-time monitoring results of gas composition.

[0036] As a further solution of the present invention, for calculating the gradient interpolation by the inverse distance weighted method, the formula is adopted:

[0037]

[0038] Among them, C k Measure the concentration value of the kth neighboring node, d k is the three-dimensional space distance from the kth adjacent node to the abnormal node, is the inverse squared distance weight, is the time series concentration difference, Δt is the length of the current time window, Δt ref It is the preset reference time window.

[0039] A method for online monitoring of pressurized, hot, and stuffy gas components in steel slag is provided. The method is based on the above-mentioned online monitoring system for pressurized, hot, and stuffy gas components in steel slag and comprises the following steps:

[0040] S1: Based on the gas sensor array layout inside the slag tank, the three-dimensional coordinate parameters of each sensor are detected, the time difference between adjacent sensors receiving CO and H2 gas signals is measured, and the time difference is linearly proportional to the sensor spacing to generate the sensor delay calibration coefficient including distance compensation;

[0041] S2: Calling the sensor delay calibration coefficient to correct the gas concentration signal waveform, extracting the CO and H2 concentration amplitude vectors in adjacent detection cycles, calculating the Euclidean distance ratio between the amplitude vectors, and comparing them with the steel slag hot stuffy working condition threshold preset in the dynamic calibration threshold library to generate the gas signal credibility coefficient;

[0042] S3: Based on the gas signal credibility coefficient, the gas concentration in the slag tank is screened to meet the standard node, the maximum and minimum values ​​of CO and H2 concentrations in continuous periods are extracted, the absolute difference between the extreme values ​​is calculated, and the difference is matched with the allowable fluctuation range of the slag reaction gas concentration within the calibration error range to generate a gas concentration stable platform;

[0043] S4: monitoring the CO and H2 concentration change rate curves within the first detection period after the gas concentration stabilization platform is terminated, synchronously collecting temperature and pressure sensor data in the slag tank, calculating the ratio of the temperature change rate to the pressure change rate, performing a logical judgment on the ratio and the ratio range set for the slag hot and stuffy safety working condition, and generating a gas concentration mutation event;

[0044] S5: Combined with the trigger state of the gas concentration mutation event and the gas signal credibility coefficient, the adjacent node gradient interpolation method is used to compensate for the missing CO and H2 concentration data of the abnormal node, and the compensation value is compared item by item with the steel slag pressurized hot and stuffy safety concentration threshold table to generate real-time monitoring results of gas composition.

[0045] Compared with the prior art, the advantages and positive effects of the present invention are:

[0046] In the present invention, the sensor is dynamically calibrated through coordinate and signal time difference to correct the timing inaccuracy caused by high-temperature deformation and ensure the spatiotemporal synchronization of multi-component data. A signal screening model is constructed to quantify amplitude fluctuations and phase offsets, suppress signal distortion caused by electromagnetic interference or particle adhesion, and improve noise resistance. The extreme value differences of trusted nodes are screened and the calibration errors are matched to accurately identify the concentration stability platform and distinguish between process steady state and equipment anomalies. The concentration change rate and the temperature and pressure parameter ratio are synchronously analyzed to eliminate interference from environmental mutations and reduce the risk of false alarms. Gradient interpolation compensates for abnormal data and verifies safety thresholds to achieve real-time repair and risk warning, enhancing the robustness of monitoring in closed flow fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a system flow chart of the present invention;

[0048] Figure 2 This is a flow chart for obtaining the sensor calibration module of the present invention;

[0049] Figure 3 This is a flow chart of obtaining the signal credibility module of the present invention;

[0050] Figure 4 This is a flow chart for obtaining the concentration stabilization module of the present invention;

[0051] Figure 5 This is a flowchart of obtaining the abnormality detection module of the present invention;

[0052] Figure 6 This is an acquisition flow chart of the real-time monitoring module of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0054] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0055] In the embodiments of the present invention, the terms "image" and "picture" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same. The terms "of," "corresponding," and "corresponding" may be used interchangeably. It should be noted that, when the distinction between them is not emphasized, their intended meanings are the same.

[0056] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0057] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0058] See also Figure 1 The present invention provides a technical solution: an online monitoring system for pressurized hot and stuffy gas components in steel slag, the system comprising:

[0059] The sensor calibration module obtains the three-dimensional coordinate parameters of each sensor, measures the arrival time difference of gas signals from adjacent sensors, proportionally calculates the time difference with the sensor spacing to obtain a delay scale factor, compares the scale factor with the acoustic wave propagation delay range, marks nodes outside the range as calibration abnormalities, and generates a sensor delay calibration coefficient.

