Method and device for predicting failure of adsorption device by fusing multi-modal data

Abstract

The invention provides an adsorption device failure prediction method and device fusing multi-modal data, and relates to the technical field of data processing, and the method comprises the steps: building a multi-modal database, and determining the life correlation factors of an adsorption device; a detection standard is introduced, a failure principle is established, a deep learning model is used for training a prediction model, and failure reminding is carried out, so that the problems that conventional failure prediction of the adsorption device depends on physical monitoring, comprehensive analysis on the global state of equipment is lacked, and interaction influence among different data types is neglected are solved; the method solves the technical problems of insufficient precision and reliability of failure prediction due to the fact that the adsorption device is in a non-contact state, and achieves the purposes of determining life-related factors by fusing multi-modal data, dynamically adapting to environmental changes, monitoring the health state of the adsorption device in real time through an enhanced detection test standard and a failure evaluation principle, and timely sending out a failure prompt when potential risks occur. The technical effect of comprehensively evaluating and monitoring the long-term service state of the adsorption device is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and a device for predicting failure of an adsorption device by fusing multimodal data. Background Art

[0002] With the continuous development of industrialization and modernization, adsorption devices are widely used in many industries, especially in the fields of chemical industry, petroleum, energy, etc., and undertake important gas adsorption and separation tasks. These adsorption devices are often in complex working environments, especially in hydrogen-rich (H 2 ) atmosphere, and operates under extreme conditions such as alternating loads, temperature and pressure for a long time. In this environment, the materials and structures of the adsorption device are susceptible to corrosion, fatigue and damage, which in turn leads to degradation of equipment performance, or even malfunction or failure. In order to ensure the long-term safe and reliable operation of the adsorption device, it is particularly important to provide early warning and predict its failure state.

[0003] At present, failure prediction of adsorption devices mainly relies on traditional physical monitoring, such as ultrasonic detection, magnetic detection, etc. These methods can usually only monitor external defects or local damage of the equipment, and it is difficult to fully reflect the various failure risks of the equipment in a complex environment. In addition, traditional life prediction often relies on a single data type (such as residual stress or microstructure), ignoring the special effects of the adsorption device under specific working environments, such as hydrogen-rich atmosphere and alternating loads, resulting in insufficient prediction accuracy and reliability.

[0004] In the prior art, conventional failure prediction of adsorption devices relies on physical monitoring, lacks a comprehensive analysis of the global status of the equipment, and ignores the interaction between different data types, resulting in technical problems such as insufficient accuracy and reliability of failure prediction. Summary of the invention

[0005] The present application provides a failure prediction method and device for an adsorption device that integrates multimodal data, thereby solving the technical problems that conventional failure prediction of an adsorption device relies on physical monitoring, lacks a comprehensive analysis of the global status of the equipment, and ignores the interaction between different data types, resulting in insufficient accuracy and reliability of failure prediction. The application achieves the technical effect of integrating multimodal data to determine life-related factors, dynamically adapting to environmental changes, monitoring the health status of the adsorption device in real time through enhanced detection test standards and failure evaluation principles, and promptly issuing failure reminders when potential risks arise, thereby achieving a comprehensive evaluation and monitoring of the long-term service status of the adsorption device.

[0006] In view of the above problems, the present application provides a method and device for predicting failure of an adsorption device by integrating multimodal data.

[0007] In the first aspect, the present application provides a method for predicting failure of an adsorption device by integrating multimodal data, wherein the method comprises: setting a multimodal database, the multimodal database being used to store microstructure data, defect feature data, and residual stress data of the adsorption device, establishing a mapping relationship between the microstructure of the adsorption device and service performance, defect feature and service performance, and residual stress and service performance; using the Delphi method, according to the adsorption device in H-rich 2 The service life weight factor under the environment is optimized by arranging and combining the characteristic parameters of microstructure, defect characteristics and residual stress to determine the life correlation factors of the adsorption device; 2 A multimodal detection test standard for adsorber life prediction under alternating load conditions; Establish a failure evaluation principle based on multimodal detection signals, which is used to define H-rich adsorber failure evaluation principles. 2 The life threshold value and life evaluation criteria of the adsorption device under alternating load conditions; based on the multimodal database, using a deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria, training the failure prediction model of the adsorption device, and performing failure reminders.

