A steel-based ceramic lining plate wear state online monitoring and early warning method

Abstract

The application provides a steel-based ceramic lining plate wear state online monitoring and early warning method. The method is used for a steel-based ceramic lining plate which is composed of a metal matrix and embedded ceramic particles. The method comprises the following steps: collecting multi-source monitoring signals, wherein the multi-source monitoring signals at least comprise distance signals, acoustic emission signals and equipment operation condition parameters; based on the multi-source monitoring signals, a wear analysis model with embedded physical constraints is used for processing to output decoupling evaluation results; the physical constraints are used for representing the coupling relationship between the metal matrix wear state and the protection effect of the ceramic particles; based on the evaluation results, the evolution trend of the dual-phase wear coupling of the lining plate is predicted, and the chain failure risk triggered by the failure of the ceramic particles and the accelerated wear of the metal matrix is identified; the graded early warning signals are obtained by comprehensively considering the grade of the chain failure risk and the evolution trend, and the maintenance decision suggestions are generated. The method provided by the application realizes the real-time online decoupling evaluation of the dual-phase wear, the prediction of the chain risk and the graded early warning.

Description

A method for online monitoring and early warning of wear condition of steel-based ceramic liners Technical Field

[0001] This disclosure generally relates to the field of material condition monitoring technology, and specifically to a method for online monitoring and early warning of wear condition of steel-based ceramic liners. Background Technology

[0002] With the deepening of intelligent mining construction, higher demands are being placed on the real-time perception and predictive maintenance of the operating status of key equipment. As the core equipment for underground coal transportation, the wear resistance and service life of the central trough, a core component of the scraper conveyor, directly affect the transportation efficiency, downtime maintenance costs, and safety level of the entire coal face. In recent years, steel-based ceramic composite materials, due to their superior wear resistance and toughness, have gradually replaced traditional wear-resistant steel plates in central trough liners, significantly extending the replacement cycle. However, despite the improvement in material wear resistance, its essential nature as a vulnerable part remains unchanged. Timely and accurate judgment of its wear state remains a key technical bottleneck in achieving the transformation from "periodic maintenance" to "predictive maintenance."

[0003] Currently, wear monitoring of the trough liners in scraper conveyors mainly relies on traditional manual inspection methods involving periodic shutdowns. Specifically, in the aforementioned work scenario, existing technology has the following significant drawbacks: First, it requires production to be interrupted, making real-time online monitoring during operation impossible, severely impacting production efficiency and resulting in a delay in inspection timing. Second, manual inspection relies on experience-based judgment; visual inspection or simple tool measurements make it difficult to accurately quantify and assess the wear on the liner surface, especially in areas covered by the chain and scraper, and it cannot capture the rapid development stages of wear. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide an online monitoring and early warning method for the wear condition of steel-based ceramic liners to solve the above problems.

[0005] This application provides a method for online monitoring and early warning of wear condition of steel-based ceramic liners, specifically for steel-based ceramic liners composed of a metal matrix and embedded ceramic particles. The method includes:

[0006] Synchronously online acquisition of multi-source monitoring signals, including at least distance signals reflecting the thickness of the liner, acoustic emission signals reflecting material failure, and equipment operating condition parameters;

[0007] Based on the multi-source monitoring signals, the data are processed by a wear analysis model with embedded physical constraints to output decoupled evaluation results; the physical constraints are used to characterize the coupling relationship between the wear state of the metal matrix and the protective effect of ceramic particles.

[0008] Based on the evaluation results, the evolution trend of the two-phase wear coupling of the liner is predicted, and the risk of cascading failure triggered by ceramic particle failure leading to accelerated wear of the metal matrix is ​​identified.

[0009] By combining the level of the cascading failure risk with the evolution trend, a graded early warning signal is obtained, and maintenance decision recommendations are generated based on the graded early warning signal.

[0010] According to the technical solution provided in the embodiments of this application, the acquisition of the acoustic emission signal used to capture material failure includes:

[0011] The raw acoustic emission signal is collected and matched with a preset feature spectrum library for identification; the feature spectrum library includes at least signal features characterizing three types of failure modes: microcracks in ceramic particles, single particle detachment, and mass detachment.

[0012] Based on the matching results, failure event information is generated as the acoustic emission signal, and the failure event information includes at least one failure mode and corresponding spatial distribution information.

[0013] According to the technical solution provided in the embodiments of this application, before processing the multi-source monitoring signals through a wear analysis model embedded with physical constraints to output decoupled evaluation results, the method further includes:

[0014] A pulsed thermal excitation was applied to the surface of the liner, and the infrared temperature field sequence of its surface was acquired after the excitation ended;

[0015] The changes of the infrared temperature field sequence over time are analyzed, and parameter maps reflecting the thermal property distribution at different locations on the liner surface are obtained through inversion calculation. These parameter maps are used to indirectly characterize the coverage density and distribution uniformity of ceramic particles on the liner surface.

[0016] Based on the parameter diagram, the failure mode and / or spatial distribution information of the failure event information are verified and corrected.

[0017] According to the technical solution provided in the embodiments of this application, the wear analysis model with embedded physical constraints is a physical information neural network model, and the physical constraints include coupling the wear rate of the metal matrix with the protection coefficient of the ceramic particles; the decoupling evaluation results include the wear rate of the metal matrix of the liner, the protection coefficient of the ceramic particles, and the corresponding spatial distribution information.

[0018] According to the technical solution provided in the embodiments of this application, predicting the two-phase wear coupling evolution trend of the liner based on the decoupling evaluation results includes:

[0019] A two-phase wear state-space model is constructed with the ceramic particle protection coefficient as the core state variable;

[0020] The current state and historical trend of the wear rate of the metal matrix and the protection coefficient of the ceramic particles, as well as the operating parameters of the equipment, are used as inputs to the state space model for iterative calculation, and the transmission process is simulated through the physical constraints.

[0021] Output the evolution map of the liner over a future set time period. The evolution map includes the predicted spatial distribution and evolution sequence of the wear depth of the metal matrix and the protection coefficient of the ceramic particles at different time points.

[0022] According to the technical solution provided in the embodiments of this application, the identification of the cascading failure risk triggered by the failure of ceramic particles leading to accelerated wear of the metal matrix includes:

[0023] Based on the failure event information, the spatial clustering degree of individual and group detachment failure modes of ceramic particles in the evaluation area is calculated in real time.

[0024] When the spatial aggregation degree is greater than the first threshold, and the rate of decrease of the ceramic particle protection coefficient in the region is greater than the second threshold as identified from the evolution map, the region is determined to have entered a cascading failure state.

[0025] Based on the determination of the chain failure state, the ceramic particle protection coefficient in the region is simulated to decrease by a corresponding magnitude in the state space model, and the chain reaction process of accelerated wear of the metal matrix and the associated decrease of the ceramic protection coefficient in adjacent regions is iteratively calculated.

