Multi-parameter fusion diagnosis and maintenance method for health state of blast furnace taphole machine bearing seat

Abstract

The present application belongs to the technical field of equipment maintenance, and provides a health state multi-parameter fusion diagnosis and maintenance method for a blast furnace opening machine bearing seat, comprising: dividing the opening operation of the blast furnace opening machine bearing seat into driving loading, slag impact, advancing bearing and backstroke unloading stages, and constructing a mechanical response chain; dividing the operation process into high-impact, disturbance and low-load recovery furnace condition windows; mapping and aligning the current mechanical response chain under the corresponding furnace condition window with the historical stable mechanical response chain in terms of force transmission path consistency, and identifying abnormal response sections according to the path deviation degree and the characteristic out-of-bound condition; performing chain tracking on the abnormal response sections, collecting impact peak increment, unloading duration and residual vibration retention amplitude according to the window constraint, and dividing the bearing seat deterioration state; performing hierarchical maintenance under the constraint of the bearing seat deterioration state and the furnace condition window type, updating the furnace condition constraint boundary and the path deviation determination rule.

Description

Technical Field

[0001] This invention belongs to the field of equipment maintenance technology, specifically a multi-parameter integrated diagnosis and maintenance method for the health status of blast furnace opening machine bearing housing. Background Technology

[0002] The blast furnace tapping machine is a key piece of equipment in the blast furnace tapping operation. Its bearing housing bears the responsibility of supporting, transmitting force and guiding during the opening and closing process. Due to long-term exposure to high temperature, impact and alternating loads, the bearing housing is prone to wear, loosening and deformation. Therefore, it is usually necessary to monitor its operating status and carry out maintenance management.

[0003] In existing technologies, bearing housing condition monitoring typically involves analyzing parameters such as vibration, load, temperature, or displacement, and using fixed thresholds or historical statistical thresholds to determine whether the equipment is abnormal. However, in actual blast furnace production, furnace conditions such as slag impact intensity, slag landing point, tapping rhythm, and opening frequency continuously change, causing synchronous fluctuations in the load, vibration, and displacement responses of the bearing housing. This further alters the response sequence and attenuation process during force transmission. These changes may originate from furnace condition disturbances or from structural performance degradation such as bearing housing wear, loosening, or deformation. Existing technologies mostly rely on single-parameter amplitudes, statistical characteristics, or fixed thresholds for condition judgment, lacking correlation analysis of the force transmission relationship between load, vibration, and displacement, and also lacking a constraint mechanism for the consistency of mechanical response paths under different furnace conditions. It is difficult to determine whether abnormal responses originate from changes in furnace conditions or deterioration of the bearing housing itself. Especially when furnace condition fluctuations and early damage coexist, misjudgments or omissions are prone to occur, leading to distorted bearing housing health status assessment results and affecting the accurate formulation of maintenance timing and strategies.

[0004] Therefore, this invention provides a multi-parameter integrated diagnosis and maintenance method for the health status of blast furnace opening machine bearing housings. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is:

[0007] The blast furnace tapping machine bearing seat opening operation is divided into drive loading, slag impact, propulsion bearing and return unloading stages. Load, vibration and displacement signals are acquired synchronously in each stage according to the same force direction and sampling time window. Based on the force transmission direction, the multi-source signals are rearranged to construct a mechanical response chain.

[0008] Based on the slag impact intensity, impact point deviation, tapping rhythm and opening frequency, the operation process is divided into high impact, disturbance and low load recovery furnace condition windows, and the allowable load amplitude and attenuation characteristic boundary of the bearing seat are limited in each furnace condition window.

[0009] The current mechanical response chain is aligned with the historical stable mechanical response chain under the corresponding furnace condition window by mapping the force transmission path consistency. The response segment is traced in reverse along the impact peak path, load transmission path and unloading attenuation path, and abnormal response segments are identified according to the degree of path deviation and characteristic out-of-bounds conditions.

[0010] Chain tracking is performed on the abnormal response segment. The impact peak increment, unloading duration and residual vibration retention amplitude are collected according to window constraints. The cumulative path deviation is used as the degradation criterion to construct an irreversible degradation path and classify the bearing housing degradation state according to the cumulative offset level.

[0011] Under the constraints of bearing housing deterioration state and furnace condition window type, graded maintenance is performed, and after maintenance, mechanical response chains are collected according to the corresponding window constraint conditions to update furnace condition constraint boundaries and path deviation judgment rules.

[0012] As a further aspect of the present invention:

[0013] The process of constructing a mechanical response chain is as follows:

[0014] The load signals, vibration acceleration signals, and displacement signals collected at each stage are processed with a unified timestamp alignment.

[0015] A principal force coordinate system is established based on the force direction of the bearing housing structure, and the signals from different measuring points are reprojected directionally according to the force transmission path.

[0016] The reprojected multi-source signal is segmented by a sliding window, and the key response segments are enhanced and calibrated by combining the impact trigger point to form a staged response unit.

[0017] Based on the force transmission sequence and structural connection topology between the phased response units, a multi-source coupled mechanical response structure representing the force transmission relationship is constructed.

[0018] Normalization and noise suppression are performed on the multi-source coupled mechanical response structure, and a force transmission weighting coefficient is introduced for weighted sorting to generate a mechanical response chain that characterizes the entire opening and closing process.

[0019] The process of defining the allowable load amplitude and attenuation characteristic boundary of the bearing housing within each furnace condition window is as follows:

[0020] Energy equivalent conversion is performed on the impact intensity of slag, and impact space offset parameters are constructed by combining the impact point offset distance;

[0021] Based on the iron tapping rhythm cycle and the opening frequency, the operating rhythm characteristics are extracted, and the periodic disturbance components are identified by frequency domain decomposition.

[0022] Multidimensional correlation analysis was performed on impact energy parameters, spatial offset parameters, and rhythmic characteristics to construct a furnace condition distribution characterization space.

[0023] Based on the density aggregation characteristics and fluctuation separation degree in the furnace condition distribution characterization space, the high impact window, disturbance window and low load recovery window are divided;

[0024] Load envelope constraint curves are constructed within each window, and the upper limit of allowable load amplitude and the boundary of vibration decay time constant are obtained by combining historical stable operation sample statistics.

[0025] The process of aligning the current mechanical response chain with the historical stable mechanical response chain under the corresponding furnace condition window through a consistent force transmission path mapping is as follows:

[0026] Standard force transmission path templates are extracted from historical stable mechanical response chains to construct a path structure library containing impact peak nodes, transmission relay nodes, and unloading termination nodes;

[0027] The current mechanical response chain is divided into corresponding path sub-segments according to the same stage, and the timing alignment is performed using a dynamic time warping method;

[0028] Based on the force transmission direction consistency constraint, a standard path template corresponding to the current furnace condition window is retrieved from the path structure library, and node-level matching is performed between the current path sub-segment and the standard path template to calculate the path structure similarity.

[0029] By introducing structural connection topology constraints, inconsistent nodes are remapped and corrected to generate path deviation vectors.

[0030] Based on the path deviation vector and window constraints, determine the path correspondence and the degree of path deviation, and output the force transmission path consistency alignment result.

[0031] The process of tracing the response segment in reverse along the peak impact path, load transfer path, and unloading attenuation path is as follows:

[0032] Based on the consistency alignment results of the force transmission path, the correspondence between the impact peak node, load transmission node and unloading attenuation node is determined.

[0033] Starting from the peak impact node, the force transmission source node is traced along the load gradient direction to form the peak impact path.

[0034] Based on the correspondence between load transfer nodes, the coordinated change characteristics of vibration response and displacement response are analyzed, the energy conduction direction during load transfer is traced, and the load transfer path is formed.

[0035] Based on the vibration attenuation amplitude and displacement recovery characteristics during the unloading phase, the residual vibration diffusion process and energy release source are traced in reverse to form the unloading attenuation path;

[0036] By associating the peak impact path, load transfer path, and unloading attenuation path, analyzing the load, vibration, and displacement response characteristics of each node, constructing the force transfer relationship between nodes, and generating inverse response results for path deviation determination.