[0060] The signal credibility module extracts the gas concentration amplitude of the calibration node based on the sensor delay calibration coefficient, constructs a multi-cycle three-dimensional amplitude vector, calculates the ratio of the dot product of adjacent cycle vectors to the product of the module length, compares the ratio with the dynamic calibration threshold cycle by cycle, selects nodes whose consecutive cycle ratios are lower than the dynamic calibration threshold, determines whether the phase offset angle of the adjacent node signal derivative trajectory exceeds the correction tolerance, and generates the gas signal credibility coefficient;

[0061] The concentration stabilization module calls the gas signal credibility coefficient to filter sensor nodes whose coefficients exceed the set credibility threshold, extracts the periodic extreme values ​​of CO and H2 concentrations at the filtered nodes, calculates the concentration difference, and compares it with the sensor calibration error range window by window. If the difference between multiple consecutive windows is within the error range, the end time of the current window is recorded to generate a gas concentration stabilization platform.

[0062] The anomaly detection module extracts the CO and H2 concentration change rates in the first cycle after the gas concentration stabilization platform ends, simultaneously obtains the temperature and pressure change rates in the current cycle and calculates the ratio. It determines whether the CO and H2 change rates exceed the set change rate threshold and the temperature-pressure ratio exceeds the ratio range, generating a gas concentration mutation event.

[0063] The real-time monitoring module combines gas concentration mutation events with gas signal credibility coefficients to perform adjacent node gradient interpolation compensation on abnormal node CO and H2 concentration data, determines whether each concentration value exceeds the safety concentration threshold after compensation, and generates real-time monitoring results of gas composition.

[0064] The sensor delay calibration coefficient includes the three-dimensional coordinate calibration parameters and the delay scaling factor. The gas signal credibility coefficient includes the credibility threshold range, the dynamic amplitude deviation ratio, and the node credibility weight. The gas concentration stability platform includes the stable concentration baseline value, the qualified node screening identifier, and the number of platform duration cycles. The gas concentration mutation event includes the mutation type classification identifier, the temperature and pressure correlation anomaly coefficient, and the trigger judgment threshold. The real-time monitoring results of the gas composition include the interpolation compensation concentration value, the exceeding node positioning index, and the real-time safety assessment level.

[0065] See also Figure 2 , the sensor calibration module includes:

[0066] The spatial parameter acquisition submodule obtains the three-dimensional coordinate parameters of each sensor, calculates the Euclidean distance between adjacent sensors, and generates a three-dimensional coordinate distance matrix containing the distances between all nodes;

[0067] Take the four sensors installed in a hot and stuffy tank in a steel plant as an example. The three-dimensional coordinates of the sensors are S1 (2.3, 5.1, 0.8), S2 (5.7, 3.9, 1.2), S3 (8.4, 6.2, 0.5), and S4 (1.8, 9.6, 1.0). The x, y, and z axis coordinate components of each node are extracted, and the Euclidean distance between adjacent node pairs is calculated. For example, the x difference between S1 and S2 is 5.7-2.3 = 3.4 meters, the y difference is 3.9-5.1 = -1.2 meters, and the z difference is 1.2-0.8 = 0.4 meters. Substituting them into the formula Meters, calculate the distance between S1 and S3 in the same way: x difference 8.4-2.3=6.1 meters, y difference 6.2-5.1=1.1 meters, z difference 0.5-0.8=-0.3 meters, and we get All non-adjacent node pairs (such as S2-S4) need to be excluded from the spacing calculation, and finally a three-dimensional coordinate spacing matrix containing valid node spacings such as S1-S2 (3.63), S1-S3 (6.21), S2-S3 (3.15), and S3-S4 (8.03) is generated. The matrix dimension is 4×4, the diagonal elements are 0, and the non-adjacent node spacing is marked as an invalid value NaN.

[0068] The delay scale factor calculation submodule calls the three-dimensional coordinate spacing matrix, synchronously collects the arrival time difference of the gas signals of adjacent sensors, divides the time difference by the corresponding spacing value, and obtains the delay scale factor set of each node pair;

[0069] Assume that the S1 node sends an acoustic signal at timestamp t = 10:00:00.000, and the S2 node receives the signal at t = 10:00:00.012. The time difference Δt = 0.012 seconds. The S1-S2 distance in the three-dimensional coordinate distance matrix is ​​3.63 meters. Calculate the delay scaling factor If the distance between another pair of nodes S3-S4 is 8.03 meters and the measured time difference Δt = 0.025 seconds, then The system traverses all valid node pairs and removes invalid NaN values. When it detects that the time difference between a node pair (such as S2-S3) is abnormal (such as Δt = 0.005 seconds, corresponding to a distance of 3.15 meters, This value will be temporarily stored in the intermediate data set, and the delay scaling factor set will eventually be generated. The data format is {(S1, S2): 0.003306, (S1, S3): 0.002254, (S2, S3): 0.001587, (S3, S4): 0.003113}, where the k value of the S2-S3 node pair deviates significantly from the normal range.

[0070] The abnormal node calibration submodule calls the delay scale factor set and compares each scale factor with the theoretical range of acoustic wave propagation delay. It selects node pairs that exceed the upper or lower limit of the theoretical range and marks their status as abnormal. It then generates the sensor delay calibration coefficient based on the scale factor deviation value of the abnormal node.