[0008] In a second aspect, the present application provides a failure prediction device for an adsorption device integrating multimodal data, wherein the device comprises: a database setting unit, the database setting unit is used to set a multimodal database, the multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and establish a mapping relationship between the microstructure of the adsorption device and service performance, defect feature and service performance, and residual stress and service performance; an optimization calculation unit, the optimization calculation unit is used to use the Delphi method to predict the failure of the adsorption device according to the adsorption device in H-rich 2 The service life weight factor under the environment is used to optimize the arrangement and combination of characteristic parameters of microstructure, defect characteristics and residual stress to determine the life correlation factors of the adsorption device; the standard introduction unit is used to introduce the service life weight factor suitable for H-rich 2 A multimodal detection test standard for adsorber life prediction under alternating load conditions; an evaluation principle establishment unit, the evaluation principle establishment unit is used to establish a failure evaluation principle based on multimodal detection signals, the failure evaluation principle is used to define H-rich 2 The life threshold value and life evaluation criteria of the adsorption device under alternating load conditions; a failure reminder unit, which is used to train the failure prediction model of the adsorption device and provide failure reminders based on the multimodal database, using a deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The multimodal database was established to store the adsorption device data, the Delphi method was used to determine the life-related factors, the multimodal detection test standards were introduced, the failure evaluation principles were established, and the failure prediction model was trained by integrating multiple factors using a deep learning model to provide failure reminders to evaluate the H-rich 2 The life of the adsorption device under alternating loads, this application has achieved the goal of comprehensively considering multiple factors such as microstructure, defect characteristics, residual stress, etc., to determine the life-related factors. In a hydrogen-rich environment, especially under extreme working conditions such as alternating loads, the failure prediction model can dynamically adapt to environmental changes, and through enhanced detection test standards and failure evaluation principles, the health status of the adsorption device is monitored in real time. When potential risks occur, failure reminders are issued in time, thereby achieving the technical effect of comprehensive evaluation and monitoring of the long-term service status of the adsorption device. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of a flow chart of a method for predicting failure of an adsorption device by fusing multimodal data in this application;

[0012] Figure 2 This is a schematic diagram of the structure of the adsorption device failure prediction device that integrates multimodal data in this application.

[0013] Explanation of the accompanying drawings: database setting unit 11, optimization calculation unit 12, standard introduction unit 13, evaluation principle establishment unit 14, failure reminder unit 15. DETAILED DESCRIPTION

[0014] The present application provides a failure prediction method and device for an adsorption device that integrates multimodal data, thereby solving the technical problems that conventional failure prediction of an adsorption device relies on physical monitoring, lacks a comprehensive analysis of the global status of the equipment, and ignores the interaction between different data types, resulting in insufficient accuracy and reliability of failure prediction. The application achieves the technical effect of integrating multimodal data to determine life-related factors, dynamically adapting to environmental changes, monitoring the health status of the adsorption device in real time through enhanced detection test standards and failure evaluation principles, and promptly issuing failure reminders when potential risks arise, thereby achieving a comprehensive evaluation and monitoring of the long-term service status of the adsorption device.

[0015] Embodiment 1

[0016] like Figure 1 As shown, the present application provides a method for predicting failure of an adsorption device by fusing multimodal data, wherein the method comprises:

[0017] S100: Setting a multimodal database, wherein the multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and establish a mapping relationship between the microstructure of the adsorption device and service performance, defect feature and service performance, and residual stress and service performance; S200: Using the Delphi method, according to the adsorption device in H-rich 2 The service life weight factor under the environment is used to optimize the arrangement and combination of characteristic parameters of microstructure, defect characteristics, and residual stress to determine the life correlation factors of the adsorption device; S300: Introduce the service life weight factor suitable for H-rich 2 Multimodal testing standard for adsorber life prediction under alternating load conditions.

[0018] Specifically, a multimodal database is a database system that integrates multiple types of data and is capable of storing and managing different types of data, such as microstructure data, defect characteristic data, and residual stress data. These data types reflect different aspects and characteristics of the adsorption device. The purpose of the multimodal database is to provide a comprehensive data platform to facilitate the analysis and prediction of the failure risk of the adsorption device; microstructure data refers to the microstructural information of the adsorption device material, including but not limited to grain size, phase distribution, and microscopic defects, etc. This information is crucial for understanding the mechanical properties and durability of the material; defect characteristic data refers to defects that may exist in the adsorption device, such as cracks, holes or other types of defects, which may affect the service performance and life of the device.