[0037] The process of identifying abnormal response segments based on the degree of path deviation and characteristic out-of-bounds conditions is as follows:

[0038] Based on the inverse response results, the impact peak path, load transfer path and unloading attenuation path are compared with the corresponding paths in the historical stable mechanical response chain, and the transfer offset and response deviation of each path node are calculated.

[0039] Based on the allowable load amplitude boundary and vibration attenuation characteristic boundary corresponding to the current furnace condition window, the load response, vibration response and displacement response of each path node are constrained and verified to identify abnormal feature points that exceed the window constraint range.

[0040] When the deviation of a path node exceeds the allowable offset range of the corresponding path, and abnormal feature points continue to appear on the same force transmission path, an abnormal response is determined to exist.

[0041] The abnormal propagation trajectory is traced along the mechanical response chain corresponding to the abnormal path to determine the abnormal extension section, and the abnormal extension section is marked as the abnormal response section.

[0042] The process of constructing an irreversible degradation path is as follows:

[0043] Within each furnace condition window, the abnormal response segment is traced along the mechanical response chain to extract the cumulative node offset and vibration energy along the impact peak path, load transfer path and unloading attenuation path.

[0044] The attenuation and retention amplitude and duration of each path node during the statistical unloading phase are combined with the cumulative path deviation results to form the path cumulative feature vector for each window.

[0045] The cumulative feature vectors of each window path are accumulated and fused in chronological order to obtain the cumulative path deviation index corresponding to each force transmission path.

[0046] Identifying irreversible evolutionary segments based on the monotonic growth trend of cumulative path deviation indicators;

[0047] The irreversible evolution segment is correlated and traced along the mechanical response chain to determine the continuous expansion range of the abnormal response segment, and the corresponding path is identified as the irreversible degradation path.

[0048] The process of classifying bearing housing deterioration status according to cumulative offset level is as follows:

[0049] The global offset score is calculated based on the irreversible degradation path, and a comprehensive degradation index is constructed by combining the impact frequency and vibration energy density.

[0050] Segmented cluster analysis was performed on the comprehensive degradation index to identify the boundary intervals between low degradation, moderate degradation, and severe degradation.

[0051] A window constraint correction factor is introduced to normalize and adjust the degradation index under different furnace condition windows;

[0052] Based on the adjusted cumulative offset level, the bearing housing status is divided into three categories: stable, early warning, and unstable. The bearing housing deterioration status is output with furnace condition labels and stage weights.

[0053] The process of performing graded maintenance under the constraints of bearing housing deterioration state and furnace condition window type is as follows:

[0054] Based on the deterioration state of the bearing housing and the current furnace condition window type, determine the corresponding maintenance intervention conditions;

[0055] When the bearing housing is in a stable state, perform lubrication compensation and inspection and maintenance of the connection parts within the low load recovery window;

[0056] When the bearing housing is in a warning state, local reinforcement, clearance adjustment, or replacement and maintenance of worn parts are carried out on the corresponding stress parts that deviate from the path within the disturbance window.

[0057] When the bearing housing is in an unstable state, the operation should be restricted under the high impact window, and the critical load-bearing parts corresponding to the irreversible deterioration path should be shut down for inspection and maintenance.

[0058] After maintenance, the mechanical response chain under the corresponding furnace condition window was re-acquired, and the load transfer characteristics, vibration attenuation characteristics, and displacement recovery characteristics were compared and verified.

[0059] The maintenance effect is confirmed based on the verification results, and a graded maintenance result corresponding to the deterioration state of the bearing housing is formed.

[0060] The process of updating furnace condition constraint boundaries and path deviation judgment rules is as follows:

[0061] After completing the corresponding maintenance work according to the graded maintenance results, load, vibration and displacement signals are re-acquired in the corresponding furnace condition window, and the mechanical response chain after maintenance is reconstructed according to the force transmission direction.

[0062] The path consistency of the maintained mechanical response chain is compared with that of the mechanical response chain before maintenance and the historical stable mechanical response chain. The recovery deviations corresponding to the impact peak path, load transfer path and unloading attenuation path are calculated.

[0063] Based on the load amplitude distribution and vibration attenuation characteristic distribution under the corresponding furnace condition window according to the recovery deviation statistics, update the load amplitude boundary and attenuation characteristic boundary corresponding to each furnace condition window.

[0064] Based on the changes in path deviation and abnormal response segment in the mechanical response chain after maintenance, the deviation judgment thresholds corresponding to each force transmission path are adjusted.

[0065] The updated furnace condition window constraint boundaries and path deviation judgment thresholds are written back to the mechanical response chain diagnosis process to form continuously iteratively updated furnace condition window constraint boundaries and path deviation judgment rules.

[0066] The beneficial effects of this invention are as follows:

[0067] By dividing the opening and closing process of the blast furnace tapper bearing housing into different stages such as drive loading, slag impact, propulsion bearing, and return unloading, and constructing a mechanical response chain based on multi-parameter signals such as load, vibration, and displacement that conforms to the actual force transmission law, the operating state of the bearing housing can be continuously characterized according to the force transmission process. This avoids the problem that traditional single vibration or temperature monitoring methods cannot accurately reflect complex force states. At the same time, different furnace condition windows are constructed by combining furnace condition factors such as slag impact intensity, impact point deviation, tapping rhythm, and tapping frequency. Differentiated constraint boundaries are established within the corresponding windows, effectively eliminating the interference of furnace condition fluctuations on diagnostic results and improving the accuracy and reliability of anomaly identification. Furthermore, by aligning the current mechanical response chain with the historical stable mechanical response chain through path consistency mapping, and performing path consistency mapping along the impact peak path, load transmission path, and unloading attenuation path, the system can continuously characterize the bearing housing operating state according to the force transmission process. Reverse tracing can accurately locate the source and propagation process of abnormal responses, enabling traceability analysis from abnormal results to the root cause. Simultaneously, by constructing irreversible degradation paths using accumulated path deviations, it can not only identify the current health status of the bearing housing but also reveal degradation evolution trends, improving early damage detection and lifespan assessment capabilities. Based on this, targeted graded maintenance is performed according to the bearing housing degradation status and furnace condition window type. Furthermore, the mechanical response chain after maintenance is used to continuously revise the furnace condition constraint boundaries and path deviation judgment rules, forming a closed-loop operation mechanism combining diagnosis, assessment, maintenance, and rule updates. This achieves dynamic monitoring, accurate diagnosis, and scientific maintenance of the blast furnace opener bearing housing health status, reducing the risk of false alarms and missed alarms, improving equipment operational stability and the rationality of maintenance decisions, reducing unplanned downtime, and extending the service life of the bearing housing and related mechanisms. Attached Figure Description

[0068] The invention will now be further described with reference to the accompanying drawings.

[0069] Figure 1 This is a flowchart of the steps in Embodiment 1 of the present invention; Detailed Implementation

[0070] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0071] Example 1: Please refer to Figure 1 As shown in the embodiment of the present invention, the multi-parameter fusion diagnosis and maintenance method for the health status of the blast furnace opening machine bearing housing includes the following steps:

[0072] Step S10: Divide the blast furnace tapping machine bearing seat opening operation into the stages of drive loading, slag impact, propulsion bearing and return unloading. In each stage, load, vibration and displacement signals are acquired synchronously according to the same force direction and sampling time window. Based on the force transmission direction, the multi-source signals are rearranged to construct a mechanical response chain.

[0073] In step S10, the process of constructing the mechanical response chain is as follows:

[0074] The load signals, vibration acceleration signals, and displacement signals collected at each stage are aligned with a unified timestamp. Load sensors, vibration acceleration sensors, and displacement sensors are arranged at the corresponding positions on the bearing housing, and a unified sampling clock is used for data acquisition. When the sampling frequencies of the sensors are inconsistent, a preset reference sampling frequency is used as a unified time reference. The collected raw data are sorted according to the timestamp, and missing sampling points are filled in using linear interpolation. The load signals, vibration acceleration signals, and displacement signals are mapped to the same time series to form a synchronous data matrix. Each moment in the matrix corresponds to the load value, vibration value, and displacement value under the same operating state.