[0071] Assuming the site temperature T = 25°C, relative humidity RH = 60%, atmospheric pressure P = 101.325kPa, the sound velocity calculation formula is c = 331.4 + 0.6T + 0.0124RH ≈ 331.4 + 15 + 0.744 = 347.144 m / s, corresponding to the theoretical delay scale factor range (Considering ±1% measurement tolerance), compare each k value in the set with the theoretical range. For example, k of S2-S3 = 0.001587 s / m, which is lower than the lower limit of 0.002881×0.99≈0.002852 s / m. It is determined to be an abnormal node pair, and its deviation value Δk = |0.001587-0.002881| = 0.001294 is calculated. If the system sets the calibration coefficient formula as Assume that only S2-S3 is abnormal and its k i,理论 =0.002881, then This coefficient will be used to correct the delay data of the S2-S3 node pair. For example, the original delay of 0.005 seconds is corrected to 0.005×1.449≈0.007245 seconds, so that the corresponding k value is corrected to Enter the theoretical realm.

[0072] See also Figure 3, the signal credibility module includes:

[0073] The amplitude vector construction submodule calls the sensor delay calibration coefficient, extracts the gas concentration amplitude data of the calibration node, integrates the three-dimensional spatial distribution characteristics of each node according to the time period, and generates a three-dimensional amplitude vector set containing a multi-period amplitude sequence;

[0074] Taking the six calibration nodes in a hot and stuffy tank in a steel plant as an example, the nodes are numbered N1-N6 and distributed on the top and side walls of the tank. The sensor delay calibration coefficient β = 1.12 is used to perform delay correction on the N2 and N4 nodes. The CO concentration amplitude of the nodes after correction in the t1-t3 period is extracted. For example, the amplitude of N2 in the t1 period is [120ppm, 115ppm, 118ppm] (x, y, z directions), and the amplitude of N4 in the t1 period is [98ppm, 105ppm, 102ppm]. The amplitudes of the three spatial dimensions of each node are arranged in periodic order to construct the three-dimensional amplitude vector of N2 as V N2,t1 =[120,115,118], V N2,t2 =[125,112,123], V N2,t3 =[118,120,116], the vector of N4 is V N4,t1 =[98,105,102], V N4,t2 =[95,108,100], V N4,t3 =[101,103,99], integrating all node data to generate a three-dimensional amplitude vector set containing 6 nodes × 3 cycles × 3 dimensions, with a storage structure of {node ID: [cycle 1 vector, cycle 2 vector, cycle 3 vector]}.

[0075] The vector similarity analysis submodule calculates the ratio of the dot product of adjacent periodic vectors to the product of the modulus lengths of the two vectors based on a set of three-dimensional amplitude vectors. It compares the ratio of each node with the dynamic calibration threshold on a period-by-period basis, and selects nodes whose ratios are below the threshold for three consecutive periods to generate a set of low-similarity nodes.

[0076] Select the adjacent periodic vector V of node N2 N2,t1 With V N2,t2 , calculate the dot product value = 120×125+115×112+118×123=15000+12880+14514=42394,

[0077]

[0078] Similarly, the dot product of the t2-t3 period vector of N2 is calculated as 125×118+112×120+123×116=14750+13440+14268=42458.

[0079]

[0080] If the dynamic calibration threshold is set to 0.85, the ratio must be lower than the threshold for three consecutive cycles to be marked as a low-similarity node. The ratio sequence of N2 in the t1-t3 period is [0.9993, 0.9981, 0.9975], which is higher than the threshold of 0.85 and is not included in the set. The ratio sequence of node N5 is [0.82, 0.79, 0.76], which is lower than the threshold for three consecutive cycles. Therefore, the generated low-similarity node set includes N5.

[0081] The offset tolerance judgment submodule calls the low-similarity node set, extracts the signal derivative trajectories of its adjacent nodes, calculates the absolute value of the phase offset angle, compares the offset angle with the correction tolerance threshold, and marks any node that exceeds the threshold as an abnormal node. The gas signal credibility coefficient is generated based on the ratio of the number of abnormal nodes to the total number of nodes.

[0082] Assume that the CO concentration of the adjacent node N4 is 95ppm at t1 and 98ppm at t2, and the derivative value ΔA1 = 98-95 = 3ppm. Similarly, calculate ΔA2 (y direction), ΔA3 (z direction), and ΔB j is the derivative value of the low similarity node N5 in the same dimension, for example, the derivative value of N5 is ΔB1=5ppm, ΔB2=-2ppm, ΔB3=4ppm, ω j is the inverse weight of the Euclidean distance from the adjacent node to the low similarity node, calculated by the three-dimensional coordinate difference. For example, the coordinate difference from N4 to N5 is (3.2-3.0, 5.6-4.8, 1.0-1.2) = (0.2, 0.8, -0.2), and the Euclidean distance Weight ω4 = 1 / 0.85 ≈ 1.176, total weight is the sum of the weights of all adjacent nodes. For example, the weights of adjacent nodes N1, N3, and N4 are 1.754, 1.389, and 1.124 respectively. The normalized weight is For example, the normalized weight of N1 = 1.754 / 4.267≈0.411.