[0026] Based on the simulation results of the chain reaction process, the spatial expansion rate of the chain reaction process is extracted, and the wear increment of the metal matrix caused by the process is calculated; based on the spatial expansion rate and the wear increment, the chain failure risk index is calculated.

[0027] According to the technical solution provided in the embodiments of this application, the step of calculating the cascading failure risk index based on the spatial expansion rate and the wear increment includes:

[0028] The spatial expansion rate and the wear increment are normalized.

[0029] The normalized spatial expansion rate and wear increment are weighted and summed according to preset weighting coefficients to obtain the cascading failure risk index; wherein the weighting coefficient of the spatial expansion rate is greater than the weighting coefficient of the wear increment.

[0030] According to the technical solution provided in the embodiments of this application, obtaining a graded early warning signal by comprehensively considering the level of the cascading failure risk and the evolution trend includes:

[0031] The cascading failure risk index is compared with multiple preset risk thresholds to determine the cascading failure risk level;

[0032] Based on the evolution map, the future time point when the predicted wear depth of the metal matrix first reaches the preset thickness threshold is determined, and the geometric loss level is determined according to the proximity of this time point to the current time; and the future time point when the ceramic particle protection coefficient first falls below the preset functional threshold is determined, and the functional degradation level is determined according to the proximity of this time point to the current time.

[0033] Based on the preset fusion decision rules, the cascading failure risk level, geometric loss level, and functional degradation level are fused to generate the graded early warning signal;

[0034] The fusion decision rules include:

[0035] If the risk level of the cascading failure reaches the highest level, then the highest level warning signal will be generated directly.

[0036] Otherwise, based on the current combination of the geometric loss level and the functional degradation level, a preset biphasic risk decision matrix is ​​queried, and a corresponding early warning signal is generated.

[0037] According to the technical solution provided in the embodiments of this application, the step of generating maintenance decision suggestions based on the graded early warning signals includes:

[0038] Obtain production plan information for a future preset period;

[0039] The graded early warning signals, evolution maps, and production plan information are matched and analyzed to dynamically recommend maintenance time windows and corresponding maintenance methods;

[0040] The maintenance methods include at least one of partial repair, overall replacement, or adjustment of operating parameters.

[0041] According to the technical solution provided in the embodiments of this application, the distance signal is acquired by a non-contact laser ranging sensor array. The sensor array is fixedly installed on the outside of the equipment housing on the back of the liner and is measured through an observation window.

[0042] Compared with existing technologies, the advantages of this application are as follows: By synchronously acquiring multi-source signals such as distance, acoustic emission, and operating conditions online, the wear state of steel-based ceramic liners is perceived online in real time, fundamentally avoiding the drawbacks of traditional methods that require shutdown for inspection, thus ensuring production continuity. Furthermore, by processing the signals through a wear analysis model embedded with physical constraints, the independent wear states and coupling relationships of the metal matrix and ceramic particles can be decoupled and quantitatively evaluated, overcoming the inaccuracy of manual experience judgment and achieving in-depth analysis of the failure mechanism of composite materials. Based on this, the method can proactively predict the evolution trend of two-phase wear coupling and identify the risk of cascading failures, thereby capturing the rapid development stage of wear and systemic risks. Finally, by comprehensively considering risk levels and evolution trends to generate graded early warnings and maintenance decisions, the equipment maintenance mode is transformed from delayed periodic maintenance to precise predictive maintenance, significantly improving maintenance efficiency, equipment reliability, and safety production levels. Attached Figure Description

[0043] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0044] Figure 1 is a flowchart of the steps of the online monitoring and early warning method for the wear status of steel-based ceramic liners provided in Example 1. Detailed Implementation

[0045] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0046] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] Example 1

[0048] Please refer to Figure 1. This embodiment provides an online monitoring and early warning method for the wear condition of steel-based ceramic liners. It is applicable to steel-based ceramic liners composed of a metal matrix and embedded ceramic particles. This type of liner is widely used in easily worn equipment scenarios such as mining machinery chutes, crusher inner walls, and material conveying pipelines.

[0049] The method includes:

[0050] S100: Synchronously acquire multi-source monitoring signals online, including at least distance signals reflecting the thickness of the liner, acoustic emission signals reflecting material failure, and equipment operating condition parameters.

[0051] Specifically, in step S1, a real-time synchronous online acquisition of multi-source monitoring signals is performed. The acquired multi-source monitoring signals include three types of core data. The distance signal reflects the geometric thickness change of the liner plate and is acquired through a non-contact laser rangefinder array. The sensor array is fixedly installed on the outside of the equipment housing on the back of the liner plate, and measurements are taken through an observation window. The sensor array is uniformly distributed along the surface of the liner plate; optionally, one sensor is placed every 5 centimeters, and the acquisition frequency is set to once every 10 minutes to ensure real-time capture of thickness changes in various areas of the liner plate. The acoustic emission signal is used to capture microscopic failures within the material. Acoustic emission sensors are installed at the edges of the liner plate and key stress points to acquire raw acoustic emission signals online. During acquisition, the sensors are kept in close contact with the liner plate surface to avoid signal attenuation. The acquired raw acoustic emission signals need to be processed to obtain acoustic emission signals for subsequent steps. Equipment operating parameters are acquired in real time through the equipment control system, specifically including equipment operating load, material flow rate, material hardness, operating time, and number of start-stop cycles. All monitoring signals are marked with a unified timestamp to ensure complete synchronization of the three types of signals in the time dimension, providing a consistent data foundation for subsequent analysis.

[0052] Further, in step S100, acquiring the acoustic emission signal used to capture material failure specifically includes:

[0053] The raw acoustic emission signal is collected and matched with a preset feature spectrum library for identification; the feature spectrum library includes at least signal features characterizing three types of failure modes: microcracks in ceramic particles, single particle detachment, and mass detachment.

[0054] Based on the matching results, failure event information is generated as the acoustic emission signal, and the failure event information includes at least one failure mode and corresponding spatial distribution information.

[0055] Specifically, in step S100, the raw acoustic emission signal is first acquired online. Acoustic emission sensors are placed in key stress areas such as the edges and middle of the liner, as well as in vulnerable parts prone to failure. The number of sensors is determined according to the liner size to ensure signal acquisition covers the entire surface of the liner. The sensor sampling frequency is set to 1MHz, and the sampling duration is consistent with the acquisition cycle of distance signals and operating parameters, with each acquisition unit consisting of 10 minutes. The raw acoustic emission signal within each acquisition unit is recorded synchronously. During the acquisition process, the raw signal is initially amplified and filtered by a signal conditioner to remove interference signals such as environmental vibration and equipment operating noise, ensuring the purity of the raw signal.