[0075] A principal force coordinate system is established based on the force direction of the bearing housing structure. The signals from different measuring points are reprojected directionally along the force transmission path. The installation structure parameters of the bearing housing and the layout parameters of the opening machine drive mechanism are obtained. The principal force coordinate system is established with the direction of the driving force as the X-axis, the radial direction perpendicular to the driving force as the Y-axis, and the bearing housing axis as the Z-axis. The coordinate transformation matrix of each measuring point is constructed according to the installation angle of each measuring point. The vibration acceleration vector and displacement vector collected by each measuring point are transformed to the principal force coordinate system using the coordinate transformation matrix. The response component along the X-axis is extracted as the principal force response quantity, forming the load sequence, vibration sequence, and displacement sequence of each measuring point along the force transmission direction.

[0076] The reprojected multi-source signal is segmented by a sliding window, and key response segments are enhanced and calibrated by combining the impact trigger point to form a staged response unit. A fixed-length sliding window is used to continuously scan each response sequence along the time axis. When the load change rate within the window exceeds a preset threshold or the vibration peak exceeds a preset impact threshold, the corresponding time point is marked as the impact trigger point. An impact response interval is formed by extending forward and backward with the impact trigger point as the center. The maximum load value, maximum vibration value, displacement change, vibration decay rate, and response duration within the impact response interval are extracted as the characteristic parameters of the interval, and the corresponding interval is defined as a staged response unit.

[0077] Based on the force transmission sequence and structural connection topology between staged response units, a multi-source coupled mechanical response structure characterizing the force transmission relationship is constructed. A structural topology table is established according to the structural connection relationship between bearing housing, connecting flange, support base and drive mechanism. The response units are arranged according to the time sequence of their appearance. The transmission relationship between adjacent structural nodes is determined according to the topology table. When the response time of the later node is later than that of the earlier node and the time difference is within the preset transmission delay range, the connection edge between the corresponding nodes is established. Using the staged response unit as the structural node and the transmission relationship between the nodes as the connection edge, a multi-source coupled mechanical response structure containing load transmission relationship, vibration propagation relationship and displacement change relationship is constructed.

[0078] Normalization and noise suppression are performed on the multi-source coupled mechanical response structure. Force transmission weighting coefficients are introduced for weighted sorting to generate a mechanical response chain characterizing the entire opening and closing process. The maximum and minimum values ​​of load characteristics, vibration characteristics, and displacement characteristics under historical stable conditions are statistically analyzed, and the range normalization method is used to transform each characteristic to a unified numerical range. Median filtering is used to eliminate isolated outliers, and low-pass filtering is used to remove high-frequency noise. The contribution of each structural node to the final fault determination is statistically analyzed based on historical stable samples, and the contribution is converted into the force transmission weighting coefficient of the corresponding node. The characteristics of each node are weighted according to the weighting coefficients, and the node order is rearranged according to the force transmission sequence to finally form a mechanical response chain including an impact transmission segment, a load transmission segment, and an unloading recovery segment.

[0079] Understandably, the significance of step S10 lies in the following: by dividing the bearing housing opening and closing operation into typical stress stages such as drive loading, slag impact, propulsion bearing, and return unloading, and simultaneously collecting multi-source signals of load, vibration, and displacement under a unified force transmission direction, a consistent expression of different physical quantities in the time dimension and force direction can be achieved. Furthermore, by rearranging the signals based on the force transmission path, the originally dispersed multi-source monitoring data can be transformed into a mechanical response chain with structural significance, so that the load transmission process, vibration propagation process, and displacement response process of the bearing housing within a complete opening and closing cycle can be continuously expressed, providing a unified data foundation for path-level analysis and anomaly tracking.

[0080] Step S20: Based on the slag impact intensity, impact point offset, tapping rhythm and opening frequency, the operation process is divided into high impact, disturbance and low load recovery furnace condition windows, and the allowable load amplitude and attenuation characteristic boundary of the bearing seat are defined in each furnace condition window.

[0081] In step S20, the process of defining the allowable load amplitude and attenuation characteristic boundary of the bearing housing within each furnace condition window is as follows:

[0082] Energy equivalent conversion of slag impact intensity is performed, and impact spatial offset parameters are constructed by combining the impact point offset distance. Slag flow rate data, slag velocity data, and slag impact point monitoring data are acquired during the tapping process. The slag impact kinetic energy is calculated based on the slag mass flow rate and slag velocity per unit time. The impact kinetic energy at each moment is accumulated according to the sampling period to obtain the impact energy parameters for the corresponding time period. At the same time, the spatial offset distance between the actual slag impact point and the standard impact point is calculated using the preset standard impact point position as a reference. The longitudinal offset and lateral offset are calculated according to the offset direction. The impact energy parameters and the impact point offset distance are normalized and combined according to the preset weights to generate impact spatial offset parameters that characterize the degree of change in the position of slag impact.

[0083] Based on the iron tapping rhythm cycle and the opening frequency, the operating rhythm characteristics are extracted. Periodic disturbance components are identified through frequency domain decomposition. The operation record data of the opening machine in multiple consecutive iron tapping cycles are obtained. The time interval between two adjacent opening operations and the number of openings per unit time are statistically analyzed. The opening frequency sequence is segmented with a fixed time window to form an operating rhythm time series. The operating rhythm time series is processed by fast Fourier transform to convert the time domain change signal to the frequency domain space. The main frequency component, frequency energy distribution and frequency fluctuation amplitude in the spectrum are extracted. The periodic frequency peaks that exist continuously are identified. The corresponding frequency peaks and their energy proportions are used as periodic disturbance components, thereby forming an operating rhythm feature parameter set that characterizes the iron tapping rhythm change characteristics.

[0084] Multidimensional correlation analysis is performed on impact energy parameters, spatial bias parameters, and rhythmic characteristics to construct a furnace condition distribution representation space. Using impact energy parameters, impact spatial bias parameters, and operating rhythmic characteristic parameters as input variables, multidimensional feature vectors are constructed. Standardization processing is performed on each dimension of the features to transform features of different dimensions into a unified numerical range. The feature vectors corresponding to each time point are mapped to the multidimensional state space in chronological order, and the Euclidean distance and local density values ​​between adjacent samples are calculated. The furnace condition distribution representation space is established based on the degree of sample aggregation, so that each sample point corresponds to a furnace condition in the actual operation process, thereby obtaining the distribution law and changing trend between different furnace conditions.

[0085] Based on the density clustering characteristics and fluctuation separation degree in the furnace condition distribution space, the high impact window, disturbance window, and low load recovery window are divided. Cluster analysis is performed on the sample points in the furnace condition distribution space to calculate the local neighborhood density and inter-cluster distance of each sample point. When a sample has high impact energy, high spatial bias parameter, and high frequency disturbance characteristics, the corresponding sample is assigned to the high impact region. When a sample has obvious rhythmic fluctuations but the impact energy is at a medium level, the corresponding sample is assigned to the disturbance region. When the overall impact energy of the sample is low and the frequency fluctuation is small, the corresponding sample is assigned to the low load recovery region. Samples of the same type are merged according to the principle of time continuity to form the high impact window, disturbance window, and low load recovery window.

[0086] Load envelope constraint curves are constructed within each window, and the upper limit of allowable load amplitude and the boundary of vibration decay time constant are obtained by combining historical stable operation samples. Load signals and vibration signals within the corresponding windows under historical stable operation conditions are extracted respectively. For load signals, the maximum, minimum, and average load values ​​at each sampling time are statistically analyzed, and spline interpolation is used to connect the maximum value points to form the upper envelope curve, and the minimum value points to form the lower envelope curve, thereby constructing the load envelope constraint curve for the corresponding window. For vibration signals, the moment of impact peak occurrence is used as the starting point, and the vibration decay process is fitted with exponential decay to obtain the corresponding vibration decay time constant. The distribution range of vibration decay time constant in historical stable samples is statistically analyzed, and the boundary of vibration decay time constant is determined according to a preset confidence interval. Finally, the upper boundary of the load envelope constraint curve is used as the upper limit of allowable load amplitude, and the boundary of vibration decay time constant is used as the decay characteristic boundary under the corresponding furnace condition window, providing constraint conditions for mechanical response chain anomaly identification and path deviation determination.