[0019] Residual stress data refers to the stress generated during the manufacturing and processing process and existing inside the material. These stresses may affect the structural integrity of the device during service. The mapping relationship refers to the correlation between different data types (microstructure, defect characteristics, residual stress) and the service performance of the adsorption device established through data analysis. This relationship helps to understand which factors have an important impact on the life and performance of the device. The Delphi method refers to collecting expert opinions through multiple rounds of anonymous surveys and feedback to reach a consensus or predict results. In the embodiments of this application, it is used to 2 The service life weight factor under the environment is used to optimize the characteristic parameters; 2 Environment refers to a hydrogen-rich environment, which may have a special impact on the material properties and service performance of the adsorption device; the service life weight factor refers to the relative importance of various factors affecting the life of the adsorption device, which may include material properties, working environment and operating conditions; the multimodal detection test standard refers to the H-rich environment. 2 A series of standardized tests and evaluation methods for predicting adsorber life under alternating load conditions.

[0020] The purpose of establishing a multimodal database is to integrate and store various data related to the performance of the adsorption device. This step is crucial in the entire program because it provides a basis for subsequent data analysis and model training. Through the comprehensive analysis of microstructure, defect characteristics and residual stress data, a more comprehensive understanding of the actual state and potential failure risk of the adsorption device can be achieved. In addition, the establishment of a mapping relationship is helpful to identify the key factors affecting the service performance of the device and provide a scientific basis for subsequent failure prediction. The Delphi method is used to gather expert wisdom and conduct a comprehensive analysis of the adsorption device in H-rich water. 2 The service life weight factors under the environment are evaluated and optimized. This process helps to determine which characteristic parameters of microstructure, defect characteristics and residual stress have the greatest impact on the life of the adsorption device. In this way, the specific impact of each factor on the life can be quantified, thereby providing accurate input parameters for the prediction model.

[0021] Introducing H-rich 2 The purpose of the multimodal testing standard under alternating load conditions is to ensure that the prediction model can adapt to the specific working environment. The role of this step is to provide a standardized framework for evaluating and predicting the performance and life of the adsorption device under actual working conditions. Through these standards, the performance of the adsorption device in H-rich conditions can be simulated and predicted more accurately. 2 The failure behavior in the environment can be analyzed to improve the practicality and accuracy of the prediction model.

[0022] S400: Establishing a failure evaluation principle based on multimodal detection signals, wherein the failure evaluation principle is used to define H-rich 2 Life threshold value and life evaluation criteria of the adsorption device under alternating load conditions; S500: Based on the multimodal database, using a deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold value and life evaluation criteria, training a failure prediction model for the adsorption device, and performing failure reminders.

[0023] Specifically, the failure evaluation principle is an evaluation system based on multimodal detection signals, which is used to evaluate the failure of adsorption devices under specific conditions (such as H-rich 2 The performance status and failure risk of the adsorption device under alternating load conditions) include a series of standards and indicators to determine whether the device reaches or exceeds the predetermined life threshold value, so as to judge whether it has failed; the life threshold value refers to a preset threshold value. When the performance parameters of the adsorption device drop below this value, it is considered that the device has reached the limit of its expected life and needs to be replaced or repaired.

[0024] The life evaluation criteria are used to evaluate the life and performance status of the adsorption device under specific conditions, including the evaluation of the service performance, durability and reliability of the device, and the prediction of potential failure risks; the deep learning model can learn complex patterns and relationships from large amounts of data. The deep learning model is used to analyze multimodal data to predict the failure risk of the adsorption device; failure reminder means that when the performance status of the adsorption device approaches or reaches the life threshold, the system will issue a reminder so that appropriate maintenance or replacement measures can be taken.

[0025] Establishing failure evaluation principles based on multimodal detection signals is one of the key steps, which involves defining and determining H-rich 2 The life threshold value and life evaluation criteria of the adsorption device under alternating load conditions. Through this principle, the failure risk of the device can be quantified and the evaluation criteria can be provided for the subsequent prediction model. The role of this step is to provide a clear evaluation framework for the health management of the adsorption device, making the maintenance and decision-making process more scientific and systematic.

[0026] After establishing the failure evaluation principles, a deep learning model is used to train the failure prediction model of the adsorption device based on the rich data in the multimodal database. This step involves integrating key information such as multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria. The deep learning model can learn patterns to predict failure risks from these complex multimodal data. In the above steps, the health status of the adsorption device is monitored in real time, and an early warning is issued when the device approaches the failure threshold, thereby achieving a comprehensive evaluation and monitoring of the long-term service status of the adsorption device and improving its safety and reliability.