[0056] The acoustic emission characteristic spectrum library for ceramic particle failure is a standard database pre-installed in the monitoring system. This library was constructed and pre-stored through laboratory simulation experiments before the monitoring work commenced. During construction, acoustic emission signals were collected for three typical failure modes of ceramic particles in steel-based ceramic liners, with at least 50 sets of repeated tests performed for each failure mode to ensure data reliability. Feature extraction was performed on the collected test signals, focusing on key parameters such as peak amplitude, signal rise time, duration, signal energy, and frequency distribution. The acoustic emission signals corresponding to microcracks showed lower peak amplitude, shorter duration, and energy concentrated in the high-frequency band. Individual detachment signals had higher peak amplitude and energy than microcracks, with a moderate rise time. Group detachment signals exhibited large peak amplitude, strong energy, long duration, and a wide frequency distribution. Statistical analysis of these characteristic parameters determined the value range and typical feature combinations for each characteristic parameter under each failure mode, forming a complete acoustic emission characteristic spectrum library for ceramic particle failure, serving as the standard basis for subsequent matching and identification.

[0057] Next, the preprocessed acoustic emission raw signal is split into acquisition units, and characteristic parameters such as peak amplitude, rise time, energy, and frequency distribution of the signal in each unit are extracted and compared one by one with the typical characteristic parameters of three types of failure modes in a preset feature spectral library. A similarity calculation method is used to quantify the degree of fit between the raw signal features and the spectral library features. A similarity threshold of 85% is set. When the similarity between a certain raw signal feature and a certain type of failure mode in the spectral library is higher than the threshold, the raw signal is determined to correspond to that type of failure mode. If the similarity is lower than the threshold, it is determined to be an invalid signal and is not included in subsequent analysis. If the feature similarity requirements of two or more types of failure modes are met simultaneously, the matching result of the energy feature is given priority to ensure the uniqueness and accuracy of the identification results.

[0058] Finally, failure event information is generated based on the matching results, serving as a crucial component of the multi-source monitoring signal. The failure event information must clearly label the identified failure mode. If the matching result indicates a microcrack, it is recorded as a microcrack failure; if a single ceramic particle detaches from the liner surface, it is recorded as a single detachment failure; if multiple ceramic particles detach in a concentrated manner, it is recorded as a group detachment failure. Simultaneously, the spatial distribution information of the failure events is determined. Combined with the positions of the deployed acoustic emission sensor array, the location is calculated using the time difference of signal arrival at different sensors to accurately pinpoint the specific area where the failure event occurred. The failure modes are correlated and integrated with the corresponding spatial distribution information to form complete failure event information, which is stored synchronously with distance signals and equipment operating parameters, providing accurate acoustic emission data support for subsequent model processing and condition assessment.

[0059] S200: Based on the multi-source monitoring signals, the wear analysis model with embedded physical constraints is used to process the signals and output the decoupled evaluation results; the physical constraints are used to characterize the coupling relationship between the wear state of the metal matrix and the protective effect of ceramic particles.

[0060] Specifically, in step S200, after the acquisition of multi-source monitoring signals, these signals are processed using a wear analysis model embedded with physical constraints. The physical constraints embedded in this model characterize the coupling relationship between the wear state of the metal matrix and the protective effect of the ceramic particles. This coupling relationship is determined based on the physical mechanism of liner wear, ensuring that the model analysis conforms to the wear law under actual working conditions. Before model processing, the acquired multi-source monitoring signals need to be preprocessed, including removing abnormal fluctuation data from the distance signal, filtering and denoising the raw acoustic emission signal, and normalizing the working parameters. The preprocessed signals are used as model input data. During model operation, through the fusion analysis of multi-source signals, the decoupled evaluation of the wear state of the metal matrix and ceramic particles is achieved. The final output decoupled evaluation results include quantitative evaluation information on the equivalent wear state of the metal matrix and the effective protective state of the ceramic particles, while clarifying the spatial distribution characteristics of the two. By matching the installation position of the sensor array, the wear state differences in each region of the liner are accurately located.

[0061] Further, in step S200, the wear analysis model embedded with physical constraints is a physical information neural network model, and the physical constraints include coupling the wear rate of the metal matrix with the ceramic particle protection coefficient; the decoupling evaluation results include the wear rate of the metal matrix of the liner, the ceramic particle protection coefficient, and the corresponding spatial distribution information.

[0062] Specifically, the wear analysis model provided in this embodiment employs a physical information neural network architecture. This model is not purely data-driven; instead, it embeds prior physical knowledge describing the wear process of steel-based ceramic composite materials into the structure and training process of the neural network in the form of mathematical equations (i.e., physical constraints), thereby ensuring that its output not only fits the data but also conforms to physical laws. This physical information neural network model comprises two core parts: a deep neural network backbone for feature extraction and fitting, and a physical constraint layer serving as physical rule constraints. The model uses the preprocessed multi-source monitoring signals from step S100 as input.

[0063] The key to this embodiment lies in the physical constraint layer. This layer implements a coupled equation based on the material wear mechanism to constrain the dynamic relationship between the wear rate of the metal matrix and the protection coefficient of the ceramic particles. Specifically, this physical constraint can be expressed by the following formula:

[0064] ,

[0065] In the formula, Indicates time Liner surface position The wear rate of the metal substrate at that location is a spatiotemporal variable. Represents a set of parameters under specific operating conditions. The theoretical wear rate of a metal matrix under conditions such as load, speed, and abrasive characteristics is obtained by mapping the working condition parameters through a branch of a neural network, and it characterizes the intrinsic wear rate without ceramic protection. Indicates time ,Location The ceramic particle protection coefficient at the location, its value range is ,in This indicates that the ceramic particles are intact, providing complete protection. This indicates that the ceramic's protective function has been completely lost. Represents the coupling strength coefficient It quantifies the maximum possible contribution of the protective effect of ceramic particles to reducing the wear rate of the matrix in practice by calibrating historical data. Represents nonlinear exponent In this embodiment, the value is 2 or 3. This index physically characterizes the decrease in the ceramic protection coefficient (i.e., The nonlinear relationship between the accelerated increase in the wear rate of the metal matrix when the wear rate decreases (i.e., when the wear rate decreases). This represents a small perturbation term used to characterize unmodeled dynamic factors and is determined through training.

[0066] During model training and inference, the aforementioned physical constraint equations act as a strong constraint, working in conjunction with the data fitting ability of the neural network. The model searches for an optimal set of... and The spatiotemporal distribution of these components allows them to both fit the input monitoring signals well and strictly satisfy the coupling laws described by the aforementioned physical equations. This design enables the model to decouple and quantify the independent wear states of the metal matrix and ceramic particles from mixed multi-source signals, rather than simply providing a general overall wear amount. This achieves a deep analysis of the failure mechanism of composite materials, fundamentally overcoming the shortcomings of traditional monitoring methods that can only assess macroscopic results and cannot perceive the contributions of internal phases. Furthermore, by constraining the nonlinear coupling relationship between the two, the model possesses the inherent ability to simulate the key physical process of accelerated matrix wear caused by ceramic protection failure, laying a solid theoretical foundation for accurately predicting evolution trends and identifying cascading risks in subsequent steps.