[0087] It is understandable that the significance of step S20 is as follows: by introducing parameters that are strongly related to the operating conditions, such as slag impact intensity, impact point offset, tapping rhythm and opening frequency, the furnace condition window is divided into the operating process, so that the mechanical response of the bearing housing can establish a correspondence with the specific production conditions. At the same time, the allowable load amplitude and vibration attenuation characteristic boundary are limited within different furnace condition windows, thereby structurally constraining the normal operating condition fluctuation range, solving the problem of indistinguishable response differences under different furnace conditions, and enabling subsequent anomaly identification to no longer rely on fixed thresholds, but to have the ability to adapt to the operating conditions.

[0088] Step S30: Align the current mechanical response chain with the historical stable mechanical response chain under the corresponding furnace condition window by performing a force transmission path consistency mapping. Track the response segment in reverse along the impact peak path, load transmission path and unloading attenuation path, and identify abnormal response segments based on the degree of path deviation and characteristic over-limit conditions.

[0089] In step S30, the process of aligning the current mechanical response chain with the historical stable mechanical response chain under the corresponding furnace condition window through a consistent force transmission path mapping is as follows:

[0090] Standard force transmission path templates are extracted from historical stable mechanical response chains to construct a path structure library containing impact peak nodes, transmission relay nodes, and unloading termination nodes. Mechanical response chains corresponding to high impact windows, disturbance windows, and low load recovery windows under historical stable operating conditions are selected as benchmark samples. In each mechanical response chain, the location of load peak occurrence, vibration energy transmission location, and response recovery end location are identified and defined as impact peak nodes, transmission relay nodes, and unloading termination nodes, respectively. The load amplitude, vibration energy, displacement change, node duration, and transmission time difference between nodes are recorded for each node. Node sequences are established according to the order of node occurrence, and standard force transmission path templates are generated based on the connection relationship between nodes. Statistical analysis is performed on multiple stable samples under the same furnace condition window, and path structures with a frequency exceeding a preset proportion are retained to construct a path structure library corresponding to the furnace condition window.

[0091] The current mechanical response chain is divided into corresponding path segments according to the same stage, and time alignment is performed using a dynamic time warping method. Based on the time boundaries corresponding to the driving loading stage, slag impact stage, propulsion bearing stage, and return unloading stage, the current mechanical response chain is divided into stages. Continuous node sequences are extracted within each stage to form path segments. The node time series of the current path segment and the standard path template are obtained respectively, and a time distance matrix is ​​constructed. The optimal matching path between the two sets of time series is calculated using a dynamic time warping algorithm. Time offsets caused by differences in running speed are eliminated through insertion point mapping, compression mapping, and expansion mapping. After alignment, a node correspondence table is output so that the current path segment and the standard path template are under a unified time reference.

[0092] Based on the constraint of consistent force transmission direction, a standard path template corresponding to the current furnace condition window is retrieved from the path structure library. Node-level matching is performed between the current path sub-segment and the standard path template to calculate the path structure similarity. The furnace condition window type to which the current mechanical response chain belongs is read, and a standard path template is extracted from the corresponding path structure library. A node transmission direction matrix is ​​established based on the force direction of the bearing seat structure. The node transmission direction of the current path sub-segment is compared node by node with that of the standard path template. When the node transmission direction is consistent and the node category is the same, a node matching relationship is established. For nodes with established matching relationships, load deviation, vibration deviation, displacement deviation, and transmission time deviation are calculated respectively. After normalizing each deviation value, a weighted calculation is performed according to a preset weight to obtain the node matching score. The scores of all matching nodes are accumulated to generate a path structure similarity index between the current path and the standard path.

[0093] Structural connection topology constraints are introduced to remap and correct inconsistent nodes, generating path deviation vectors. A topology connection matrix is ​​established based on the structural connection relationships between bearing housings, support structures, connecting flanges, and drive mechanisms. When node matching is interrupted by missing nodes, abnormal node order, or inconsistent node types, adjacent nodes with direct connection relationships to the current node are searched according to the topology connection matrix. If the adjacent node satisfies the transmission time difference constraint and force direction constraint, the adjacent node is remapped to the corresponding standard node position. After all nodes are corrected, the deviation values ​​of each node in the load dimension, vibration dimension, displacement dimension, and time dimension are calculated respectively, and the nodes are combined according to the node order to form a path deviation vector. The path deviation vector is used to characterize the deviation between the current mechanical response chain and the standard force transmission path.

[0094] Based on the path deviation vector and window constraints, the path correspondence and path deviation degree are determined, and the force transmission path consistency alignment result is output. The allowable load amplitude boundary and vibration decay time constant boundary corresponding to the current furnace condition window are obtained. The deviation values ​​of each dimension in the path deviation vector are compared with the corresponding boundaries to identify abnormal deviation items that exceed the constraint range. The number of abnormal deviation items, the proportion of abnormal nodes, and the cumulative deviation value in the path deviation vector are counted, and the path deviation index is calculated. The correspondence level between the current path and the standard path is determined according to the path deviation index. When the path deviation index is lower than the first threshold, it is judged as a consistent path. When it is higher than the first threshold but lower than the second threshold, it is judged as a slightly deviated path. When it is higher than the second threshold, it is judged as an abnormal deviation path. Finally, the force transmission path consistency alignment result containing node correspondence, path structure similarity, path deviation vector, and path deviation index is output and used as input data for abnormal response segment identification.

[0095] In step S30, the process of reverse tracing the response segment along the impact peak path, load transfer path, and unloading attenuation path is as follows:

[0096] Based on the force transmission path consistency alignment results, the correspondence between impact peak nodes, load transfer nodes, and unloading attenuation nodes is determined. The node correspondence table is read from the force transmission path consistency alignment results, and the node mapping between the current mechanical response chain and the standard path template is extracted. Nodes where the load reaches a local maximum value are marked as impact peak nodes, nodes located at structural connection points and undertaking load transfer functions are marked as load transfer nodes, and nodes located in the return unloading stage and representing vibration recovery are marked as unloading attenuation nodes. A correspondence table between the current nodes and standard nodes is established, and the load value, vibration value, displacement value, and time information of each node are recorded to provide basic data for reverse tracing.

[0097] Starting from the peak impact node, the force transmission source node is traced along the load gradient direction to form the peak impact path. The peak impact node is selected as the starting point for tracing. The load change sequence within the continuous time window before the node is read, and the load gradient of adjacent nodes is calculated. The search is performed step by step in the direction of gradient from high to low. If there is a structural connection between adjacent nodes and the load change direction is consistent with the direction of the main force, it is determined as the source node of the next level. The tracing continues until the driving force input node is reached or the gradient is lower than the preset threshold. All traced nodes are connected in time order and structural connection order to form the peak impact path. At the same time, the load increment and transmission time of each node in the path are recorded.

[0098] Based on the correspondence between load transfer nodes, the coordinated change characteristics of vibration response and displacement response are analyzed, the energy conduction direction during load transfer is traced, and a load transfer path is formed. The vibration and displacement sequences corresponding to the load transfer nodes are extracted, the vibration energy change rate and displacement change rate of adjacent nodes are calculated, and the node response sequence is established by combining the order of vibration peak occurrence and displacement delay time. When the vibration energy change rate of a node is higher than that of the adjacent nodes and the response time is earlier than that of the adjacent nodes, it is determined that the energy is transferred from that node to the subsequent nodes. The nodes are connected sequentially according to the energy conduction direction, and node chains that do not conform to the structural topology are eliminated. Finally, a load transfer path representing the load and vibration energy conduction is formed, and the node vibration attenuation rate, displacement change and transmission delay time are recorded.

