A membrane structure abnormality prediction method and system based on artificial intelligence

Abstract

The present invention relates to the field of artificial intelligence technology, and provides a membrane structure abnormality prediction method and system based on artificial intelligence. By comprehensively using technical means such as deep residual network, monitoring knowledge integration and updating, and abnormal risk prediction, a comprehensive, real-time and accurate prediction of membrane structure performance is achieved, providing a strong guarantee for the safe operation and maintenance of membrane structures.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based membrane structure abnormality prediction method and system. Background Art

[0002] As one of the most representative architectural forms in the 21st century, membrane structures have the advantages of large span, economy, artistry, safety, and short construction period. However, with the development of time and aging of the structure, membrane structures will have problems such as deformation and damage, and the requirements for their normal use and maintenance technology are very high. If potential structural problems cannot be discovered in time, accidents may occur, affecting their safety performance and normal use functions, and increasing operating and management costs.

[0003] In view of the above problems, it is necessary to conduct real-time abnormal monitoring of the built membrane structure to detect problems in time, but manual monitoring is time-consuming and labor-intensive, and its accuracy and real-time performance are difficult to guarantee. Summary of the invention

[0004] In order to improve the above problems, the present invention provides a membrane structure abnormality prediction method and system based on artificial intelligence.

[0005] In a first aspect, an embodiment of the present invention provides a membrane structure abnormality prediction method based on artificial intelligence, which is applied to a membrane structure abnormality prediction system. The method includes:

[0006] Acquire structural state sensor monitoring data of the membrane structure to be monitored and at least one environmental state sensor monitoring data;

[0007] Performing X-order monitoring knowledge mining on the structural state sensor monitoring data through a first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data through the first artificial intelligence algorithm to obtain a second sensor monitoring knowledge relationship network set, wherein the first artificial intelligence algorithm is a deep residual network, the first sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, and the second sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, where X is a positive integer;

[0008] Integrating the monitoring knowledge of the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set to obtain a first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored;

[0009] updating monitoring knowledge according to the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set by a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set;

[0010] Abnormal risk prediction is performed based on the first linkage sensor monitoring knowledge relationship network set and the second linkage sensor monitoring knowledge relationship network set to obtain an abnormal risk prediction viewpoint of the membrane structure to be monitored.

[0011] Preferably, the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set are integrated to obtain the first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored, which includes:

[0012] Determining a feature enhancement factor according to a first sensor monitoring knowledge relationship network and a second sensor monitoring knowledge relationship network, wherein the first sensor monitoring knowledge relationship network and the second sensor monitoring knowledge relationship network have consistent feature dimensions, the first sensor monitoring knowledge relationship network is included in the first sensor monitoring knowledge relationship network set, and the second sensor monitoring knowledge relationship network is included in the second sensor monitoring knowledge relationship network set;

[0013] Determine the splicing coefficient by combining the second sensor monitoring knowledge relationship network with the feature enhancement factor to obtain the knowledge splicing coefficient;

[0014] Splicing the knowledge splicing coefficient with the first sensor monitoring knowledge relationship network to obtain a first linkage sensor monitoring knowledge relationship network, wherein the first linkage sensor monitoring knowledge relationship network belongs to the first linkage sensor monitoring knowledge relationship network set;

[0015] The first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set are sequentially walked to obtain a first linkage sensor monitoring knowledge relationship network set.

[0016] Preferably, the monitoring knowledge is updated according to the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set by a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set including:

[0017] Respectively optimizing the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network to obtain a first sensor monitoring optimization knowledge relationship network and a second sensor monitoring optimization knowledge relationship network, wherein the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network have consistent feature dimensions, the third sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the fourth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm;

[0018] Integrating monitoring knowledge on the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network to obtain a first linkage sensor monitoring optimization knowledge relationship network;

[0019] performing monitoring knowledge updating according to the first linkage sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring knowledge relationship network, wherein the second linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the second linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the first artificial intelligence algorithm at the Xth order;

[0020] Respectively optimizing the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network to obtain a third sensor monitoring optimization knowledge relationship network and a fourth sensor monitoring optimization knowledge relationship network, wherein the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network have consistent feature dimensions, the fifth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X-order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the sixth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm;

[0021] Integrating monitoring knowledge on the third sensor monitoring optimization knowledge relationship network and the fourth sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring optimization knowledge relationship network;

[0022] performing monitoring knowledge updating according to the second linkage sensor monitoring optimization knowledge relationship network to obtain a third linkage sensor monitoring knowledge relationship network, wherein the third linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the third linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the X-1th order of the first artificial intelligence algorithm;

[0023] Through X rounds of cyclic processing, the second linkage sensor monitoring knowledge relationship network set is obtained.