[0067] S300: Based on the evaluation results, predict the evolution trend of the two-phase wear coupling of the liner and identify the risk of cascading failures triggered by ceramic particle failure that accelerates wear of the metal matrix.

[0068] Specifically, in step S300, during the prediction process, the coupling relationship between the metal matrix and ceramic particles in the physical constraints is combined to analyze the correlation law of their wear states, clarify the changing trend of two-phase wear under different working conditions, and form a judgment on the wear evolution trend of the entire liner and each key area. The identification of cascading failure risk focuses on the correlation effect between ceramic particle failure and accelerated wear of the metal matrix. By monitoring the attenuation of the effective protection state of ceramic particles, the degree of influence on the equivalent wear state of the metal matrix is ​​judged. When the effective protection capability of ceramic particles decreases significantly and may cause an abnormal increase in the wear rate of the metal matrix, a cascading failure risk is determined.

[0069] Further, in step S300, based on the decoupling evaluation results, the two-phase wear coupling evolution trend of the liner is predicted, specifically including:

[0070] S310: Construct a two-phase wear state-space model with the ceramic particle protection coefficient as the core state variable.

[0071] Specifically, the purpose of step S310 is to quickly obtain a digital twin of the wear evolution that highly matches the current state of the liner. The system pre-stores a general two-way wear extrapolation model framework. This framework is based on the general wear physics mechanism of steel-based ceramic composites, and its form includes the physical constraints described above.

[0072] After obtaining the decoupling evaluation results at the current moment, the system uses these results to configure key parameters and initialize the state of the general model: the spatial distribution of the ceramic particle protection coefficient measured at the current moment is then used. The initial field, serving as the core state variable, is directly embedded into the model. This ensures that the starting point of the model simulation is completely consistent with the current true state of the ceramic layer on the liner. The wear rate distribution of the metal substrate at the current moment is utilized. and The correspondence between the coupling strength coefficients in the model. or nonlinear exponent Fine-tuning or validation is performed to ensure the model's output at the initial moment best matches the current observation data. After completing the above configuration, the general model is instantiated into a personalized two-phase wear state-space model specifically designed to simulate the future wear evolution of this particular liner from now on.

[0073] S320: The current state and historical trend of the wear rate of the metal matrix and the protection coefficient of the ceramic particles, as well as the operating parameters of the equipment, are used as inputs to the state space model for iterative calculation, and the transmission process is simulated through the physical constraints.

[0074] Specifically, step S320 is the process of performing personalized predictive simulation. First, the initialized state-space model is placed in the current state. Simultaneously, the expected equipment operating condition parameter sequences for the current and future prediction periods are set. As an external driving condition input model.

[0075] Subsequently, the model uses discrete time steps. Perform iterative calculations. At each time step:

[0076] The model is based on the current Distribution and operating conditions The wear rate distribution of the metal matrix at that moment is calculated and updated through the physical constraints embedded within it. Meanwhile, the model is updated based on the Operating conditions and current status The spatial distribution of the protection coefficient is calculated and updated in the next time step. This process continues in a cycle, with physical constraints ensuring that each step of the evolution follows the basic laws of material wear, while the spatial term calculation simulates the transmission process of wear and failure effects on the liner surface.

[0077] S330: Output the evolution map of the liner over a future set time period. The evolution map includes the predicted spatial distribution and evolution sequence of the wear depth of the metal matrix and the protection coefficient of the ceramic particles at different time points.

[0078] Specifically, in step S330, the model runs to a preset future time endpoint. Then, a dynamic coupling evolution map is output. This map is a collection of time slices (e.g., every 8 hours) over a future period of time in the time dimension; and covers the entire liner surface in the spatial dimension.

[0079] For each predicted time point Corresponding to the wear depth of the metal substrate and ceramic particle protection coefficient . By analyzing the results generated in iterative calculations The sequence is obtained by time integration, which visually displays the predicted thickness loss; This is the distribution of the protection coefficient predicted at that moment, directly output by the model.

[0080] The evolutionary map is essentially a sequence of the aforementioned spatiotemporal distribution arranged chronologically. It clearly reveals a complete and dynamic future picture of how the ceramic layer in each region of the liner decays and how the substrate wears down under given operating conditions, starting from the current state. For example, the map can predict that the ceramic protection coefficient in region A will drop below 0.5 in 3 days, triggering accelerated substrate wear in that region, causing it to reach the maintenance threshold in 5 days. This high-fidelity predictive map, generated by a model with personalized configuration based on the current state, is the core basis for this solution to achieve accurate early warning and forward-looking decision-making.

[0081] Further, in step S300, the risk of a cascading failure triggered by the failure of ceramic particles leading to accelerated wear of the metal matrix is ​​identified, including:

[0082] S340: Based on the failure event information, calculate in real time the spatial aggregation degree of individual and group detachment failure modes of ceramic particles within the evaluation area.

[0083] Specifically, in step S340, the system receives and processes the generated failure event information, which records the failure mode (including ceramic particle microcracks, single particle detachment, and group detachment) and the location coordinates of each failure event. To achieve specialized analysis of cascading failure risks, the system first filters out two types of events from this event information: single ceramic particle detachment and group ceramic particle detachment. These two types of events directly represent the physical loss of the protective ceramic layer and are direct causes that may trigger a chain reaction. Next, the system divides the liner surface into multiple evaluation areas (e.g., a grid with a side length of 10 cm) and, for each evaluation area, calculates the spatial clustering of the two types of failure events within a set sliding time window (e.g., the most recent hour).

[0084] Clustering is a quantitative indicator, and its calculation can involve the following steps: First, count the total number of the two types of events occurring within the time window, and assign higher weight to group detachment events to reflect their more severe initial impact; then, calculate the weighted event areal density (number of events / area); finally, compare this areal density with the background random event rate obtained based on long-term historical data, and generate a standardized spatial clustering index through ratio or difference operations. The higher the index, the more abnormally concentrated the failure events are in the spatial distribution of the area, exceeding the random fluctuation range under normal operation, strongly suggesting that the ceramic layer in this local area may have structural weakening or is undergoing intensive damage.

[0085] S350: When the spatial aggregation degree is greater than the first threshold, and the rate of decrease of the ceramic particle protection coefficient in the region is greater than the second threshold as identified from the evolution map, it is determined that the region has entered a cascading failure state.

[0086] Specifically, in step S350, when the system determines that the spatial clustering degree of a certain evaluation area exceeds a preset first threshold, it considers that the area has entered a state of dense event occurrence that constitutes the physical basis of a chain reaction. The first threshold is determined through offline simulation analysis and calibration with a large amount of field data, characterizing the spatial clustering of failure events to a critical level sufficient to significantly weaken the structural integrity of the local area. However, the static or quasi-static characteristic of event clustering alone is insufficient to confirm the existence of risk, so the system introduces a second key and dynamic judgment condition.