[0083] Example substitution:

[0084] Taking the adjacent node N4 and the low similarity node N5 as an example, ΔA1 = 3ppm, ΔB1 ​​= 5ppm, the absolute difference |3-5| = 2, the normalized weight 0.411, the weighted term = 2×0.411≈0.822, ΔA2 = -2ppm (y-direction derivative difference), ΔB2 = -2ppm, the absolute difference |(-2)-(-2)| = 0, the weighted term = 0×0.311=0, ΔA3 = 1ppm (z-direction derivative difference), ΔB3 = 2ppm, the absolute difference |1-2| = 1, the normalized weight 0.278, the weighted term = 1×0.278≈0.278, the sum of the numerators = 0.822+0+0.278≈1.100, and the denominator calculation is:

[0085] Denominator = 3.741 × 5.744 ≈ 21.50, θ = arccos(1.100 / 21.50) = arccos(0.0511) ≈ 87.1°, the correction tolerance threshold is set to 15°, 87.1°>15°, and N5 is determined to be an abnormal node.

[0086] Parameter rationality: The sensor spacing in a steel mill environment is typically 0.5-2 meters. The derivative value change rate refers to the actual monitoring data fluctuation range of ±10ppm / cycle. The weight coefficient is calculated based on the actual distance to avoid subjective assumptions. The phase offset angle threshold of 15° is set based on the sound wave propagation delay and spatial distribution characteristics, which meets the error tolerance standards of industrial scenarios.

[0087] The result θ = 87.1° indicates that the difference in signal trajectories between adjacent nodes and low-similarity nodes significantly exceeds the tolerance range. The gas signal credibility coefficient is generated by combining the proportion of abnormal nodes. For example, if the number of abnormal nodes among 6 nodes is 1, the credibility coefficient = 1-1 / 6≈0.833.

[0088] See also Figure 4 , the concentration stabilization module includes:

[0089] The node screening submodule calls the gas signal credibility coefficient, compares each node coefficient with the set credibility threshold one by one, screens the sensor nodes whose coefficient exceeds the threshold, and generates a qualified node set containing valid monitoring points;

[0090] Taking the eight sensor nodes deployed in a hot and stuffy tank in a steel plant as an example, the credibility threshold is set to α = 0.8, and the gas signal credibility coefficient set {N1: 0.85, N2: 0.75, N3: 0.92, N4: 0.79, N5: 0.83, N6: 0.68, N7: 0.88, N8: 0.81} is called. The coefficients of each node are compared with the threshold. For example, if the N1 coefficient is 0.85> 0.8, it is marked as qualified, and if the N2 coefficient is 0.75< 0.8, it is marked as qualified. .8, marked as unqualified, and after traversal, five nodes N1, N3, N5, N7, and N8 are screened out, and the qualified node set storage structure is generated as {N1, N3, N5, N7, N8}. At the same time, the original data of CO and H2 concentrations of each node during the screening period are recorded. For example, the CO concentration sequence of N1 is [220ppm, 218ppm, 225ppm], and the H2 concentration sequence is [1.8%LEL, 2.1%LEL, 2.0%LEL].

[0091] The extreme value difference calculation submodule extracts the maximum and minimum CO concentrations and the maximum and minimum H2 concentrations of each node within the period based on the qualified node set, calculates the absolute value of the extreme value difference of the two types of gas concentrations, and generates the extreme value difference sequence of all nodes;

[0092] For the N3 node in the qualified node set, the maximum and minimum CO concentrations in its three monitoring cycles are extracted. For example, the CO concentrations in cycles 1-3 are [215ppm, 230ppm, 210ppm], with a maximum of 230ppm and a minimum of 210ppm. The extreme value difference is ||230-210||=20ppm. The H2 concentration sequence is [1.5%LEL, 2.4%LEL, 1.9%LEL], with a maximum of 230ppm and a minimum of 210ppm. The CO extreme value difference of other nodes, such as N5, is ||228-215||=13ppm, and the H2 extreme value difference is ||2.2-1.7||=0.5%LEL. After traversing all qualified nodes, an extreme value difference sequence is generated. The storage format is {node ID: [CO extreme value difference (ppm), H2 extreme value difference (%LEL)]}, for example:

[0093] {N1:[7,0.3],N3:[20,0.9],N5:[13,0.5],N7:[8,0.4],N8:[10,0.6]}.

[0094] The stable window determination submodule calls the extreme value difference sequence and compares the difference of each node with the calibration error range of its sensor window by window. It counts the number of nodes whose difference of three consecutive windows is within the error range. If the proportion of the number of nodes exceeds the node number threshold, the current window end timestamp is recorded to generate a gas concentration stability platform.