[0024] Preferably, the method further comprises:

[0025] Acquire an algorithm debugging sample set and an artificial intelligence algorithm to be debugged, wherein the algorithm debugging sample set includes active sensor monitoring data samples and environmental state sensor monitoring data;

[0026] Performing X-order monitoring knowledge mining on the active sensor monitoring data sample based on the first artificial intelligence algorithm to obtain X sensor monitoring knowledge relationship networks of the active sensor monitoring data sample;

[0027] Based on the first artificial intelligence algorithm, the environmental state sensor monitoring data is subjected to X-order monitoring knowledge mining to obtain X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data;

[0028] Performing monitoring knowledge integration on the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data to obtain X linkage sensor monitoring knowledge relationship network samples;

[0029] Transmitting the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data into the artificial intelligence algorithm to be debugged to update the monitoring knowledge, so as to obtain X updated sensor monitoring knowledge relationship networks;

[0030] Performing abnormal risk prediction based on the X updated sensor monitoring knowledge relationship networks and the X linked sensor monitoring knowledge relationship network examples to obtain a debugging error variable;

[0031] The artificial intelligence algorithm to be debugged is debugged based on the debugging error variable to obtain the second artificial intelligence algorithm.

[0032] Preferably, the environmental state sensing monitoring data includes temperature and humidity state sensing monitoring data and ice and wind load sensing monitoring data of the membrane structure building sample, and the active sensing monitoring data sample is stress state sensing monitoring data of the membrane structure building sample.

[0033] Preferably, the method further comprises:

[0034] Obtain at least two basic normalized subnets of the first artificial intelligence algorithm, wherein the at least two basic normalized subnets have a one-to-one matching relationship with the active sensor monitoring data sample and the environmental state sensor monitoring data;

[0035] The at least two basic normalized distribution subnetworks are supervised and debugged based on the debugging error variable to obtain at least two target normalized distribution subnetworks, wherein the at least two target normalized distribution subnetworks have different algorithm model variables.

[0036] Preferably, the step of performing X-order monitoring knowledge mining on the structural state sensor monitoring data by a first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm to obtain a second sensor monitoring knowledge relationship network set includes:

[0037] Performing monitoring knowledge mining on the structural state sensor monitoring data through the first artificial intelligence algorithm and processing through the corresponding target normal distribution subnet to obtain the first sensor monitoring knowledge relationship network set;

[0038] The monitoring knowledge mining of the at least one environmental status sensor monitoring data is performed through the first artificial intelligence algorithm and processed through the corresponding target normal distribution subnet to obtain the second sensor monitoring knowledge relationship network set.

[0039] Preferably, the method further comprises:

[0040] A maintenance strategy for the membrane structure to be monitored is generated based on the abnormal risk prediction viewpoint.

[0041] In a second aspect, an embodiment of the present invention provides a membrane structure abnormality prediction system, comprising at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the method described in the first aspect.

[0042] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method described in the first aspect when running.

[0043] In terms of accuracy and real-time performance, the above technical solution has the following beneficial effects:

[0044] In terms of accuracy, the deep residual network (the first artificial intelligence algorithm) is used to perform X-order knowledge mining on the structural state and environmental state sensor monitoring data, which can deeply extract the sensor monitoring knowledge relationship network with multiple feature dimensions. This deep mining helps to more accurately capture the subtle changes in the membrane structure and its relationship with environmental factors.

[0045] In terms of real-time performance, the solution can quickly respond to changes in the state of the membrane structure by integrating and updating monitoring knowledge. The construction of the first and second linkage sensor monitoring knowledge relationship network sets enables the system to integrate new monitoring data in real time and continuously update and optimize the understanding of the membrane structure state.

[0046] In summary, this technical solution improves the accuracy and response speed of abnormal risk prediction of monitored membrane structures through precise data mining and real-time knowledge updating, which is of great significance for ensuring the safety and stability of membrane structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A flowchart of a membrane structure abnormality prediction method based on artificial intelligence provided in an embodiment of the present invention.