[0027] Furthermore, the multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and the method further includes:

[0028] Preprocess the microstructure data, defect feature data, and residual stress data to obtain multimodal preprocessed data; enhance the multimodal preprocessed data and select the optimal splitting point; 2 According to the law of environmental changes in the environment, a model of the mutual construction mechanism of multiple types of damage modes is constructed based on the optimal splitting point.

[0029] Specifically, preprocessing refers to the processing of directly collected microstructure data, defect feature data, and residual stress data to improve data quality and make it more suitable for subsequent analysis and model training. The preprocessing steps include data cleaning, normalization, denoising and other operations; multimodal preprocessed data refers to preprocessed data sets from different data sources (such as microstructure data, defect feature data, residual stress data), which are integrated together to facilitate multimodal analysis; data augmentation is a technique to increase the size of a data set by creating variants of the data, with the aim of improving the generalization ability of the model and reducing overfitting. In this scenario, enhancement may involve applying specific transformations to the preprocessed data to simulate different damage modes.

[0030] The optimal split point refers to the best position under which feature or condition to divide the data set into subsets when building the tree model. The optimal split point is selected to maximize the performance of the model, such as classification accuracy. The multi-class damage mode interaction mechanism model aims to understand and simulate the adsorption device in H-rich 2 The model can help us deeply understand how different damage modes affect the overall performance and life of the device.

[0031] Preprocessing is an important step in the data preparation stage. By preprocessing microstructure data, defect feature data, and residual stress data, cleaner and more consistent multimodal preprocessed data can be obtained. The purpose of this step is to improve the quality of the data, reduce the impact of noise and outliers, and make the data more suitable for subsequent analysis and model training. The preprocessed data can more accurately reflect the actual state of the adsorption device and provide reliable input for failure prediction.

[0032] The multimodal preprocessed data were enhanced and the optimal splitting point was selected. Furthermore, in order to simulate the adsorption device in H-rich 2 The role of this step is to increase the diversity and complexity of the data set and improve the model's ability to recognize and predict different damage modes in the environment. Through data enhancement, the model can learn a wider range of damage characteristics, thereby improving its generalization ability and prediction accuracy.

[0033] Construct a multi-type damage mode interaction mechanism model. Specifically, based on the rich H 2 The environmental change laws under the environment are studied, and the optimal splitting point is used to construct a model of the mutual construction mechanism of multiple types of damage modes. This can further understand and simulate the interaction and conversion mechanism between different damage modes, and accurately predict the behavior and life of the adsorption device under specific conditions, thereby providing a scientific basis for maintenance and replacement decisions.

[0034] Furthermore, for H-rich2 According to the law of environmental changes in the environment, a multi-type damage mode mutual construction mechanism model is constructed according to the optimal splitting point, and the method includes:

[0035] The KNN classifier is trained and a distance measurement method is set; based on the distance measurement method, a distance distribution map is drawn up; according to the distance distribution map, clustering characteristics under different damage modes are analyzed to determine multiple types of damage modes.

[0036] Specifically, the KNN classifier is an instance-based machine learning algorithm used for classification and regression. It predicts the category of a sample by voting based on the categories of its K nearest neighbor samples. In the KNN algorithm, distance measurement refers to the method used to calculate the similarity between samples. Common distance measurement methods include Euclidean distance, Manhattan distance, Minkowski distance, etc.

[0037] The distance distribution map is used to display the distance distribution of different samples in the feature space. Through the distance distribution map, the similarities and differences between samples can be intuitively observed; the clustering characteristic refers to the aggregation of data points in the feature space. By analyzing the clustering characteristics, the natural groupings or categories in the data can be identified; multi-class damage modes refer to different types of damage that the adsorption device may suffer, such as cracks, corrosion, fatigue, etc. Each damage mode has its specific characteristics and influences.

[0038] Train the KNN classifier. Specifically, select a suitable distance measurement method and use training data to train the KNN model. The role of the KNN classifier is to classify new samples based on known damage samples and predict their damage types. By setting a suitable distance measurement method, the accuracy and robustness of the classifier can be improved, making it better adapted to the characteristics of multimodal data.

[0039] Based on the selected distance metric, a distance distribution map is drawn up. This step involves calculating the distance between samples and displaying the distance information on the map. The role of the distance distribution map is to provide an intuitive way to observe and analyze the similarities and differences between samples, which is crucial for understanding the structure and characteristics of the data. Through the distance distribution map, we can better understand the distribution of different damage patterns in the feature space, providing a basis for subsequent clustering analysis and damage pattern recognition.