[0087] The system accesses the dynamic coupling evolution map generated in step S330, extracts the predicted curve of the ceramic particle protection coefficient corresponding to the evaluation area, and calculates its current instantaneous decline rate (i.e., the negative value of the time derivative). If this decline rate simultaneously exceeds a preset second threshold, it indicates that the ceramic protection function in this area is being lost at an accelerated rate, rather than simply decaying slowly according to the baseline trend predicted by the model. The second threshold corresponds to the critical rate of change when the protection function enters the nonlinear accelerated failure stage. The system only determines that the area has entered a cascading failure state when both the spatial aggregation degree and the rate of decline of the protection coefficient exceed the limit are met simultaneously. This dual-judgment logic effectively avoids misjudgment based on a single indicator. Its technical meaning is that it confirms that the area is in a specific stage where ceramic particles are densely detaching and the detachment rate has caused its protection function to enter an accelerated collapse. This is a clear precursor that may trigger the positive feedback loop of "ceramic detachment - accelerated matrix wear - inducing more detachment".

[0088] S360: Based on the determination of the chain failure state, simulate the decrease in the ceramic particle protection coefficient of the region in the state space model, and iteratively calculate the chain reaction process of accelerated wear of the metal matrix and the associated decrease in the ceramic protection coefficient of adjacent regions.

[0089] Specifically, in step S360, after the risk status is determined, the system does not immediately issue an alarm, but instead performs a proactive worst-case scenario simulation based on digital simulation to accurately quantify the severity of potential chain reactions.

[0090] This test is conducted in the personalized two-phase wear state space model constructed and initialized to reflect the current true state of the liner in step S310. Based on the dominant failure mode (single detachment or group detachment) within the triggering region during the evaluation time window, the system queries a pre-established mapping relationship between failure modes and disturbance amplitudes, calibrated in the laboratory, to determine the simulated disturbance amplitude to be applied to the model. This mapping relationship is determined in the laboratory by simulating the actual impact of different failure modes on the ceramic particle protection coefficient of a local area. For example, the mapping rule stipulates that if single detachment is the dominant event within a region, a small instantaneous decrease in the ceramic particle protection coefficient of the model for that region should be applied. If the primary issue is group shedding, then apply a relatively large decrease value. Subsequently, in the state-space model, the initial value of the ceramic particle protection coefficient of the decision region is forcibly subtracted from the corresponding value. This was to simulate the impact of a hypothetical, additional severe detachment event on the area.

[0091] Next, the system uses this perturbed state as a new starting point and re-runs the model for iterative calculations. Because the protection coefficient in this region is artificially reduced, the wear rate of the metal matrix is ​​immediately recalculated and significantly increased based on the model's embedded physical constraints. The model further uses its built-in spatial transmission algorithm, which simulates mechanical interactions and stress redistribution, to calculate the impact of the accelerated matrix wear on the adhesion stability of ceramic particles in adjacent mesh units. This iteratively simulates a positive feedback loop process: "local ceramic protection failure → rapid wear of exposed metal matrix → wear pits altering local geometry and stress field → weakening the support of adjacent ceramic particles leading to their detachment," representing a complete chain reaction spatiotemporal propagation dynamic process.

[0092] S370: Based on the simulation results of the chain reaction process, extract the spatial expansion rate of the chain reaction process and calculate the wear increment of the metal matrix caused by the process; based on the spatial expansion rate and the wear increment, calculate the chain failure risk index.

[0093] Specifically, in step S370, the system performs precise quantitative analysis on the result data generated by the above simulation process to generate a comprehensive and comparable risk measurement index.

[0094] First, the spatial propagation characteristics of the chain reaction during the simulation process are analyzed, specifically by tracking the ceramic particle protection coefficient in the model below a certain critical danger value (e.g., The area of ​​a dangerous region is measured by its area change curve over time. The maximum instantaneous rate of expansion of the dangerous region is calculated, which is then quantified as the spatial expansion rate of the chain reaction. Its physical meaning reflects the urgency of the potential damage spread.

[0095] Secondly, by precisely comparing the predicted metal matrix wear depth distribution maps at the same future moment when the simulation terminates, under two scenarios—one involving a chain reaction simulation and the other involving no disturbance and only following the normal baseline trend—the depth difference between the two scenarios in the risk area and the surrounding influence zone is calculated. This difference is then integrated within this spatial range to obtain the additional metal matrix wear increment caused by the chain reaction process. This increment quantifies the severity of the additional material loss that would occur if the chain reaction actually happened.

[0096] Finally, the system obtains the cascading failure risk index according to the preset calculation method. This index is a single scalar value that comprehensively characterizes the urgency and potential severity of the risk of cascading failures in a specific region at a specific point in time, providing a direct and quantitative core input basis for generating objective and accurate graded early warning signals in subsequent steps.

[0097] Further, in step S370, based on the spatial expansion rate and the wear increment, the cascading failure risk index is calculated, including:

[0098] The spatial expansion rate and the wear increment are normalized.

[0099] The normalized spatial expansion rate and wear increment are weighted and summed according to preset weighting coefficients to obtain the cascading failure risk index; wherein the weighting coefficient of the spatial expansion rate is greater than the weighting coefficient of the wear increment.

[0100] Specifically, the system receives the chain reaction spatial expansion rate and the metal substrate wear increment from the simulation output of step S370. To enable comprehensive comparison, the system normalizes both values ​​according to a preset benchmark range, converting them into dimensionless scalars. After normalization, the system performs the core comprehensive calculation. This calculation follows preset rules: the normalized spatial expansion rate value and the wear increment value are weighted and summed according to their respective preset weighting coefficients to synthesize a single chain failure risk index. The weighting coefficient for the spatial expansion rate is set to be greater than that for the wear increment. This weighting allocation reflects the priority given to the urgency of dynamic risk propagation in this scheme.

[0101] The formula for calculating the weighted sum is: Risk Index = ( × Normalized space expansion rate) + ( × Normalized wear increment), where and Let be the weight coefficient, and satisfy... The specific values ​​of the weighting coefficients are determined through domain knowledge and historical data calibration. Finally, the system outputs the calculated cascading failure risk index.

[0102] S400: By combining the level of the cascading failure risk with the evolution trend, a graded early warning signal is obtained, and maintenance decision recommendations are generated based on the graded early warning signal.

[0103] Specifically, in step S400, different early warning levels are defined based on the severity and evolution trend of the cascading failure risk. Early warning signals are then pushed to relevant operation and maintenance management platforms and personnel through appropriate information transmission channels. Simultaneously, based on the wear evolution pattern of the liner plates and the equipment's operational needs, targeted maintenance decision-making suggestions are formulated to provide scientific operational guidance to operation and maintenance personnel, ensuring the timeliness and effectiveness of maintenance measures and guaranteeing the continuous safe operation of the equipment.