[0095] Set the sensor calibration error range to CO ± 15ppm, H2 ± 0.7% LEL (based on the calibration parameters in the high temperature environment of the steel plant), call the CO difference of the N3 node in the extreme value difference sequence 20ppm, and compare it with the error range of 15ppm. 20>15, which is judged to be out of range. The H2 difference 0.9% LEL> 0.7% LEL is also out of range. The CO difference of N5 is 13ppm≤15ppm, and the H2 difference is 0.5% LEL≤0.7% LEL, which is judged to be valid. Count the number of valid nodes in three consecutive time windows. For example, the number of valid nodes in window 1 is 3 (N1, N5, N7), window 2 is 4 (N1, N5, N7, N8), and window 3 is 4 (N1, N5, N7, N8). The total number of nodes is 5. If the node number threshold is set to 70% (i.e. ), window 3 satisfies 4≥3.5, the recording window end timestamp is 2025-04-1814:35:00, and a gas concentration stable platform is generated. The storage parameters include the timestamp, the valid node list and the corresponding stable concentration range (such as CO: 215-228ppm, H2: 1.7-2.2%LEL).

[0096] See also Figure 5 , the anomaly detection module includes:

[0097] The change rate extraction submodule calls the end time point of the gas concentration stabilization platform, extracts the CO concentration change rate and the H2 concentration change rate in the first cycle thereafter, and generates a concentration change rate set containing the change rates of the two types of gases;

[0098] The end timestamp of the gas concentration stabilization platform is called 2025-04-18 14:35:00, and the CO and H2 concentration data of each node in the first monitoring period (14:35-14:50) are extracted. For example, the CO concentration of node N1 increases from 220ppm to 235ppm, and the change rate is The H2 concentration increases from 1.8%LEL to 2.1%LEL, the change rate CO change rate of node N3 H2 change rate After traversing all nodes, a concentration change rate set is generated, and the storage format is {node ID: [CO change rate (%), H2 change rate (%)]}, for example, {N1: [6.8, 16.7], N3: [7.0, 60.0], N5: [5.5, 18.2], N7: [3.2, 9.1], N8: [4.8, 12.5]}.

[0099] The parameter ratio calculation submodule synchronously obtains the temperature change rate and pressure change rate in the current cycle, divides the absolute value of the temperature change rate by the absolute value of the pressure change rate, and obtains the temperature-pressure ratio sequence at each time point;

[0100] Synchronously obtain the temperature and pressure data within the current cycle. For example, the temperature is 850°C at 14:35 and rises to 870°C at 14:50. The temperature change rate is The pressure increases from 1.2MPa to 1.25MPa, and the pressure change rate Calculate the temperature-pressure ratio If the temperature change rate is 1.8% and the pressure change rate is 3.3% at another time point 14:40, then A sequence of temperature-pressure ratio values ​​is generated by traversing all time points within the period, for example, {14:35:0.563,14:40:0.545,14:45:0.588,14:50:0.571}.

[0101] The mutation event determination submodule calls the concentration change rate set and temperature-pressure ratio sequence, compares the CO change rate with the set change rate threshold, and the H2 change rate with the set change rate threshold, and determines whether the temperature-pressure ratio exceeds the preset ratio range. If the change rates of both gases exceed the threshold and the temperature-pressure ratio exceeds the range, a gas concentration mutation event is generated;

[0102] Set the CO change rate threshold to 5%, the H2 change rate threshold to 15%, and the preset temperature-pressure ratio range to 0.5-0.6. Call the N3 node data (CO7.0%, H260.0%) in the concentration change rate set, compare CO7.0%>5%, H260.0%>15%, and check that the temperature-pressure ratio of 0.563 at the corresponding time point 14:35 is within the range of 0.5-0.6. The double conditions are not met. The CO change rate of node N5 is 5.5%>5%, The H2 change rate of 18.2% is greater than 15%, and the temperature-pressure ratio of 0.545 is within the range, but the condition is still not met. If the CO change rate of a certain node is 6.5%, the H2 change rate is 20.0%, and the temperature-pressure ratio is 0.62>0.6, then the condition is determined to be met, and a gas concentration mutation event is generated. The event parameters are recorded as {timestamp: 2025-04-18 14:50, trigger node: NX, CO change rate: 6.5%, H2 change rate: 20.0%, temperature-pressure ratio: 0.62}.

[0103] See also Figure 6 , the real-time monitoring module includes:

[0104] The abnormal node screening submodule calls the gas concentration mutation event and the gas signal credibility coefficient, screens the nodes whose credibility coefficient is lower than the credibility threshold and has mutation events, and generates an abnormal node set containing spatial location and concentration data;

[0105] Taking the six nodes deployed in a hot and stuffy tank of a steel plant as an example, the credibility threshold is set to α = 0.8, and the gas concentration mutation event set {timestamp 2025-04-18 14:50, triggering nodes N2, N5} and gas signal credibility coefficients {N1: 0.85, N2: 0.75, N3: 0.92, N4: 0.79, N5: 0.76, N6: 0.83} are called. Double screening is performed on each node: First, check whether the credibility coefficient is lower than 0.8, for example, N2 has a coefficient of 0.75. <0.8, N5 coefficient 0.76<0.8, then verify whether the node exists in the mutation event list. N2 and N5 both meet the conditions. After traversal, generate an abnormal node set with the storage structure of {N2: [coordinates (3.2, 5.6, 1.0), CO2 40ppm, H2 2.8% LEL], N5: [coordinates (6.8, 4.3, 0.9), CO2 55ppm, H2 3.1% LEL]}, excluding nodes N1, N3, and N6 that meet the credibility standards but have no mutation events.