[0040] The clustering characteristics under different damage modes are analyzed according to the distance distribution map to determine multiple types of damage modes. Specifically, the data clustering in the feature space is identified and analyzed to discover different damage modes, identify the different types of damage that the adsorption device may suffer, and establish a feature model for each damage mode. By analyzing the clustering characteristics, new damage samples can be more accurately predicted and identified, thereby improving the accuracy and practicality of the failure prediction model.

[0041] Furthermore, according to the distance distribution map, clustering characteristics under different damage modes are analyzed to determine multiple types of damage modes. The method includes:

[0042] An incremental learning mechanism is introduced to allow the distance distribution map to be updated synchronously; based on the distance distribution map, a regular calibration mechanism is established.

[0043] Specifically, the incremental learning mechanism is a machine learning paradigm that allows the learning algorithm to learn gradually, that is, the model is able to update its knowledge when it receives new data without the need to retrain from scratch. This mechanism is particularly useful for processing continuously incoming data streams; in incremental learning, synchronous update means that the model updates its parameters or knowledge base immediately after receiving new data to reflect the latest data features; the periodic calibration mechanism is a periodic verification and adjustment process used to ensure that the performance of a model or system remains at the expected level. During calibration, the model may be retrained with new data or its parameters may be adjusted to adapt to changes in the distribution of data.

[0044] The purpose of introducing an incremental learning mechanism into the failure prediction model is to enable the model to adapt to the data distribution and environmental conditions that change over time, and to allow the model to dynamically update its parameters and knowledge base, including the distance distribution map, when receiving new multimodal data. The role of the incremental learning mechanism is to improve the adaptability and flexibility of the model, so that it can capture the latest damage patterns and changing trends, thereby maintaining prediction accuracy and relevance.

[0045] Establish a regular calibration mechanism to periodically evaluate the performance of the model and adjust the model based on the latest data. The role of the regular calibration mechanism is to ensure the long-term stability and accuracy of the model. Through regular calibration, it can be detected whether the model is still suitable for the current data distribution, and whether parameters need to be adjusted or retrained to adapt to new data features, maintain the predictive performance of the model, and reduce potential risks caused by outdated models.

[0046] Furthermore, based on the distance distribution map, a regular calibration mechanism is established, and the method includes:

[0047] Obtain a calibration cycle of the periodic calibration mechanism; within the calibration cycle, collect timestamp-marked microstructure data, timestamp-marked defect feature data, and timestamp-marked residual stress data as calibration samples; use the calibration samples to retrain the KNN classifier and obtain a calibrated distance distribution map.

[0048] Specifically, the calibration cycle refers to the time interval set in the regular calibration mechanism. At the end of this time interval, the calibration process will be executed to update and adjust the model; timestamp data means that each data record is accompanied by a timestamp, which records the specific time when the data was generated, which is very important for tracking data changes over time and trend analysis; in machine learning, calibration samples refer to the data set used for model calibration, which is used to adjust model parameters to ensure that the model can accurately reflect the latest data features; retraining refers to additional training of an already trained model using new data or calibration samples to update model parameters and improve the accuracy and adaptability of the model.

[0049] The purpose of obtaining a regular calibration mechanism is to ensure that the model can continuously adapt to new data. Specifically, determine an appropriate time interval, at the end of which the model will be calibrated. The setting of the calibration cycle needs to take into account the speed of data changes and business needs to ensure that the model can be updated in a timely manner while avoiding waste of resources caused by overly frequent calibration. Determination of the calibration cycle helps maintain the long-term performance and accuracy of the model.

[0050] Timestamped microstructure data, defect feature data, and residual stress data are collected during each calibration cycle. These data are used as calibration samples to update and adjust the model. These data are collected to capture the state changes of the adsorption device at different time points to ensure that the model can reflect the latest equipment status and environmental conditions. Using timestamped data can help analyze trends and patterns in the data over time, which is crucial to the accuracy of the prediction model.

[0051] Retraining the KNN classifier using the collected calibration samples is the core step of model calibration. Specifically, the model parameters are updated using the latest calibration samples to ensure that the model can accurately reflect the latest data features and trends. The role of retraining is to improve the adaptability and accuracy of the model so that it can better predict the failure risk of the adsorption device. Through retraining, the model can learn new damage patterns and change trends, thereby improving the accuracy and reliability of failure prediction. After retraining, the calibrated distance distribution map can be obtained, which will provide a more accurate basis for subsequent analysis and decision-making.