[0104] Further, in step S400, by combining the level of the cascading failure risk with the evolution trend, a graded early warning signal is obtained, including:

[0105] S410: Compare the cascading failure risk index with multiple preset risk thresholds to determine the cascading failure risk level.

[0106] Specifically, in step S410, the system internally presets three risk thresholds: a low-risk threshold, a medium-risk threshold, and a high-risk threshold. These thresholds are determined through a combination of historical failure case analysis and simulation verification, and are used to classify risks into four states: acceptable, requiring attention, requiring vigilance, and requiring emergency handling.

[0107] During execution, the system reads the cascading failure risk index of all currently calculated assessment regions. For each region, the following judgment logic is executed:

[0108] If the chain failure risk index is less than or equal to the low risk threshold, the chain failure risk level of the area is determined to be low, indicating that the risk of a chain reaction occurring in the area is extremely small, the ceramic particles detach at a sparse time and have not formed an accelerated failure trend.

[0109] If the cascading failure risk index is greater than the low risk threshold and less than or equal to the medium risk threshold, the cascading failure risk level of the area is determined to be medium, indicating that there are certain signs of clustering failure events or accelerated decline in the protection coefficient in the area, which need to be included in the observation scope, but have not yet reached the level of triggering an emergency response.

[0110] If the cascading failure risk index is greater than the medium risk threshold and less than or equal to the high risk threshold, the judgment level is high, indicating that the area has met or is close to the judgment conditions for cascading failure, there is a clear potential risk of cascading reaction, and contingency plans need to be prepared and the monitoring frequency increased.

[0111] If the cascading failure risk index is greater than the high-risk threshold, the level is determined to be emergency, indicating that the risk of cascading failure in the area is extremely high. Simulations show that once triggered, the destructive spread is rapid, and immediate intervention measures must be taken.

[0112] The system stores the identifier of each area in association with its risk level, serving as one of the key inputs for generating the final warning signal.

[0113] S420: Based on the evolution map, determine the future time point when the predicted wear depth of the metal matrix first reaches the preset thickness threshold, and determine the geometric loss level according to the proximity of this time point to the current time; and determine the future time point when the ceramic particle protection coefficient first falls below the preset functional threshold, and determine the functional degradation level according to the proximity of this time point to the current time.

[0114] Specifically, in step S420, based on the evolution map output in step S330, the geometric loss level and the functional degradation level are determined respectively, wherein the geometric loss level is determined.

[0115] For each geometric wear level, a preset maintenance threshold for the wear depth of the metal substrate is established. This threshold is determined based on the safety margin of the liner design, equipment operating requirements, and industry standards; for example, it can be set to 30% of the original thickness. The system traverses the predicted wear depth sequence for each evaluation region in the evolution map. For each location, it identifies the future time point when the predicted value first reaches or exceeds the maintenance threshold. The geometric wear level is then classified based on the time difference between the future time point and the current time. A preset time window threshold is also established, for example:

[0116] If the time difference is greater than 30 days, the level is normal;

[0117] If the time difference is greater than 7 days and less than or equal to 30 days, the level is "attention".

[0118] If the time difference is greater than 1 day and less than or equal to 7 days, the level is warning.

[0119] If the time difference is less than or equal to 1 day, the level is emergency.

[0120] The system considers the overall condition of the liner or key areas, and usually takes the level corresponding to the minimum time difference among all locations as the overall geometric loss level.

[0121] For each functional degradation level, a preset functional failure threshold for the ceramic particle protection coefficient is established. This threshold characterizes the critical point at which the ceramic particles lose their effective protective capability; for example, it can be set to 0.3. The system iterates through the predicted protection coefficient sequence for each evaluation region in the evolution map. For each location, it identifies the future time point at which the predicted value first falls below the functional failure threshold. Based on the time difference between the future time point and the current time, the functional degradation level is classified. The preset time window threshold can be similar to that for geometric loss, for example:

[0122] If the time difference is greater than 30 days, the level is normal;

[0123] If the time difference is greater than 15 days and less than or equal to 30 days, the level is "attention".

[0124] If the time difference is greater than 3 days and less than or equal to 15 days, the level is warning.

[0125] If the time difference is less than or equal to 3 days, the level is emergency.

[0126] Similarly, the system determines the overall level of functional degradation based on the worst-case scenario.

[0127] S430: Based on the preset fusion decision rules, the cascading failure risk level, geometric loss level, and functional degradation level are fused to generate the graded early warning signal;

[0128] The fusion decision rules include:

[0129] If the risk level of the cascading failure reaches the highest level, then the highest level warning signal will be generated directly.

[0130] Otherwise, based on the current combination of the geometric loss level and the functional degradation level, a preset biphasic risk decision matrix is ​​queried, and a corresponding early warning signal is generated.

[0131] Specifically, in step S430, the system first checks whether there are any areas in all assessment areas where the cascading failure risk level reaches the emergency level. If one or more areas are at the emergency level, the system ignores geometric loss and functional degradation levels and directly outputs the highest level warning. This ensures an absolute priority response to sudden, highly destructive cascading risks.

[0132] If the risk level of cascading failure does not reach the emergency level, the system will then make decisions based on the gradual loss trend. The specific method is as follows:

[0133] The overall geometric loss level and overall functional degradation level determined in step S420 are used as input. A preset biphasic risk decision matrix is ​​queried. This matrix is ​​a two-dimensional lookup table, with row indices representing geometric loss levels (normal, watch out, warning, emergency) and column indices representing functional degradation levels (normal, watch out, warning, emergency). Each cell defines the corresponding final warning signal level. For example:

[0134] When both the geometric loss level and the functional degradation level are normal, the lowest level warning is output.

[0135] When one dimension enters the attention level while the other dimension is normal, a low-level warning is output.

[0136] When any dimension enters the warning level, or both dimensions are at the attention level, a medium-level warning is output.

[0137] When any dimension enters the emergency level, or both dimensions are at the warning level, a high-level warning is output.

[0138] When both dimensions are at the emergency level, a high-level warning is also output. Since the chain risk has not reached the emergency level, it does not jump to the highest level.

[0139] The system maps corresponding early warning signals from a matrix based on the current combination of geometric loss level and functional degradation level. The final generated graded early warning signal includes the early warning level, the main risk type that triggered the warning (such as urgent cascading failure risk, urgent geometric loss and functional degradation warning, etc.), the specific liner area number with the highest risk or about to reach the threshold, and the key prediction time point. This signal is transmitted to equipment maintenance personnel in real time through various methods such as the monitoring system interface, audible and visual alarms, and mobile terminal push notifications.