[0106] The gradient interpolation compensation submodule extracts the CO and H2 concentration values ​​of its adjacent nodes in three-dimensional space based on the abnormal node set, calculates the gradient interpolation using the inverse distance weighted method, and overwrites the original data of the abnormal node with the compensation value to generate a compensated concentration set containing the corrected concentration values;

[0107] For example, the CO concentration at node N1 is 235ppm, N3 is 228ppm, and N4 is 242ppm. The data comes from the sensor readings during the monitoring period of 14:50-14:55. k Calculate by three-dimensional coordinate difference. For example, the coordinates of abnormal node N2 are (3.2, 5.6, 1.0) and the coordinates of adjacent node N1 are (2.8, 5.2, 1.1). Then:

[0108]

[0109] Weight Similarly, calculate N3's d3 = 0.72 meters, w3 = 1 / (0.72) 2 ≈1.93, total weight The normalized weight terms are 3.08 / 6.26≈0.492, 1.93 / 6.26≈0.308, and 1.25 / 6.26≈0.200;

[0110] ΔC k is the concentration difference between the adjacent nodes in the current and previous time windows. For example, the CO concentration of N1 in the period 14:45-14:50 is 230ppm, and the current period is 235ppm, then ΔC1=235-230=5ppm, Δt=5 minutes, Δt ref = 5 minutes, the time scale factor is 5 / 5 = 1, the denominator

[0111] Example substitution and calculation process:

[0112] The first part calculates the weighted concentration:

[0113]

[0114] The second part calculates the adjustment items:

[0115] Numerator:

[0116]

[0117] Denominator: 4.319;

[0118] Adjustment item: 4.20 / 4.319≈0.97ppm;

[0119] Final compensation concentration:

[0120] C comp =234.24+0.97=235.21ppm;

[0121] Parameter setting basis and rationality: weight It reflects the attenuation law of the concentration gradient due to distance. The closer the distance, the higher the weight. The reasonable range of sensor spacing in steel plants is 0.5-2 meters. The measured distance of 0.57 meters is in line with reality. The concentration difference ΔC k Calculated based on monitoring period data, time window Δt ref It is set to 5 minutes to meet industrial monitoring standards. The denominator of the adjustment item uses the sum of the reciprocals of the distances to balance the spatial distribution weights.

[0122] The result C comp =235.21ppm represents the CO concentration compensation value of the abnormal node N2. The compensation values ​​of all abnormal nodes are combined to generate a compensation concentration set. For example, the corrected concentration of N2 is 235.21ppm, and the corrected concentration of N5 is 248ppm. When the compensation value exceeds the safety threshold, an alarm is triggered.

[0123] The safety threshold determination submodule calls the compensation concentration set and compares the CO concentration and H2 concentration of each node with the corresponding concentration threshold one by one. If any concentration value exceeds the corresponding threshold, it is marked as a risk node. The time period when the proportion of risk nodes exceeds the proportion threshold is counted to generate real-time monitoring results of gas composition;

[0124] The CO concentration threshold is set to 250ppm and the H2 threshold is set to 3.0%LEL. The N2 data in the compensated concentration set (234.5ppm < 250ppm, 2.57%LEL < 3.0%LEL) is called and judged as safe. The CO value of N5 after compensation is 248ppm < 250ppm, and the H2 value is 2.95%LEL < 3.0%LEL, which is also safe. If the CO value of a node after compensation is 255ppm > 250ppm, it is marked as a risk node. The proportion of risk nodes in three monitoring periods (14:50-15:05) is calculated. For example, the number of risk nodes in period 1 is 0 / 6, the number of risk nodes in period 2 is 1 / 6 (accounting for 16.7%), and the number of risk nodes in period 3 is 2 / 6 (accounting for 33.3%). If the proportion threshold is set to 30%, only the condition is triggered in period 3, generating the real-time monitoring results of gas composition {time period: 15:00-15:05, number of risk nodes: 2, proportion: 33.3%}.

[0125] An online monitoring method for pressurized hot and stuffy gas components in steel slag comprises the following steps:

[0126] S1: Based on the gas sensor array layout inside the slag tank, the three-dimensional coordinate parameters of each sensor are detected, the time difference between adjacent sensors receiving CO and H2 gas signals is measured, and the time difference is linearly proportional to the sensor spacing to generate the sensor delay calibration coefficient including distance compensation;

[0127] S2: Call the sensor delay calibration coefficient to correct the gas concentration signal waveform, extract the CO and H2 concentration amplitude vectors in adjacent detection cycles, calculate the Euclidean distance ratio between the amplitude vectors, and compare it with the steel slag hot and stuffy working condition threshold preset in the dynamic calibration threshold library to generate the gas signal credibility coefficient;

[0128] S3: Based on the gas signal credibility coefficient, the gas concentration in the slag tank is screened to meet the standard node, the maximum and minimum values ​​of CO and H2 concentrations in continuous cycles are extracted, the absolute difference between the extreme values ​​is calculated, and the difference is matched with the allowable fluctuation range of the slag reaction gas concentration within the calibration error range to generate a gas concentration stable platform;