[0052] Furthermore, the KNN classifier is retrained using the calibration sample, and the method further includes:

[0053] Performance evaluation indicators are set, and the performance evaluation indicators include accuracy, recall rate, and F1 score; the performance of the distance distribution map before and after calibration is evaluated by the performance evaluation indicators, and the performance evaluation changes before and after calibration are compared to perform calibration verification.

[0054] Specifically, performance evaluation indicators are used to measure the performance of the model, including accuracy, recall and F1 scores, which quantify the quality of the model's prediction results; accuracy refers to the proportion of samples predicted correctly by the model to the total samples, reflecting the accuracy of the model's prediction; recall, also known as true positive rate or sensitivity, refers to the proportion of positive samples successfully identified by the model to all actual positive samples, and the recall reflects the model's ability to identify positive samples; the F1 score is the harmonic mean of accuracy and recall, which is used to measure the overall performance of the model, especially in the case of class imbalance; calibration verification refers to the process of verifying the calibration effect by comparing the performance changes of the model before and after calibration. If the performance of the calibrated model is improved, it means that the calibration is effective.

[0055] During the model calibration process, setting performance evaluation indicators is an important step in measuring and monitoring model performance. By setting indicators such as accuracy, recall rate and F1 score, the prediction effect of the model can be quantified and optimized accordingly. The role of these indicators is to provide an objective standard to evaluate the predictive ability of the model, understand the performance of the model in identifying different damage modes, and make necessary adjustments.

[0056] Before and after calibration, using the set performance evaluation indicators to evaluate the performance of the distance distribution map is a key step in verifying the calibration effect. Specifically, the model prediction results before calibration are compared with the actual results, and then this process is repeated to evaluate the calibrated model. The purpose of the evaluation is to determine whether the calibration has improved the prediction accuracy and reliability of the model. By comparing the performance indicators before and after calibration, we can understand the specific impact of calibration on model performance and thus determine whether the calibration is successful.

[0057] Analyze the changes in performance evaluation indicators to determine whether the calibration is effective. If the calibrated model has improved in indicators such as accuracy, recall and F1 score, it means that the calibration is effective and the predictive performance of the model has been improved, thereby ensuring that the model can continue to adapt to new data and environmental conditions and maintain its ability to predict failure risks. Through calibration verification, the long-term performance and reliability of the model are guaranteed, ensuring that the failure prediction of the adsorption device is more accurate and timely.

[0058] Furthermore, the method further comprises:

[0059] Based on the integrated sensor monitoring module, a data transmission network is determined, wherein the integrated sensor monitoring module is used to collect microstructure data, defect characteristic data, and residual stress data of the adsorption device; an encrypted communication protocol is used to perform data security protection on the data transmission network.

[0060] Specifically, the integrated sensor monitoring module is used to collect key data of the adsorption device in real time, such as microstructure data, defect characteristic data and residual stress data; the data transmission network refers to the network that transmits the collected data from the monitoring module to the data processing center or storage system. This network can be wireless or wired and is responsible for data transmission and communication; the encrypted communication protocol is used to protect data from unauthorized access or tampering during the data transmission process. The encrypted communication protocol encrypts the data through an encryption algorithm to ensure the security and integrity of data transmission.

[0061] An integrated sensor monitoring module is deployed to collect the microstructure data, defect feature data and residual stress data of the adsorption device. The data transmission network is determined to ensure that these key data can be safely and efficiently transmitted to the data processing center. The role of the data transmission network is to provide a reliable data flow channel so that the collected data can be processed and analyzed in time, thereby realizing real-time monitoring of the status of the adsorption device; an encrypted communication protocol is used to protect the data transmission network to prevent the data from being intercepted or tampered with during the transmission process. Through the encrypted communication protocol, it can be ensured that only authorized users or systems can access and interpret the transmitted data. The role of data security protection is to improve the security of the system and prevent the leakage of sensitive data.