[0099] Based on the vibration attenuation amplitude and displacement recovery characteristics during the unloading phase, the residual vibration diffusion process and energy release source are traced in reverse to form an unloading attenuation path. The vibration attenuation curve and displacement recovery curve of each node during the unloading phase are obtained, and the vibration attenuation rate, residual vibration amplitude and displacement recovery time are calculated. Starting from the node with the largest residual vibration amplitude, adjacent nodes are searched step by step along the direction of decreasing vibration attenuation rate. If there is a structural connection between adjacent nodes and the residual vibration propagates continuously, the diffusion relationship between nodes is established. The energy release sequence is analyzed in combination with the displacement recovery time difference to determine the diffusion path of residual energy. The process is traced until the vibration amplitude drops to the background noise level to form an unloading attenuation path. At the same time, the vibration diffusion amount and recovery delay amount of each node are recorded.

[0100] The impact peak path, load transfer path, and unloading attenuation path are correlated, and the load, vibration, and displacement response characteristics of each node are analyzed to construct the force transfer relationship between nodes, forming the inverse response result for path deviation determination. The nodes in the three paths are mapped according to a unified time base to establish a node correspondence matrix. The load change, vibration energy change, and displacement change of each node are statistically analyzed, and the parameter transfer ratio of adjacent nodes is calculated. The force transfer relationship network between nodes is constructed by combining the load transfer order, vibration diffusion order, and displacement recovery order. The transfer efficiency, energy attenuation rate, and response lag time between nodes are calculated, and these parameters are combined into an inverse response feature vector. Finally, the node correlation, force transfer network, and inverse response feature vector are output for path deviation determination and abnormal response segment identification.

[0101] In step S30, the process of identifying abnormal response segments based on the degree of path deviation and characteristic out-of-bounds conditions is as follows:

[0102] Based on the inverse response results, the impact peak path, load transfer path, and unloading attenuation path are compared with the corresponding paths in the historical stable mechanical response chain. The transfer offset and response deviation of each path node are calculated. Based on the inverse response results, each node in the impact peak path, load transfer path, and unloading attenuation path is compared with the corresponding node in the historical stable mechanical response chain. For each node, the load peak difference, vibration amplitude difference, and displacement difference are calculated, and further weighted combinations are used to obtain node-level deviation indices. These deviation indices are used to generate path deviation vectors, and the deviation distribution characteristics of nodes in the path are statistically analyzed to form a complete path deviation matrix, providing a quantitative basis for anomaly judgment. For each node, the load response, vibration response, and displacement response of the node are compared with the allowable load amplitude boundary and vibration attenuation characteristic boundary under the corresponding furnace condition window. Using a threshold judgment method, nodes exceeding the boundaries are identified as abnormal feature points. At the same time, for the vibration and displacement co-features, the vibration energy density and displacement recovery time can be calculated to see if they exceed the preset safety range. Nodes that violate multiple constraints simultaneously are weighted and marked to clarify the degree of anomaly and key locations.

[0103] Based on the allowable load amplitude boundary and vibration attenuation characteristic boundary corresponding to the current furnace condition window, the load response, vibration response and displacement response of each path node are constrained and verified to identify abnormal feature points that exceed the window constraint range. The degree of deviation of the path nodes is analyzed for continuity. When the deviation of consecutive nodes exceeds the allowable offset range of the corresponding path, and these abnormal feature points continue to appear along the same path, it is determined that there is an abnormal response in the path. The sliding window method is used to scan the path to ensure that short-term occasional anomalies are not misjudged. At the same time, the intensity of the anomaly can be evaluated by statistically analyzing the cumulative amount of path deviation, providing a basis for maintenance.

[0104] When the deviation of a path node exceeds the allowable offset range of the corresponding path, and abnormal feature points continue to appear on the same force transmission path, an abnormal response is determined to exist. Along the identified abnormal path, the deviation of the force-bearing nodes in the mechanical response chain is traced forward and backward level by level. Combining the structural connection relationship between nodes and the force transmission direction, the propagation range of the abnormal response in the link is determined. Vibration attenuation, load transmission and displacement recovery characteristics are dynamically analyzed. Node areas that continuously deviate or accumulate deviations exceeding the threshold are marked as abnormal extension segments, thereby forming a complete abnormal response segment and providing data support for the construction of deteriorated paths.

[0105] The abnormal propagation trajectory is traced along the mechanical response chain corresponding to the abnormal path to determine the abnormal extension segment, and the abnormal extension segment is identified as an abnormal response segment. The traced abnormal extension segments are integrated to generate an abnormal response segment list. Each segment includes a start node, an end node, a path type, a deviation index, and abnormal characteristic statistics. These abnormal response segments are output to the diagnostic system for subsequent chain tracking, degradation assessment, and graded maintenance, ensuring that the abnormal detection results are quantifiable, traceable, and engineering feasible.

[0106] Understandably, the significance of step S30 lies in: by mapping and aligning the current mechanical response chain with the historical stable mechanical response chain through a consistent force transmission path mapping, and by tracing back along the impact peak path, load transmission path, and unloading attenuation path, a structured comparative analysis of the bearing housing's mechanical response process is achieved. This enables the differentiation between "normal response changes caused by operating conditions" and "path deviations caused by structural deterioration" under furnace condition constraints, and the identification of multi-dimensional anomalies is achieved through the degree of path deviation and characteristic out-of-bounds conditions, thereby improving the pertinence and reliability of anomaly judgment.

[0107] Step S40: Perform chain tracking on the abnormal response segment, collect the impact peak increment, unloading duration and residual vibration retention amplitude according to window constraints, use the cumulative path deviation as the degradation criterion, construct the irreversible degradation path, and classify the bearing housing degradation state according to the cumulative offset level.

[0108] In step S40, the process of constructing the irreversible degradation path is as follows:

[0109] Within each furnace condition window, the abnormal response segment is traced along the mechanical response chain to extract the cumulative node offset and vibration energy along the impact peak path, load transfer path, and unloading attenuation path. Within each furnace condition window, the identified abnormal response segment is traced node by node along the impact peak path, load transfer path, and unloading attenuation path. For each node, the cumulative amount of load, vibration, and displacement deviating from historical stable values ​​is calculated. At the same time, the vibration response is processed by energy integration to obtain the cumulative vibration energy. Timestamp synchronization and node number matching technology are used to ensure that the data of each node corresponds consistently. The node offset and energy accumulation information are automatically extracted and stored through programming to provide a quantitative basis for path accumulation analysis.

[0110] The attenuation and retention amplitude and duration of each path node during the unloading stage are statistically analyzed. Combined with the cumulative path deviation results, a path cumulative feature vector for each window is formed. For each node on the path, the attenuation curve of vibration amplitude over time is analyzed during the unloading stage. The retention amplitude and attenuation duration are extracted. The cumulative deviation, vibration energy and attenuation retention characteristics of the node are integrated into a vector to form the path cumulative feature vector of each path under the current furnace condition window. Normalization is used to ensure the comparability of the characteristics of different nodes. The feature information of each path is stored in matrix form to facilitate subsequent multi-window cumulative fusion and trend analysis.

[0111] The cumulative feature vectors of each window path are accumulated and fused in chronological order to obtain the cumulative path deviation index corresponding to each force transmission path; the cumulative feature vectors of the paths obtained under each furnace condition window are accumulated and fused in chronological order to obtain the cumulative force transmission path deviation index across windows. During the fusion process, a weighted average or exponential decay weighting method can be used to increase the weight of recent window anomalies to reflect the dynamic evolution trend of equipment deterioration. The output cumulative path deviation index not only quantifies the degree of path deviation but also reflects the abnormal accumulation trend, providing a basis for the identification of irreversible deterioration sections.

[0112] Identify irreversible evolution segments based on the monotonous growth trend of the cumulative path deviation index; perform monotonous growth trend analysis on the cumulative path deviation index, and use sliding window regression or difference method to determine whether the cumulative index continues to rise without significant regression. When the cumulative index continuously exceeds the set threshold and shows monotonous growth, the corresponding segment is marked as an irreversible evolution segment. Statistical tests or setting a minimum number of consecutive growth nodes are used to avoid misjudging occasional fluctuations as irreversible evolution.