[0048] Figure 2 A schematic diagram of the structure of a membrane structure abnormality prediction system 200 provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to better understand the above technical scheme, the technical scheme of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical scheme of the present invention, rather than limitations on the technical scheme of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0050] Figure 1 A membrane structure abnormality prediction method based on artificial intelligence is shown, which is applied to a membrane structure abnormality prediction system. The method includes S110-S150.

[0051] S110, acquiring structural state sensing monitoring data of the membrane structure to be monitored and at least one piece of environmental state sensing monitoring data.

[0052] S120. Perform X-order monitoring knowledge mining on the structural state sensor monitoring data through a first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and perform X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data through the first artificial intelligence algorithm to obtain a second sensor monitoring knowledge relationship network set.

[0053] Among them, the first artificial intelligence algorithm is a deep residual network, the first sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, and the second sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, where X is a positive integer.

[0054] S130, integrating monitoring knowledge of the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set to obtain a first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored.

[0055] S140. Update monitoring knowledge based on the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set by using a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set.

[0056] S150: performing abnormal risk prediction based on the first linkage sensor monitoring knowledge relationship network set and the second linkage sensor monitoring knowledge relationship network set to obtain an abnormal risk prediction viewpoint of the membrane structure to be monitored.

[0057] In large public facilities, the membrane structure on the roof needs to be continuously monitored for health to ensure safety. This monitoring involves the assessment of the structural status and the possible impact of environmental factors on it. The following is a specific application scenario combining S110-S150, detailing how to perform this task through the membrane structure abnormality prediction system.

[0058] First, the membrane structure abnormality prediction system will obtain the structural state sensor monitoring data of the membrane structure to be monitored, such as stress, strain, displacement, etc., which are collected in real time by sensors arranged on the membrane structure. At the same time, the system will also obtain at least one environmental state sensor monitoring data, such as temperature, humidity, wind speed, etc., which are also collected in real time by sensors arranged in the surrounding environment of the membrane structure.

[0059] Next, the system will use the deep residual network, an artificial intelligence algorithm, to perform X-order monitoring knowledge mining on the collected structural state sensor monitoring data. This means that the system will analyze the deep relationships in the data and extract sensor monitoring knowledge relationship networks of X different feature dimensions, which can reflect the performance of the membrane structure under different states. Similarly, the system will also perform X-order monitoring knowledge mining on the environmental state sensor monitoring data to obtain another sensor monitoring knowledge relationship network of X different feature dimensions.

[0060] Then, the system integrates the two sets of sensor monitoring knowledge relationship network sets. The purpose of this step is to comprehensively consider the influence of structural state and environmental state to form a more comprehensive membrane structure performance evaluation model. Through this integration process, the system obtains the first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored.

[0061] Subsequently, the system will use the second artificial intelligence algorithm to update the monitoring knowledge based on the previous two sets of sensor monitoring knowledge relationship network sets. This update process can reflect the latest status of the membrane structure and changes in environmental factors in real time, thereby obtaining the second linkage sensor monitoring knowledge relationship network set. This set not only contains the wisdom of historical data, but also incorporates the dynamic changes of real-time data.

[0062] Finally, based on the first linkage sensor monitoring knowledge network set and the second linkage sensor monitoring knowledge network set, the system will predict abnormal risks. This prediction process will comprehensively consider the historical performance and current status of the membrane structure, as well as the influence of environmental factors, to obtain a predicted view on the abnormal risk of the membrane structure to be monitored. This view can provide decision support for facility managers, helping them to take necessary maintenance measures in a timely manner to ensure the safety and stability of the membrane structure.

[0063] Under another design idea, combined with the above S110-S150, an introduction to the technical solution for membrane structure health monitoring of a coke oven gas comprehensive utilization project (raw coal shed) is also provided.

[0064] Application scenario overview: In a coke oven gas comprehensive utilization project, the raw coal shed is a key facility, and the safety and stability of its membrane structure are crucial to the operation of the overall project. Considering that the raw coal shed membrane structure may be affected by various factors such as internal gas pressure and external climate, it is necessary to implement a membrane structure health monitoring system.

[0065] First, structural state sensors, such as stress sensors and displacement sensors, are arranged at key locations of the original coal shed membrane structure to obtain real-time structural state sensing monitoring data such as stress distribution and deformation of the membrane structure. At the same time, environmental state sensors, such as temperature and humidity sensors, wind speed and direction sensors, are set around the coal shed to collect at least one environmental state sensing monitoring data, such as temperature, humidity, wind speed, etc.