[0140] Through steps S410 to S430, this embodiment realizes a complete decision-making chain from multi-source data to quantitative indicators, then to multi-dimensional levels, and finally integrates them into a clear and operable hierarchical early warning signal, providing a precise triggering basis for predictive maintenance.

[0141] Further, in step S400, generating maintenance decision suggestions based on the graded early warning signals includes:

[0142] S440: Obtain production plan information for a future preset period.

[0143] Specifically, in step S440, the system automatically obtains detailed production plan information for a future preset period from the enterprise's production management system, planning and scheduling system, or manual data entry platform through a preset data interface. This preset period should at least cover the most pressing risk time window indicated by the warning signal, and can typically be set to the next 30 to 90 days.

[0144] The acquired production planning information is structured data and must include at least: a clear planned production schedule, which details the specific production time and planned equipment runtime for each future workday and shift; specific production tasks and load data, including the types of materials to be transported, the expected transport volume, and key characteristic parameters reflecting the abrasiveness of the materials; pre-arranged downtime windows and their durations for equipment maintenance, process switching, or other non-production operations; and higher-level production priority and constraint information, such as special high-production periods, important supply guarantee periods, or unchangeable order requirements.

[0145] S450: Match and analyze the graded early warning signals, evolution maps and production plan information to dynamically recommend maintenance time windows and corresponding maintenance methods;

[0146] The maintenance methods include at least one of partial repair, overall replacement, or adjustment of operating parameters.

[0147] Specifically, in step S450, the system deeply integrates the graded early warning signal, the dynamic coupling evolution map and the future production plan information through intelligent matching analysis, so as to dynamically recommend the optimal maintenance time window and the corresponding maintenance method.

[0148] Specifically, the system first integrates three types of input data: a graded early warning signal that includes level, risk basis and key time prediction; an evolution map that reveals the spatiotemporal evolution of future wear depth and protection coefficient; and a structured production plan that specifies production period, task load and predetermined downtime window.

[0149] Based on this, the system first analyzes the urgency of maintenance needs, and then, according to the warning level and key time points extracted from the evolution map (such as the latest deadline for mandatory maintenance and the recommended buffer start time for maintenance), and combined with the spatial distribution characteristics of risks, it preliminarily determines the appropriate maintenance method—such as repair for local risks, replacement for overall losses, or temporary mitigation by adjusting operating parameters.

[0150] Furthermore, the system uses the suggested maintenance time as a benchmark to intelligently match feasible windows in the production plan: it prioritizes coupling the predetermined planned downtime periods to maximize the use of non-production time; if there is no predetermined window, it looks for nearby production gaps to perform short-time operations; if a longer operation time is required, it assesses the downtime cost within the risk time window and recommends a negotiated downtime window with the least impact on production.

[0151] Ultimately, the system generates structured maintenance decision recommendations, clearly defining the recommended time window, maintenance method, decision basis, and alternative solutions. These recommendations are then automatically pushed to the operations and maintenance system, thereby achieving seamless integration from technical status perception to economic and feasible operations and maintenance actions, and promoting the precise implementation of predictive maintenance.

[0152] Example 2

[0153] Based on Embodiment 1 above, this embodiment provides another method for online monitoring and early warning of wear status of steel-based ceramic liners. The same content as Embodiment 1 will not be repeated here; the difference lies in:

[0154] Before step S200, the method further includes:

[0155] A pulsed thermal excitation was applied to the surface of the liner, and the infrared temperature field sequence of its surface was acquired after the excitation ended;

[0156] The changes of the infrared temperature field sequence over time are analyzed, and parameter maps reflecting the thermal property distribution at different locations on the liner surface are obtained through inversion calculation. These parameter maps are used to indirectly characterize the coverage density and distribution uniformity of ceramic particles on the liner surface.

[0157] Based on the parameter diagram, the failure mode and / or spatial distribution information of the failure event information are verified and corrected.

[0158] Specifically, while acoustic emission signals can effectively capture microscopic failure events such as dynamic detachment of ceramic particles, they are prone to misjudgment or omission due to interference from vibrations and material impacts in the equipment's operating environment, resulting in insufficient accuracy of failure event information. Although the parameter diagrams obtained from thermophysical analysis cannot reflect the dynamic detachment process, they have the advantages of strong anti-interference ability and stable measurement. Therefore, this solution adds a thermophysical analysis verification step to improve monitoring accuracy through the complementarity of the two. This verification method is applicable to the aforementioned steel-based ceramic liner application scenario, and the specific implementation process is as follows.

[0159] After completing the initial acquisition of multi-source monitoring signals in step S100, active pulsed thermal excitation is applied to the surface of the liner. A pulsed infrared heating source is used and fixed directly in front of the monitoring area of ​​the liner to ensure uniform energy coverage. The thermal excitation parameters can be set to a pulse power of 500-800W, a single duration of 3-5 seconds, and applied once every 2 hours, coordinated with the signal acquisition cycle. The temperature feedback module ensures that the initial operating conditions are consistent for each excitation.

[0160] After thermal excitation, an infrared thermal imager is activated to acquire a thermal imaging sequence. The thermal imager is coaxially positioned with the heating source, and its parameters can be set to an imaging resolution of 640×480 pixels, a sampling frame rate of 10 frames / second, and an acquisition duration of 30 minutes to completely record the entire cooling process. During acquisition, the timestamps are synchronized with the laser rangefinder and acoustic emission sensor, and interference is eliminated through ambient temperature compensation to ensure data consistency in the temporal dimension.

[0161] Temperature-time cooling curves of each pixel in the infrared thermal imaging sequence are extracted. Based on the principle of heat conduction, the differences in thermal conductivity and heat capacity between ceramic particles and the metal matrix are utilized. When ceramic particles detach, the surface thermal properties at that location will approach those of the metal matrix, leading to characteristic changes in the local cooling curve. By analyzing the spatiotemporal heterogeneity of the cooling curve distribution, the coverage density and uniformity of ceramic particles on the liner surface can be inverted and imaged, thereby generating a two-dimensional thermophysical property distribution parameter map. This parameter map is unaffected by operating vibrations, material impacts, etc., and can objectively reflect the actual coverage state of ceramic particles on the liner surface, providing a reliable basis for correcting deviations in acoustic emission signals.

[0162] Determining the threshold for thermal conductivity: 10-30 W / (m²) in the area covered by ceramic particles. K), the exposed area of ​​the metal substrate is 40-60 W / (m²). K). The thermal property distribution parameter map is compared region by region with the generated failure event information spatial distribution map: If the parameter map of a certain region shows that the thermal conductivity is higher than the metal matrix threshold, and the region is initially judged to have no failure events, it means that the acoustic emission signal missed the ceramic particle detachment, and the failure event information needs to be corrected, adding single detachment or group detachment failure modes; If the parameter map of a certain region shows that the thermal conductivity is within the ceramic particle threshold range, but it is initially judged to be a group detachment failure, it means that the acoustic emission signal has misjudged, and the group detachment label of the region needs to be canceled, while retaining the original effective protection status record of the ceramic particles; If the failure range marked by the failure event information of a certain region deviates from the boundary of the thermal property anomaly region in the parameter map, the regional boundary of the parameter map shall be used as the standard to correct the spatial distribution position of the failure event.