[0113] The irreversible evolution segment is traced along the mechanical response chain to determine the continuous expansion range of the abnormal response segment, and the corresponding path is identified as the irreversible degradation path. The identified irreversible evolution segment is traced along the impact peak path, load transfer path, and unloading attenuation path to analyze the node overlap and continuity with the abnormal response segment, determine its actual range on the mechanical response chain, and mark the path corresponding to the range as the irreversible degradation path. At the same time, information such as the start and end nodes of the path, cumulative offset, vibration energy, and duration are recorded to provide data support for the classification of bearing housing degradation levels and graded maintenance, and realize the engineering implementation of path tracing.

[0114] In step S40, the process of classifying the bearing housing deterioration state according to the cumulative offset level is as follows:

[0115] A global offset score is calculated based on the irreversible deterioration path, and a comprehensive deterioration index is constructed by combining impact frequency and vibration energy density. Statistical analysis is performed on the identified irreversible deterioration path, extracting the cumulative path offset, cumulative vibration energy, and abnormal duration corresponding to each node in the path. Different weights are assigned according to the path's position in the overall force transmission chain, with higher weights given to nodes closer to the impact input end and the main load-bearing parts. The offsets of each node are weighted and summed to obtain the global offset score. At the same time, the number of impact peak events within a preset period is counted to calculate the impact frequency parameter. Energy integration processing is performed on the vibration signal, and vibration energy density parameters are calculated by combining the volume of the structural area corresponding to the node. The global offset score, impact frequency parameter, and vibration energy density parameter are normalized and combined according to preset weights to form a comprehensive deterioration index to quantitatively characterize the overall deterioration degree of the bearing housing.

[0116] Segmented cluster analysis was performed on the comprehensive degradation index to identify the boundary intervals between low, moderate, and severe degradation. Historical comprehensive degradation indices and actual maintenance records were collected to establish a degradation sample database. The comprehensive degradation indices were arranged into a sample sequence according to time sequence, and K-means clustering or hierarchical clustering methods were used to group the samples. Based on the clustering results, the average offset, average vibration energy, and actual wear of each category of samples were statistically analyzed. Samples with relatively light wear and long-term stability were defined as the low degradation interval; samples with significant wear but not yet affecting operation were defined as the moderate degradation interval; and samples with structural loosening, increased gaps, or local failure risks were defined as the severe degradation interval. Finally, the comprehensive degradation index range corresponding to each interval was determined, forming the boundary intervals between low, moderate, and severe degradation.

[0117] A window constraint correction factor is introduced to normalize the degradation index under different furnace condition windows. The furnace condition window type corresponding to the current comprehensive degradation index is obtained, and the load amplitude boundary, vibration attenuation boundary, and impact energy level parameters corresponding to the window are read. The influence of various furnace condition windows on mechanical response is statistically analyzed based on historical stable samples, and the window correction coefficient is calculated. When the current window belongs to the high impact window, the impact response weight is increased and the influence of short-term vibration anomalies on degradation judgment is reduced. When the current window belongs to the disturbance window, the path deviation persistence weight is increased. When the current window belongs to the low load recovery window, the residual vibration and recovery delay characteristic weights are increased. The comprehensive degradation index is normalized and adjusted using the window correction coefficient to eliminate the influence of different furnace condition conditions on the degradation assessment results, so that the degradation index obtained under different windows has a unified evaluation standard.

[0118] Based on the adjusted cumulative offset level, the bearing housing status is divided into three categories: stable, warning, and unstable. The bearing housing deterioration status is output with furnace condition labels and stage weights. The adjusted comprehensive deterioration index is compared with a pre-established deterioration level range. When the comprehensive deterioration index is in the low deterioration range, the bearing housing is classified as stable; when it enters the moderate deterioration range, it is classified as warning; and when it reaches the severe deterioration range, it is classified as unstable. Simultaneously, the main deterioration characteristics originating from the operation stages are statistically analyzed, including the drive loading stage, slag impact stage, propulsion bearing stage, and return unloading stage. The contribution ratio of each stage to the comprehensive deterioration index is calculated to generate stage weight parameters. Combined with the current furnace condition window type, corresponding furnace condition labels are added to the diagnostic results. Finally, the bearing housing deterioration status result, including deterioration level, furnace condition label, stage weight, comprehensive deterioration index, and main abnormal path information, is output, providing a basis for graded maintenance decisions.

[0119] It is understandable that the significance of step S40 is as follows: by chain-tracking the abnormal response segment and introducing dynamic features such as the impact peak increment, unloading duration and residual vibration amplitude, the abnormal propagation process can be continuously characterized. At the same time, the cumulative path deviation is used as the degradation criterion, which improves the equipment condition assessment from single-point abnormality detection to full-path cumulative analysis. This enables the identification of the process of the bearing housing evolving from a slight abnormality to irreversible degradation, and further constructs the irreversible degradation path, thereby realizing a structured expression and traceable description of the long-term degradation trend of the equipment.

[0120] Step S50: Perform graded maintenance under the constraints of bearing housing deterioration state and furnace condition window type, and collect mechanical response chain according to the corresponding window constraint conditions after maintenance, update furnace condition constraint boundary and path deviation judgment rule;

[0121] In step S50, the process of performing graded maintenance under the constraints of bearing housing deterioration state and furnace condition window type is as follows:

[0122] Based on the bearing housing deterioration status and the current furnace condition window type, the corresponding maintenance intervention conditions are determined; the current deterioration status judgment result of the bearing housing and the furnace condition window type corresponding to the current operation are obtained, and a maintenance decision table corresponding to the deterioration level and furnace condition window is established. The maintenance decision table pre-records the maintenance trigger conditions corresponding to the stable state, early warning state and unstable state, as well as the allowed maintenance methods corresponding to the high impact window, disturbance window and low load recovery window. When the system receives the latest deterioration status result, it first matches the current furnace condition window type, then reads the intervention conditions in the corresponding maintenance decision table, and combines the irreversible deterioration path location, the distribution range of abnormal response segments and the comprehensive deterioration index to sort the maintenance priorities, determine the maintenance level and maintenance object to be performed on the current bearing housing, and form a maintenance task instruction.

[0123] When the bearing housing is in a stable state, lubrication compensation and connection inspection and maintenance are performed within the low load recovery window. When the bearing housing is determined to be in a stable state and is currently in the low load recovery window, the bearing housing lubrication record and running time data are retrieved to determine the grease consumption. Grease is added to the rolling contact parts of the bearing housing according to the preset lubrication cycle. The tightness of the connection parts between the bearing housing and the support structure, connecting flange and fixing bolts is checked. The preload of the connecting parts is measured using a torque measuring tool and compared with the standard preload range. For connection parts with slight looseness, re-tightening is performed. After maintenance is completed, the lubrication compensation amount and connection status parameters are recorded as maintenance history data.

[0124] When the bearing housing is in a warning state, local reinforcement, clearance adjustment, or replacement of worn parts are performed on the stress-bearing parts corresponding to the path deviation within the disturbance window. When the bearing housing is determined to be in a warning state, the path deviation analysis results and the distribution of irreversible deterioration paths are read to locate the stress nodes that have continuously deviated. The maintenance objects are determined based on the structural positions corresponding to the abnormal nodes. Reinforcement is performed on the support structure that has become loose. Clearance adjustment is performed on the bearing mating clearance that exceeds the allowable range. Replacement is performed on the bushings, connecting pins, or bearing assemblies whose wear exceeds the preset wear threshold. During the maintenance process, the maintenance location, maintenance method, and replacement parts information are recorded simultaneously and correlated with the corresponding path deviation data to form a targeted maintenance record.

[0125] When the bearing housing is in an unstable state, the opening and closing operation is restricted under the high impact window, and the key load-bearing parts corresponding to the irreversible deterioration path are shut down for inspection and maintenance; when the bearing housing is determined to be in an unstable state, an operation restriction command is sent to the opening machine control system to reduce the frequency of opening and closing operations or suspend high-intensity iron tapping operations. The key load-bearing parts are located according to the irreversible deterioration path, including the main load-bearing bearing area, the support connection area and the impact load concentration area. The corresponding parts are shut down for disassembly and inspection, and the bearing wear, structural cracks, connection deformation and clearance changes are detected. Parts with severe wear or structural damage are replaced, and parts with structural deformation are corrected or reinforced. After the maintenance is completed, the equipment is reassembled and the trial operation inspection is completed according to the prescribed process.