[0062] In summary, the adsorption device failure prediction method and device provided by the present application that integrates multimodal data have the following technical effects:

[0063] Due to the use of a multimodal database, the multimodal database is used to store the microstructure data, defect feature data, and residual stress data of the adsorption device, and to establish the mapping relationship between the microstructure of the adsorption device and the service performance, defect feature and service performance, and residual stress and service performance; the Delphi method is used to calculate the relationship between the microstructure of the adsorption device and the service performance, defect feature and service performance, and residual stress and service performance according to the adsorption device in H-rich 2 The service life weight factor under the environment is optimized by arranging and combining the characteristic parameters of microstructure, defect characteristics and residual stress to determine the life correlation factors of the adsorption device; 2 Multimodal detection test standard for adsorber life prediction under alternating load conditions; Establish failure evaluation principle based on multimodal detection signal, which is used to define H-rich adsorber failure evaluation principle. 2The life threshold value and life evaluation criteria of the adsorption device under alternating load conditions; based on the multimodal database, using the deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria, training the failure prediction model of the adsorption device, and performing failure reminders. This application has achieved the technical effect of integrating multimodal data to determine life-related factors, dynamically adapting to environmental changes, and monitoring the health status of the adsorption device in real time through enhanced detection test standards and failure evaluation principles. When potential risks arise, failure reminders are issued in a timely manner, thereby achieving a comprehensive evaluation and monitoring of the long-term service status of the adsorption device.

[0064] Embodiment 2

[0065] Based on the same inventive concept as the adsorption device failure prediction method integrating multimodal data in the aforementioned embodiment, Figure 2 As shown, the present application provides an adsorption device failure prediction device integrating multimodal data, wherein the device comprises:

[0066] The database setting unit 11 is used to set a multimodal database, which is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and establish a mapping relationship between the microstructure of the adsorption device and service performance, defect features and service performance, and residual stress and service performance.

[0067] The optimization calculation unit 12 is used to use the Delphi method to calculate the adsorption capacity of the adsorption device in the rich H 2 The service life weight factor under the environment is used to optimize the arrangement and combination of characteristic parameters of microstructure, defect characteristics and residual stress to determine the life-related factors of the adsorption device.

[0068] Standard introduction unit 13, the standard introduction unit is used to introduce H-rich 2 Multimodal testing standard for adsorber life prediction under alternating load conditions.

[0069] The evaluation principle establishing unit 14 is used to establish a failure evaluation principle based on the multimodal detection signal, and the failure evaluation principle is used to define the rich H 2 Life threshold value and life evaluation criteria of adsorption device under alternating load conditions.

[0070] The failure reminder unit 15 is used to train the failure prediction model of the adsorption device based on the multimodal database, using a deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria, and perform failure reminders.

[0071] Furthermore, the database setting unit 11 is also used to execute the following method:

[0072] Preprocess the microstructure data, defect feature data, and residual stress data to obtain multimodal preprocessed data; enhance the multimodal preprocessed data and select the optimal splitting point; 2 According to the law of environmental changes in the environment, a model of the mutual construction mechanism of multiple types of damage modes is constructed based on the optimal splitting point.

[0073] Furthermore, the database setting unit 11 is also used to execute the following method:

[0074] The KNN classifier is trained and a distance measurement method is set; based on the distance measurement method, a distance distribution map is drawn up; according to the distance distribution map, clustering characteristics under different damage modes are analyzed to determine multiple types of damage modes.

[0075] Furthermore, the database setting unit 11 is also used to execute the following method:

[0076] An incremental learning mechanism is introduced to allow the distance distribution map to be updated synchronously; based on the distance distribution map, a regular calibration mechanism is established.

[0077] Furthermore, the database setting unit 11 is also used to execute the following method:

[0078] Obtain a calibration cycle of the periodic calibration mechanism; within the calibration cycle, collect timestamp-marked microstructure data, timestamp-marked defect feature data, and timestamp-marked residual stress data as calibration samples; use the calibration samples to retrain the KNN classifier and obtain a calibrated distance distribution map.

[0079] Furthermore, the database setting unit 11 is also used to execute the following method:

[0080] Performance evaluation indicators are set, and the performance evaluation indicators include accuracy, recall rate, and F1 score; the performance of the distance distribution map before and after calibration is evaluated by the performance evaluation indicators, and the performance evaluation changes before and after calibration are compared to perform calibration verification.

[0081] Furthermore, the adsorption device failure prediction device integrating multimodal data is also used to perform the following method:

[0082] Based on the integrated sensor monitoring module, a data transmission network is determined, wherein the integrated sensor monitoring module is used to collect microstructure data, defect characteristic data, and residual stress data of the adsorption device; an encrypted communication protocol is used to perform data security protection on the data transmission network.