[0129] S4: Monitor the CO and H2 concentration change rate curves within the first detection cycle after the gas concentration stabilization platform is terminated, synchronously collect temperature and pressure sensor data in the slag tank, calculate the ratio of the temperature change rate to the pressure change rate, perform a logical judgment on this ratio and the ratio range set for the slag hot and stuffy safety working condition, and generate a gas concentration mutation event;

[0130] S5: Combined with the trigger status of the gas concentration mutation event and the gas signal credibility coefficient, the adjacent node gradient interpolation method is used to compensate for the missing CO and H2 concentration data of the abnormal nodes. The compensation values ​​are compared item by item with the steel slag pressurized hot and stuffy safety concentration threshold table to generate real-time monitoring results of gas composition.

[0131] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An online monitoring system for pressurized hot and stuffy gas components in steel slag, characterized in that: The system comprises: The sensor calibration module obtains the three-dimensional coordinate parameters of each sensor, measures the arrival time difference of gas signals from adjacent sensors, and calculates the ratio of the time difference to the sensor spacing to generate the sensor delay calibration coefficient. a signal credibility module, which extracts the gas concentration amplitude of the calibration node based on the sensor delay calibration coefficient, calculates the ratio of the amplitude vectors of adjacent cycles, and compares it with the dynamic calibration threshold to generate a gas signal credibility coefficient; The concentration stabilization module calls the gas signal credibility coefficient to screen the qualified nodes, extracts the CO and H2 concentration period extreme values, calculates the concentration difference and compares it with the calibration error range to generate a gas concentration stabilization platform; The anomaly detection module extracts the CO and H2 concentration change rates in the first cycle after the gas concentration stabilization platform is terminated, simultaneously obtains the temperature and pressure change rate ratio in the current cycle, compares the temperature and pressure ratio with the set ratio range, and generates a gas concentration mutation event; The real-time monitoring module combines the gas concentration mutation event with the gas signal credibility coefficient, performs gradient interpolation compensation on the abnormal node concentration data, determines whether the compensation value exceeds the safe concentration threshold, and generates real-time monitoring results of the gas composition.

2. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The sensor delay calibration coefficient includes three-dimensional coordinate calibration parameters and a delay scaling factor. The gas signal credibility coefficient specifically includes a credibility threshold range, a dynamic amplitude deviation ratio, and a node credibility weight. The gas concentration stabilization platform includes a stable concentration baseline value, a qualified node screening identifier, and a platform duration cycle number. The gas concentration mutation event specifically includes a mutation type classification identifier, a temperature-pressure correlation anomaly coefficient, and a trigger judgment threshold. The real-time monitoring results of the gas composition include an interpolated compensation concentration value, an exceeding-standard node positioning index, and a real-time safety assessment level.

3. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The sensor calibration module includes: The spatial parameter acquisition submodule obtains the three-dimensional coordinate parameters of each sensor, calculates the Euclidean distance between adjacent sensors, and generates a three-dimensional coordinate distance matrix containing the distances between all nodes; The delay scale factor calculation submodule calls the three-dimensional coordinate spacing matrix, synchronously collects the arrival time difference of the gas signals of adjacent sensors, divides the time difference by the corresponding spacing value, and obtains the delay scale factor set of each node pair; The abnormal node calibration submodule calls the delay scale factor set, compares each scale factor with the theoretical range of sound wave propagation delay one by one, filters out node pairs that exceed the upper or lower limit of the theoretical range, marks their status as abnormal, and generates a sensor delay calibration coefficient based on the scale factor deviation value of the abnormal node.

4. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The signal credibility module includes: The amplitude vector construction submodule calls the sensor delay calibration coefficient, extracts the gas concentration amplitude data of the calibration node, integrates the three-dimensional spatial distribution characteristics of each node according to the time period, and generates a three-dimensional amplitude vector set containing a multi-period amplitude sequence; A vector similarity analysis submodule calculates the ratio of the dot product of adjacent periodic vectors to the product of the modulus lengths of the two vectors based on the three-dimensional amplitude vector set, compares the ratio of each node with the dynamic calibration threshold on a period-by-period basis, and selects nodes whose ratios are lower than the threshold for three consecutive periods to generate a low-similarity node set; The offset tolerance determination submodule calls the low-similarity node set, extracts the signal derivative trajectories of its adjacent nodes, calculates the absolute value of the phase offset angle, compares the offset angle with the correction tolerance threshold, and marks the node as an abnormal node if it exceeds the threshold. The gas signal credibility coefficient is generated based on the ratio of the number of abnormal nodes to the total number of nodes.

5. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 4, characterized in that: To calculate the absolute value of the phase shift angle, use the formula: Where ΔA j Represents the difference in derivative trajectories of adjacent nodes in the jth dimension (x / y / z), ΔB j Represents the difference in derivative trajectories of low similarity nodes in the jth dimension, is the Euclidean distance l from the jth adjacent node to the low similarity node j is the inverse weight of , n is the total number of adjacent nodes, and θ is the absolute value of the weighted phase offset angle.

6. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The concentration stabilization module includes: The node screening submodule calls the gas signal credibility coefficient, compares each node coefficient with a set credibility threshold one by one, screens sensor nodes whose coefficients exceed the threshold, and generates a qualified node set containing valid monitoring points; The extreme value difference calculation submodule extracts the maximum and minimum CO concentrations and the maximum and minimum H2 concentrations of each node within the period based on the qualified node set, calculates the absolute values ​​of the extreme value differences of the two types of gas concentrations, and generates an extreme value difference sequence for all nodes; The stable window determination submodule calls the extreme value difference sequence, compares the difference of each node with the calibration error range of its sensor window by window, and counts the number of nodes whose difference values ​​of three consecutive windows are within the error range. If the proportion of the number of nodes exceeds the node number threshold, the current window end timestamp is recorded to generate a gas concentration stability platform.

7. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The anomaly detection module includes: The change rate extraction submodule calls the end time point of the gas concentration stabilization platform, extracts the CO concentration change rate and the H2 concentration change rate in the first period thereafter, and generates a concentration change rate set containing the change rates of the two types of gases; The parameter ratio calculation submodule synchronously obtains the temperature change rate and pressure change rate in the current cycle, divides the absolute value of the temperature change rate by the absolute value of the pressure change rate, and obtains the temperature-pressure ratio sequence at each time point; The mutation event determination submodule calls the concentration change rate set and temperature-pressure ratio sequence, compares the CO change rate with the set change rate threshold, and the H2 change rate with the set change rate threshold, and determines whether the temperature-pressure ratio exceeds the preset ratio range. If the change rates of both types of gases exceed the threshold and the temperature-pressure ratio exceeds the range, a gas concentration mutation event is generated.

8. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 1, characterized in that: The real-time monitoring module includes: The abnormal node screening submodule calls the gas concentration mutation event and the gas signal credibility coefficient, screens the nodes whose credibility coefficient is lower than the credibility threshold and has mutation events, and generates an abnormal node set containing spatial position and concentration data; The gradient interpolation compensation submodule extracts the CO and H2 concentration values ​​of its adjacent nodes in three-dimensional space based on the abnormal node set, calculates gradient interpolation using the inverse distance weighted method, overwrites the original data of the abnormal node with the compensation value, and generates a compensated concentration set containing the corrected concentration value; The safety threshold determination submodule calls the compensation concentration set and compares the CO concentration and H2 concentration of each node with the corresponding concentration thresholds one by one. If any concentration value exceeds the corresponding threshold, it is marked as a risk node. The time period when the proportion of risk nodes exceeds the proportion threshold is counted to generate real-time monitoring results of gas composition.

9. The online monitoring system for pressurized, hot, and stuffy gas components in steel slag according to claim 8, characterized in that: For gradient interpolation calculated by the inverse distance weighted method, the formula is: Among them, C k Measure the concentration value of the kth neighboring node, d k is the three-dimensional space distance from the kth adjacent node to the abnormal node, is the inverse squared distance weight, is the time series concentration difference, Δt is the length of the current time window, Δt ref It is the preset reference time window.

10. An online monitoring method for pressurized hot and stuffy gas components in steel slag, characterized in that: The method is used in the online monitoring system for pressurized hot and stuffy gas composition in steel slag according to any one of claims 1 to 7, comprising the following steps: S1: Based on the gas sensor array layout inside the slag tank, the three-dimensional coordinate parameters of each sensor are detected, the time difference between adjacent sensors receiving CO and H2 gas signals is measured, and the time difference is linearly proportional to the sensor spacing to generate the sensor delay calibration coefficient including distance compensation; S2: Calling the sensor delay calibration coefficient to correct the gas concentration signal waveform, extracting the CO and H2 concentration amplitude vectors in adjacent detection cycles, calculating the Euclidean distance ratio between the amplitude vectors, and comparing them with the steel slag hot stuffy working condition threshold preset in the dynamic calibration threshold library to generate the gas signal credibility coefficient; S3: Based on the gas signal credibility coefficient, the gas concentration in the slag tank is screened to meet the standard node, the maximum and minimum values ​​of CO and H2 concentrations in continuous periods are extracted, the absolute difference between the extreme values ​​is calculated, and the difference is matched with the allowable fluctuation range of the slag reaction gas concentration within the calibration error range to generate a gas concentration stable platform; S4: monitoring the CO and H2 concentration change rate curves within the first detection period after the gas concentration stabilization platform is terminated, synchronously collecting temperature and pressure sensor data in the slag tank, calculating the ratio of the temperature change rate to the pressure change rate, performing a logical judgment on the ratio and the ratio range set for the slag hot and stuffy safety working condition, and generating a gas concentration mutation event; S5: Combined with the trigger state of the gas concentration mutation event and the gas signal credibility coefficient, the adjacent node gradient interpolation method is used to compensate for the missing CO and H2 concentration data of the abnormal node, and the compensation value is compared item by item with the steel slag pressurized hot and stuffy safety concentration threshold table to generate real-time monitoring results of gas composition.

Images