[0066] The deep residual network is used as the first artificial intelligence algorithm to perform X-order monitoring knowledge mining on the collected structural state sensor monitoring data. This means that the algorithm will deeply analyze the complex relationship between the data and extract sensor monitoring knowledge relationship networks of X different feature dimensions. These relationship networks can fully reflect the performance of the membrane structure under different states. Similarly, the environmental state sensor monitoring data is also mined for X-order monitoring knowledge to obtain sensor monitoring knowledge relationship networks of another X feature dimensions.

[0067] The first sensor monitoring knowledge relationship network set obtained by mining the structural state sensor monitoring data is integrated with the second sensor monitoring knowledge relationship network set obtained by mining the environmental state sensor monitoring data. The purpose of this step is to comprehensively consider the impact of both and form a more comprehensive membrane structure performance evaluation model, namely the first linkage sensor monitoring knowledge relationship network set.

[0068] As time goes by and the operating status of the coal shed changes, the monitoring data will also change. Therefore, it is necessary to use a second artificial intelligence algorithm to update the monitoring knowledge based on the previous first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set. This update process can reflect the latest status of the membrane structure and changes in environmental factors in real time, thereby obtaining the second linkage sensor monitoring knowledge relationship network set.

[0069] Finally, abnormal risk prediction is performed based on the first linkage sensor monitoring knowledge network set and the second linkage sensor monitoring knowledge network set. This prediction process will comprehensively consider the historical performance, current state and environmental factors of the membrane structure, and judge whether there is an abnormal risk of the membrane structure through algorithm analysis, and give corresponding prediction opinions. These prediction opinions can provide decision support for project managers, helping them to take necessary maintenance measures in time to ensure the safety and stability of the original coal shed membrane structure.

[0070] The following is an exemplary explanation of the technical terms involved in the above S110-S150.

[0071] The membrane structure to be monitored refers to the membrane structure that needs to be monitored for health and performance evaluation. Usually, such structures are widely used in various engineering fields, such as architecture, aerospace, energy, etc. Due to its light weight, flexibility and efficient space utilization, membrane structures are often used to construct large roofs, shading facilities, temporary buildings or flexible solar panels.

[0072] For example, in the design of the roof of a large venue, membrane structures are often used to achieve large-span coverage while maintaining the lightness of the building. This type of roof needs to be used as a "monitored membrane structure" and sensors are installed to monitor its stress state, deformation, and possible environmental impact in real time to ensure its safety during use.

[0073] Structural state sensing monitoring data refers to the data about the performance status of the membrane structure itself collected by sensors installed on the membrane structure. These data usually include stress distribution, strain, displacement, vibration frequency, etc., which are used to evaluate the integrity, stability and safety of the membrane structure.

[0074] For example, in an exhibition hall built with membrane structure, engineers installed displacement sensors at key locations to monitor the deformation of the membrane structure under wind and snow loads. The displacement data collected by these sensors belong to "structural state sensing monitoring data", which can help engineers understand the deformation behavior of the membrane structure in real time and take measures to prevent structural damage when necessary.

[0075] Environmental state sensing monitoring data refers to data about external environmental conditions collected by sensors placed in the environment surrounding the membrane structure. These data include temperature, humidity, wind speed, wind direction, air pressure, rainfall, etc., which are crucial for understanding how the external environment affects the performance of the membrane structure.

[0076] For example, in a membrane structure park pavilion in a coastal city, managers installed wind speed and direction sensors around the pavilion to study the impact of sea breeze on the stability of the membrane structure. The wind speed and direction data recorded by these sensors belong to "environmental state sensing monitoring data". By analyzing these data, managers can better understand the effect mode of sea breeze on membrane structures and optimize maintenance strategies accordingly.

[0077] In this embodiment, the first artificial intelligence algorithm refers to a deep residual network for processing membrane structure monitoring data. The algorithm deeply mines complex patterns and associations in the data through a multi-layer neural network structure.

[0078] X-order monitoring knowledge mining refers to the use of artificial intelligence algorithms to conduct multi-level and multi-dimensional in-depth analysis of monitoring data to extract valuable information or knowledge hidden in the data. Here, "X-order" represents the depth or level of mining, and X is a positive integer.