[0083] This specification and drawings are merely exemplary illustrations of the present application. Various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. These modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, and the present application is intended to include these modifications and variations.

Claims

1. A method for predicting failure of an adsorption device by integrating multimodal data, characterized in that: The method comprises: A multimodal database is set up, wherein the multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and establish a mapping relationship between the microstructure of the adsorption device and service performance, defect feature and service performance, and residual stress and service performance; The Delphi method is used to optimize the arrangement and combination of characteristic parameters of microstructure, defect characteristics, and residual stress according to the service life weight factor of the adsorption device in a H2-rich environment to determine the life correlation factors of the adsorption device. Introducing a multi-modal testing standard suitable for adsorber life prediction under H2-rich alternating load conditions; Establishing a failure evaluation principle based on multimodal detection signals, wherein the failure evaluation principle is used to define a life threshold value and a life evaluation criterion of an adsorption device under H2-rich alternating load conditions; Based on the multimodal database, a deep learning model is used to integrate the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria to train a failure prediction model for an adsorption device and provide failure reminders.

2. The method according to claim 1, characterized in that The multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device. The method further includes: Preprocessing the microstructure data, defect feature data, and residual stress data to obtain multimodal preprocessing data; enhancing the multimodal preprocessed data and selecting an optimal splitting point; According to the law of environmental changes in a H2-rich environment and based on the optimal splitting point, a multi-type damage mode interaction mechanism model is constructed.

3. The method according to claim 2, characterized in that According to the law of environmental change in a H2-rich environment, a multi-type damage mode mutual construction mechanism model is constructed according to the optimal splitting point. The method includes: Train the KNN classifier and set the distance metric; Based on the distance measurement method, a distance distribution map is formulated; According to the distance distribution map, clustering characteristics under different damage modes are analyzed to determine multiple types of damage modes.

4. The method according to claim 3, characterized in that According to the distance distribution map, clustering characteristics under different damage modes are analyzed to determine multiple types of damage modes. The method includes: Introducing an incremental learning mechanism to allow the distance distribution map to be updated synchronously; Based on the distance distribution map, a regular calibration mechanism is established.

5. The method according to claim 4, characterized in that Based on the distance distribution map, a regular calibration mechanism is established, the method comprising: Obtaining a calibration period of the periodic calibration mechanism; During the calibration period, microstructure data marked by time stamps, defect characteristic data marked by time stamps, and residual stress data marked by time stamps are collected as calibration samples; The KNN classifier is retrained using the calibration samples to obtain a calibrated distance distribution map.

6. The method according to claim 5, characterized in that Retraining the KNN classifier using the calibration sample, the method further comprising: Setting performance evaluation indicators, wherein the performance evaluation indicators include accuracy, recall rate, and F1 score; The performance evaluation index is used to evaluate the performance of the distance distribution map before and after calibration, and the calibration verification is performed by comparing the performance evaluation changes before and after calibration.

7. The method according to claim 1, characterized in that The method further comprises: Determining a data transmission network based on an integrated sensor monitoring module, wherein the integrated sensor monitoring module is used to collect microstructure data, defect feature data, and residual stress data of the adsorption device; An encrypted communication protocol is used to provide data security protection for the data transmission network.

8. An adsorption device failure prediction device integrating multimodal data, characterized in that: The device comprises: A database setting unit, the database setting unit is used to set a multimodal database, the multimodal database is used to store microstructure data, defect feature data, and residual stress data of the adsorption device, and establish a mapping relationship between the microstructure of the adsorption device and service performance, defect feature and service performance, and residual stress and service performance; The optimization calculation unit is used to use the Delphi method to optimize the arrangement and combination of characteristic parameters of microstructure, defect characteristics, and residual stress according to the service life weight factor of the adsorption device in a H2-rich environment, and determine the life correlation factors of the adsorption device; Standard introduction unit, which is used to introduce multi-modal test standards suitable for adsorber life prediction under H2-rich alternating load conditions; An evaluation principle establishment unit, the evaluation principle establishment unit is used to establish a failure evaluation principle based on a multimodal detection signal, the failure evaluation principle is used to define a life threshold value and a life evaluation criterion of an adsorption device under H2-rich alternating load conditions; A failure reminder unit is used to train a failure prediction model of an adsorption device based on the multimodal database, using a deep learning model, integrating the multimodal detection test standards, life-related factors, life threshold values ​​and life evaluation criteria, and performing failure reminders.

Images