[0163] After verification and correction, the updated failure event information is synchronized to the multi-source monitoring signal dataset, replacing the initially generated failure event information. This provides more accurate input data for subsequent wear analysis models embedded with physical constraints. The entire process effectively corrects inaccuracies in acoustic emission signals caused by scene interference through objective feedback from thermophysical characteristics, achieving complementarity between dynamic failure capture and stable state verification, further improving the accuracy and reliability of wear condition monitoring for steel-based ceramic liners.

[0164] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for online monitoring and early warning of wear condition of steel-based ceramic liners, characterized in that, For steel-based ceramic liners composed of a metal matrix and embedded ceramic particles, the method includes: synchronously acquiring multi-source monitoring signals online, wherein the multi-source monitoring signals include at least a distance signal reflecting the liner thickness, an acoustic emission signal reflecting material failure, and equipment operating condition parameters; processing the multi-source monitoring signals using a wear analysis model embedded with physical constraints to output decoupling evaluation results; the physical constraints characterize the coupling relationship between the wear state of the metal matrix and the protective effect of the ceramic particles; based on the evaluation results, predicting the evolution trend of the two-phase wear coupling of the liner and identifying the cascading failure risk of accelerated wear of the metal matrix triggered by ceramic particle failure; combining the level of the cascading failure risk with the evolution trend to obtain graded early warning signals, and generating maintenance decision recommendations based on the graded early warning signals; the wear analysis model embedded with physical constraints is a physical information neural network model, and the physical constraints include the physical constraints on the metal matrix... The wear rate is coupled with the ceramic particle protection coefficient; the decoupling evaluation result includes the wear rate of the metal matrix of the liner, the ceramic particle protection coefficient, and the corresponding spatial distribution information; the value range of the ceramic particle protection coefficient is 0 to 1, where 1 indicates that the ceramic particles are intact and provide complete protection, and 0 indicates that the ceramic protection function is completely lost; based on the evaluation result, the two-phase wear coupling evolution trend of the liner is predicted, including: constructing a two-phase wear state space model with the ceramic particle protection coefficient as the core state variable; inputting the current state and historical change trend of the metal matrix wear rate and the ceramic particle protection coefficient, as well as the equipment operating condition parameters into the two-phase wear state space model for iterative calculation, and simulating the transmission process through the physical constraints; outputting the evolution spectrum of the liner in the future set time period, the evolution spectrum including the predicted spatial distribution and evolution sequence of the metal matrix wear depth and the ceramic particle protection coefficient at different time points.

2. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 1, characterized in that, Acquiring the acoustic emission signal used to reflect material failure includes: acquiring the original acoustic emission signal and matching it with a preset feature spectrum library; the feature spectrum library includes at least signal features characterizing three types of failure modes: ceramic particle microcracks, single particle detachment, and group detachment; based on the matching result, generating failure event information as the acoustic emission signal, the failure event information including at least one of the failure modes and corresponding spatial distribution information.

3. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 2, characterized in that, Before processing the multi-source monitoring signals using a wear analysis model embedded with physical constraints to output decoupled evaluation results, the method further includes: applying pulsed thermal excitation to the surface of the liner and acquiring an infrared temperature field sequence of the surface after the excitation ends; analyzing the change of the infrared temperature field sequence over time and obtaining a parameter map reflecting the thermal property distribution at different locations on the liner surface through inversion calculation, wherein the parameter map is used to indirectly characterize the coverage density and distribution uniformity of ceramic particles on the liner surface; and verifying and correcting the failure mode and / or spatial distribution information of the failure event information based on the parameter map.

4. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 3, characterized in that, The method for identifying the risk of cascading failures triggered by ceramic particle failure leading to accelerated wear of the metal matrix includes: based on the failure event information, calculating in real time the spatial clustering degree of single and group detachment failure modes of ceramic particles within the evaluation area; when the spatial clustering degree is greater than a first threshold, and the rate of decrease in the protection coefficient of ceramic particles in the area is greater than a second threshold as identified from the evolution map, determining that the area has entered a cascading failure state; based on the determination of the cascading failure state, simulating a corresponding decrease in the protection coefficient of ceramic particles in the area in the two-phase wear state spatial model, and iteratively calculating the cascading reaction process of accelerated wear of the metal matrix and the associated decrease in the protection coefficient of ceramic particles in adjacent areas; extracting the spatial expansion rate of the cascading reaction process based on the simulation results of the cascading reaction process, and calculating the increase in wear of the metal matrix caused by the process; and calculating the cascading failure risk index based on the spatial expansion rate and the wear increment.

5. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 4, characterized in that, The step of calculating the cascading failure risk index based on the spatial expansion rate and the wear increment includes: normalizing the spatial expansion rate and the wear increment; and weighting and summing the normalized spatial expansion rate and the wear increment according to preset weight coefficients to obtain the cascading failure risk index; wherein the weight coefficient of the spatial expansion rate is greater than the weight coefficient of the wear increment.

6. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 5, characterized in that, The step of obtaining a graded early warning signal by comprehensively considering the level of the cascading failure risk and the evolution trend includes: comparing the cascading failure risk index with multiple preset risk thresholds to determine the cascading failure risk level; determining the future time point when the predicted wear depth of the metal matrix first reaches a preset thickness threshold based on the evolution map, and determining the geometric loss level based on the proximity of this time point to the current time; determining the future time point when the ceramic particle protection coefficient first falls below a preset functional threshold, and determining the functional degradation level based on the proximity of this time point to the current time; and fusing the cascading failure risk level, geometric loss level, and functional degradation level according to a preset fusion decision rule to generate the graded early warning signal; the fusion decision rule includes: if the cascading failure risk level reaches the highest level, then directly generating the highest level early warning signal; otherwise, querying a preset biphasic risk decision matrix based on the current combination of the geometric loss level and the functional degradation level, and mapping to generate the corresponding early warning signal.

7. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 6, characterized in that, The step of generating maintenance decision recommendations based on the graded early warning signals includes: obtaining production plan information for a future preset period; matching and analyzing the graded early warning signals, evolution maps, and production plan information to dynamically recommend maintenance time windows and corresponding maintenance methods; wherein, the maintenance methods include at least one of partial repair, overall replacement, or adjustment of operating parameters.

8. The method for online monitoring and early warning of wear condition of steel-based ceramic liners according to claim 1, characterized in that, The distance signal is acquired through a non-contact laser rangefinder array, which is fixedly installed on the outside of the equipment housing on the back of the liner and measured through an observation window.

Images