[0126] After maintenance, the mechanical response chain under the corresponding furnace condition window is re-acquired, and the load transfer characteristics, vibration attenuation characteristics, and displacement recovery characteristics are compared and verified. After maintenance, the monitoring process is restarted under the same furnace condition window as before maintenance. Load signals, vibration signals, and displacement signals are collected synchronously according to the mechanical response chain construction method, and the mechanical response chain after maintenance is reconstructed. Key characteristic parameters in the impact peak path, load transfer path, and unloading attenuation path are extracted, including load transfer efficiency, vibration attenuation rate, residual vibration amplitude, and displacement recovery time. The characteristic parameters after maintenance are compared with the data before maintenance and historical stable sample data. The characteristic recovery rate and the degree of improvement of path deviation are calculated to form the maintenance verification results.

[0127] Based on the verification results, the maintenance effect is confirmed, and a graded maintenance result corresponding to the deterioration state of the bearing housing is generated. According to the maintenance verification results, the reduction rate of path deviation, the reduction ratio of abnormal response segments, and the improvement degree of comprehensive deterioration index are statistically analyzed. When the path deviation is restored to the preset allowable range and the abnormal response segments are eliminated, the maintenance is deemed effective. When some abnormal characteristics still exist but the deviation is significantly reduced, the maintenance is deemed partially effective, and a follow-up maintenance plan is generated. When the path deviation continues to increase after maintenance, the maintenance is deemed ineffective, and the maintenance level is increased or a new overhaul plan is formulated. Finally, a graded maintenance result including maintenance level, maintenance object, maintenance measures, maintenance effect evaluation, and follow-up suggestions is output, and the result is written back to the bearing housing health management database to provide a basis for updating furnace condition constraint boundaries and revising path deviation judgment rules.

[0128] In step S50, the process of updating the furnace condition constraint boundary and the path deviation determination rule is as follows:

[0129] After completing the corresponding maintenance work according to the graded maintenance results, load, vibration and displacement signals are re-acquired in the corresponding furnace condition window, and the mechanical response chain after maintenance is reconstructed according to the force transmission direction. After completing the graded maintenance work, the low-load or limited-load opening operation of the blast furnace opening machine bearing seat is started, and load sensor, vibration acceleration sensor and displacement sensor data are collected synchronously in the corresponding furnace condition window. The collected data is timestamped and aligned, and a main force coordinate system is established according to the force direction of the bearing seat. The multi-point sensor signals are reprojected according to the force transmission path. Through sliding window segmentation and impact trigger point enhancement calibration, the stage response units of each stage after maintenance are constructed, and a complete mechanical response chain after maintenance is formed based on the force transmission sequence and structural topology relationship, providing basic data for path deviation analysis.

[0130] The maintained mechanical response chain is compared with the mechanical response chain before maintenance and the historical stable mechanical response chain to calculate the recovery deviations corresponding to the impact peak path, load transfer path, and unloading attenuation path. The maintained mechanical response chain is divided into impact peak segment, load transfer segment, and unloading attenuation segment according to the operation stage. Node-level path matching is performed with the mechanical response chain before maintenance and the historical stable mechanical response chain. The time sequence is aligned using a dynamic time warping method. The deviations of the corresponding nodes at the impact peak node, transfer relay node, and unloading termination node are calculated, including load peak deviation, vibration amplitude deviation, and displacement recovery deviation. The recovery deviations of each path node are statistically analyzed to generate deviation vectors for the impact peak path, load transfer path, and unloading attenuation path, providing a quantitative basis for updating boundaries and judgment thresholds.

[0131] Based on the load amplitude distribution and vibration attenuation characteristic distribution under the corresponding furnace condition window according to the recovery deviation statistics, the load amplitude boundary and attenuation characteristic boundary corresponding to each furnace condition window are updated; based on the deviation vector of the mechanical response chain after maintenance, the load amplitude, vibration peak value and unloading attenuation time in each furnace condition window are statistically analyzed, the mean, variance and fluctuation range are calculated, the statistical results are fused with the corresponding characteristics of the historical stable mechanical response chain, the upper limit of load amplitude and vibration attenuation time constant boundary of each furnace condition window are refitted, and a new boundary curve or table is established to represent it. For windows with periodic disturbances, the frequency domain decomposition method is used to identify abnormal frequency components and the boundary is corrected to ensure that the furnace condition constraint boundary can reflect the real operating capacity of the bearing housing after maintenance.

[0132] Based on the changes in path deviation and abnormal response segments in the mechanical response chain after maintenance, the deviation judgment thresholds corresponding to each force transmission path are corrected. Using the changes in path deviation and abnormal response segments in the mechanical response chain after maintenance, the node deviation amplitude, cumulative path offset, and abnormal segment length of each force transmission path are calculated. The above indicators are compared with the corresponding values ​​before maintenance and historical stable values ​​to determine the deviation judgment threshold correction coefficients at the node and path levels. The thresholds are normalized and window constraint corrected so that the judgment rules can adapt to the changes in structural stiffness, wear state, and lubrication condition after maintenance, thereby ensuring the accuracy of subsequent abnormal response identification.

[0133] The updated furnace condition window constraint boundaries and path deviation judgment thresholds are written back to the mechanical response chain diagnostic process, forming continuously iteratively updated furnace condition window constraint boundaries and path deviation judgment rules. The updated furnace condition window load amplitude boundaries, vibration attenuation boundaries, and corrected path deviation judgment thresholds are uniformly recorded and written back to the mechanical response chain diagnostic system. An iterative update mechanism is established within the system, and the above process is executed after each graded maintenance, realizing the dynamic updating of furnace condition window constraint boundaries and path deviation judgment rules. The system can automatically call the latest constraint boundaries and judgment thresholds in subsequent operations to perform anomaly detection and deterioration assessment on the mechanical response chain, ensuring that diagnostic and maintenance decisions are based on the current true health status of the bearing housing, forming a closed-loop monitoring and continuous optimization mechanism.

[0134] Understandably, the significance of step S50 lies in: performing graded maintenance under the dual constraints of bearing housing deterioration and furnace condition window, so that the maintenance strategy can match the actual operating conditions and equipment health status, thereby avoiding the maladaptation problems caused by a uniform maintenance strategy. At the same time, by re-collecting the mechanical response chain after maintenance and comparing the changes in path deviation before and after maintenance and in historical stable states, the maintenance effect can be quantitatively verified. Furthermore, by updating the furnace condition constraint boundary and path deviation judgment rules, the system can have continuous self-learning and adaptive adjustment capabilities, thereby forming a closed-loop optimization mechanism of "monitoring-diagnosis-maintenance-recalibration".

[0135] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-parameter integrated diagnosis and maintenance method for the health status of blast furnace opening machine bearing housings, characterized in that: Includes the following steps: The blast furnace tapping machine bearing seat opening operation is divided into drive loading, slag impact, propulsion bearing and return unloading stages. Load, vibration and displacement signals are acquired synchronously in each stage according to the same force direction and sampling time window. Based on the force transmission direction, the multi-source signals are rearranged to construct a mechanical response chain. Based on the slag impact intensity, impact point deviation, tapping rhythm and opening frequency, the operation process is divided into high impact, disturbance and low load recovery furnace condition windows, and the allowable load amplitude and attenuation characteristic boundary of the bearing seat are limited in each furnace condition window. The current mechanical response chain is aligned with the historical stable mechanical response chain under the corresponding furnace condition window by mapping the force transmission path consistency. The response segment is traced in reverse along the impact peak path, load transmission path and unloading attenuation path, and abnormal response segments are identified according to the degree of path deviation and characteristic out-of-bounds conditions. Chain tracking is performed on the abnormal response segment. The impact peak increment, unloading duration and residual vibration retention amplitude are collected according to window constraints. The cumulative path deviation is used as the degradation criterion to construct an irreversible degradation path and classify the bearing housing degradation state according to the cumulative offset level. Under the constraints of bearing housing deterioration state and furnace condition window type, graded maintenance is performed, and after maintenance, mechanical response chains are collected according to the corresponding window constraint conditions to update furnace condition constraint boundaries and path deviation judgment rules.

2. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of constructing a mechanical response chain is as follows: The load signals, vibration acceleration signals, and displacement signals collected at each stage are processed to perform unified timestamp alignment. A principal force coordinate system is established based on the force direction of the bearing housing structure, and the signals from different measuring points are reprojected directionally according to the force transmission path. The reprojected multi-source signal is segmented by a sliding window, and the key response segments are enhanced and calibrated by combining the impact trigger point to form a staged response unit. Based on the force transmission sequence and structural connection topology between the phased response units, a multi-source coupled mechanical response structure representing the force transmission relationship is constructed. Normalization and noise suppression are performed on the multi-source coupled mechanical response structure, and a force transmission weighting coefficient is introduced for weighted sorting to generate a mechanical response chain that characterizes the entire opening and closing process.

3. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of defining the allowable load amplitude and attenuation characteristic boundary of the bearing housing within each furnace condition window is as follows: Energy equivalent conversion is performed on the impact intensity of slag, and impact space offset parameters are constructed by combining the impact point offset distance; Based on the iron tapping rhythm cycle and the opening frequency, the operating rhythm characteristics are extracted, and the periodic disturbance components are identified by frequency domain decomposition. Multidimensional correlation analysis was performed on impact energy parameters, spatial offset parameters, and rhythmic characteristics to construct a furnace condition distribution characterization space. Based on the density aggregation characteristics and fluctuation separation degree in the furnace condition distribution characterization space, the high impact window, disturbance window and low load recovery window are divided; Load envelope constraint curves are constructed within each window, and the upper limit of allowable load amplitude and the boundary of vibration decay time constant are obtained by combining historical stable operation sample statistics.

4. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of aligning the current mechanical response chain with the historical stable mechanical response chain under the corresponding furnace condition window through a consistent force transmission path mapping is as follows: Standard force transmission path templates are extracted from historical stable mechanical response chains to construct a path structure library containing impact peak nodes, transmission relay nodes, and unloading termination nodes; The current mechanical response chain is divided into corresponding path sub-segments according to the same stage, and the timing alignment is performed using a dynamic time warping method; Based on the force transmission direction consistency constraint, a standard path template corresponding to the current furnace condition window is retrieved from the path structure library, and node-level matching is performed between the current path sub-segment and the standard path template to calculate the path structure similarity. By introducing structural connection topology constraints, inconsistent nodes are remapped and corrected to generate path deviation vectors. Based on the path deviation vector and window constraints, determine the path correspondence and the degree of path deviation, and output the force transmission path consistency alignment result.

5. The method for multi-parameter fusion diagnosis and maintenance of the health status of the blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of tracing the response segment in reverse along the peak impact path, load transfer path, and unloading attenuation path is as follows: Based on the consistency alignment results of the force transmission path, the correspondence between the impact peak node, load transmission node and unloading attenuation node is determined. Starting from the peak impact node, the force transmission source node is traced along the load gradient direction to form the peak impact path. Based on the correspondence between load transfer nodes, the coordinated change characteristics of vibration response and displacement response are analyzed, the energy conduction direction during load transfer is traced, and the load transfer path is formed. Based on the vibration attenuation amplitude and displacement recovery characteristics during the unloading phase, the residual vibration diffusion process and energy release source are traced in reverse to form the unloading attenuation path. By associating the peak impact path, load transfer path, and unloading attenuation path, analyzing the load, vibration, and displacement response characteristics of each node, constructing the force transfer relationship between nodes, and generating inverse response results for path deviation determination.

6. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of identifying abnormal response segments based on the degree of path deviation and characteristic out-of-bounds conditions is as follows: Based on the inverse response results, the impact peak path, load transfer path and unloading attenuation path are compared with the corresponding paths in the historical stable mechanical response chain, and the transfer offset and response deviation of each path node are calculated. Based on the allowable load amplitude boundary and vibration attenuation characteristic boundary corresponding to the current furnace condition window, the load response, vibration response and displacement response of each path node are constrained and verified to identify abnormal feature points that exceed the window constraint range. When the deviation of a path node exceeds the allowable offset range of the corresponding path, and abnormal feature points continue to appear on the same force transmission path, an abnormal response is determined to exist. The abnormal propagation trajectory is traced along the mechanical response chain corresponding to the abnormal path to determine the abnormal extension section, and the abnormal extension section is marked as the abnormal response section.

7. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of constructing an irreversible degradation path is as follows: Within each furnace condition window, the abnormal response segment is traced along the mechanical response chain to extract the cumulative node offset and vibration energy along the impact peak path, load transfer path and unloading attenuation path. The attenuation and retention amplitude and duration of each path node during the statistical unloading phase are combined with the cumulative path deviation results to form the path cumulative feature vector for each window. The cumulative feature vectors of each window path are accumulated and fused in chronological order to obtain the cumulative path deviation index corresponding to each force transmission path. Identifying irreversible evolutionary segments based on the monotonic growth trend of cumulative path deviation indicators; The irreversible evolution segment is correlated and traced along the mechanical response chain to determine the continuous expansion range of the abnormal response segment, and the corresponding path is identified as the irreversible degradation path.

8. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of classifying bearing housing deterioration status according to cumulative offset level is as follows: The global offset score is calculated based on the irreversible degradation path, and a comprehensive degradation index is constructed by combining the impact frequency and vibration energy density. Segmented cluster analysis was performed on the comprehensive degradation index to identify the boundary intervals between low degradation, moderate degradation, and severe degradation. A window constraint correction factor is introduced to normalize and adjust the degradation index under different furnace condition windows; Based on the adjusted cumulative offset level, the bearing housing status is divided into three categories: stable, early warning, and unstable. The bearing housing deterioration status is output with furnace condition labels and stage weights.

9. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of performing graded maintenance under the constraints of bearing housing deterioration state and furnace condition window type is as follows: Based on the deterioration state of the bearing housing and the current furnace condition window type, determine the corresponding maintenance intervention conditions; When the bearing housing is in a stable state, perform lubrication compensation and inspection and maintenance of the connection parts within the low load recovery window; When the bearing housing is in a warning state, local reinforcement, clearance adjustment, or replacement and maintenance of worn parts are carried out on the corresponding stress parts that deviate from the path within the disturbance window. When the bearing housing is in an unstable state, the operation should be restricted under the high impact window, and the critical load-bearing parts corresponding to the irreversible deterioration path should be shut down for inspection and maintenance. After maintenance, the mechanical response chain under the corresponding furnace condition window was re-acquired, and the load transfer characteristics, vibration attenuation characteristics, and displacement recovery characteristics were compared and verified. The maintenance effect is confirmed based on the verification results, and a graded maintenance result corresponding to the deterioration state of the bearing housing is formed.

10. The method for multi-parameter fusion diagnosis and maintenance of the health status of blast furnace opening machine bearing housing according to claim 1, characterized in that, The process of updating furnace condition constraint boundaries and path deviation judgment rules is as follows: After completing the corresponding maintenance work according to the graded maintenance results, load, vibration and displacement signals are re-acquired in the corresponding furnace condition window, and the mechanical response chain after maintenance is reconstructed according to the force transmission direction. The path consistency of the maintained mechanical response chain is compared with that of the mechanical response chain before maintenance and the historical stable mechanical response chain. The recovery deviations corresponding to the impact peak path, load transfer path and unloading attenuation path are calculated. Based on the load amplitude distribution and vibration attenuation characteristic distribution under the corresponding furnace condition window according to the recovery deviation statistics, update the load amplitude boundary and attenuation characteristic boundary corresponding to each furnace condition window. Based on the changes in path deviation and abnormal response segment in the mechanical response chain after maintenance, the deviation judgment thresholds corresponding to each force transmission path are adjusted. The updated furnace condition window constraint boundaries and path deviation judgment thresholds are written back to the mechanical response chain diagnosis process to form continuously iteratively updated furnace condition window constraint boundaries and path deviation judgment rules.

Images