[0079] For example, in this embodiment, by performing X-order monitoring knowledge mining on the structural state sensing monitoring data and the environmental state sensing monitoring data of the membrane structure through a deep residual network, the deep relationship between the membrane structure performance, environmental changes and the membrane structure response can be extracted.

[0080] The first sensor monitoring knowledge relationship network set refers to a set of knowledge relationship networks obtained by mining structural state sensor monitoring data through artificial intelligence algorithms. These networks reveal the association and dependency between different parameters inside the membrane structure.

[0081] For example, in this embodiment, the first sensor monitoring knowledge relationship network set is a series of interrelated knowledge networks obtained by mining structural state data such as stress and strain of the membrane structure through a deep residual network.

[0082] The second sensor monitoring knowledge relationship network set refers to a set of knowledge relationship networks obtained by mining environmental state sensor monitoring data through artificial intelligence algorithms. These networks reflect the interaction and influence between external environmental factors and membrane structure performance.

[0083] For example, in this embodiment, the second sensor monitoring knowledge relationship network set is a series of interrelated knowledge networks obtained by mining environmental status data such as temperature, humidity, and wind speed through a deep residual network.

[0084] Deep Residual Network (ResNet) is a deep learning model that solves the gradient vanishing and model degradation problems of deep neural networks during training by introducing residual learning. ResNet allows the network to learn effectively even when it is very deep.

[0085] For example, in this embodiment, a deep residual network is used as a first artificial intelligence algorithm to perform deep mining and analysis on the monitoring data of the membrane structure.

[0086] A characteristic dimension refers to a quantity or indicator used to describe a certain aspect of the characteristics of data or an object. In multidimensional data analysis, each characteristic dimension represents a different observation angle or data attribute.

[0087] For example, in this embodiment, the structural state sensing monitoring data of the membrane structure may include multiple characteristic dimensions, such as stress, strain, displacement, etc.; similarly, the environmental state sensing monitoring data may also include multiple characteristic dimensions such as temperature, humidity, wind speed, etc.

[0088] Sensor monitoring knowledge relationship network refers to a form of knowledge representation obtained by mining and analyzing sensor monitoring data through artificial intelligence algorithms. It is usually presented in the form of a network structure, where nodes represent different data features or parameters, and edges represent the relationship or dependency between these features or parameters.

[0089] For example, in this embodiment, the sensor monitoring knowledge relationship network can reveal the complex relationships and interactions between membrane structure performance parameters and between these parameters and external environmental factors. These relationship networks help to more fully understand the performance and behavior of membrane structures.

[0090] Monitoring knowledge integration refers to the process of integrating and fusing monitoring knowledge from different sources, types or time points. This process aims to form a more comprehensive and consistent knowledge system to facilitate subsequent analysis, decision-making or application.

[0091] For example, in this case, monitoring knowledge integration is to integrate the sensor monitoring knowledge network set about the structural state of the membrane structure (reflecting the internal performance of the membrane structure) and the sensor monitoring knowledge network set about the environmental state (reflecting the impact of the external environment on the membrane structure). By integrating these two parts of knowledge, a more comprehensive membrane structure performance evaluation model can be obtained.

[0092] The first linkage sensor monitoring knowledge relationship network set refers to a set of interrelated and mutually influential knowledge relationship networks obtained after completing the monitoring knowledge integration. These networks not only reflect the information of a single sensor or data source, but also reveal the linkage relationship and joint action between them.

[0093] For example, in this case, the first linkage sensor monitoring knowledge network set is the result of the integration of structural state sensor monitoring knowledge and environmental state sensor monitoring knowledge. It shows the changes and responses of membrane structure performance under different environmental conditions, providing an important basis for predicting abnormal risks of membrane structures.

[0094] Monitoring knowledge updating refers to the process of revising, supplementing or replacing existing monitoring knowledge as time goes by and new monitoring data is generated. This process aims to maintain the timeliness and accuracy of monitoring knowledge to adapt to the dynamically changing environment and object status.

[0095] For example, in this case, as the membrane structure is used for a longer time and the external environment changes, its performance may change. Therefore, it is necessary to update the existing sensor monitoring knowledge network with new monitoring data to reflect the latest status of the membrane structure and changes in environmental factors.

[0096] The second linkage sensor monitoring knowledge relationship network set refers to a set of new, interrelated knowledge relationship networks obtained after completing the monitoring knowledge update. These networks not only contain the information of the original monitoring knowledge, but also incorporate new data and knowledge, thus providing a more comprehensive and accurate evaluation of membrane structure performance.

[0097] For example, in this case, the second linkage sensor monitoring knowledge network set is the result of updating the first linkage sensor monitoring knowledge network set and new monitoring data. It provides the latest information on the performance of the membrane structure, providing strong support for predicting abnormal risks of the membrane structure and formulating maintenance strategies.

[0098] Abnormal risk prediction refers to the process of predicting and evaluating possible abnormal situations or risk events by using monitoring knowledge, data analysis models, algorithms and other technical means. This prediction aims to discover potential problems in advance so that preventive measures or response strategies can be taken in time to reduce the losses caused by risks.

[0099] For example, in the field of membrane structure monitoring, abnormal risk prediction may involve predicting the performance changes of membrane structures under extreme weather conditions. By analyzing historical data, current structural state sensor monitoring data, and environmental state sensor monitoring data, combined with artificial intelligence algorithms, it is possible to predict abnormal conditions that may occur in membrane structures in upcoming storms, such as excessive deformation, damage, etc., so that reinforcement or other countermeasures can be taken in advance.

[0100] Abnormal risk prediction viewpoints refer to the theories, assumptions, models or analytical methods used in making abnormal risk predictions. These viewpoints reflect the forecaster's understanding and knowledge of risk sources, risk transmission mechanisms and risk impacts, and are an important basis for guiding the prediction process and explaining the prediction results.

[0101] For example, in the prediction of membrane structure abnormality risk, a common prediction viewpoint is based on the theoretical model of structural dynamics and material mechanics. This viewpoint believes that the performance change of membrane structure is mainly affected by its material properties, geometric shape and external loads (such as wind, snow, etc.). Therefore, in the prediction process, the focus will be on the changes in these factors and their interactions, and the possible abnormalities of membrane structure will be predicted by establishing corresponding mathematical models and simulation analysis. This prediction viewpoint provides an important basis for formulating targeted risk prevention and control measures.

[0102] The beneficial effect of the embodiment of the present invention is that it provides a comprehensive and efficient membrane structure monitoring and abnormal risk prediction method. First, by obtaining the structural state sensor monitoring data of the membrane structure to be monitored and at least one environmental state sensor monitoring data, a comprehensive understanding of the performance of the membrane structure to be monitored is ensured, while taking into account the influence of external environmental factors. This step lays a solid foundation for subsequent data mining and knowledge integration.

[0103] Next, the deep residual network is used as the first artificial intelligence algorithm to perform X-order monitoring knowledge mining on the structural state sensor monitoring data and the environmental state sensor monitoring data. The application of the deep residual network effectively solves the gradient vanishing and model degradation problems of traditional neural networks in deep learning, making the mining results more accurate and in-depth. Through this step, the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set containing X different feature dimensions are obtained, which provide rich materials for the subsequent monitoring knowledge integration and update.

[0104] In the monitoring knowledge integration stage, the first sensor monitoring knowledge relationship network set is combined with the second sensor monitoring knowledge relationship network set to form the first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored. This step realizes the organic integration of structural state and environmental state knowledge, making it possible to more comprehensively understand the performance of the membrane structure and its interaction with the external environment.

[0105] In addition, the monitoring knowledge of the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set is updated through the second artificial intelligence algorithm to obtain the second linkage sensor monitoring knowledge relationship network set. This step ensures the timeliness and accuracy of the monitoring knowledge, so that changes in membrane structure performance and changes in the external environment can be responded to in a timely manner.

[0106] Finally, based on the first linkage sensor monitoring knowledge relationship network set and the second linkage sensor monitoring knowledge relationship network set, abnormal risk prediction was performed to obtain the abnormal risk prediction view of the membrane structure to be monitored. This step provides strong support for preventing possible abnormal situations in the membrane structure, helps to reduce the losses caused by risks, and improves the safety and reliability of the membrane structure.

[0107] In summary, the embodiments of the present invention achieve a comprehensive, real-time and accurate prediction of membrane structure performance by comprehensively using deep residual networks, monitoring knowledge integration and updating, and abnormal risk prediction, providing a strong guarantee for the safe operation and maintenance of membrane structures.

[0108] In some other possible application scenarios, in view of the shortcomings of existing membrane structure monitoring technology, the embodiments of the present invention also propose an innovative real-time monitoring and early warning platform for membrane structure safety. The platform integrates advanced sensor networks, data acquisition and storage technology, artificial intelligence algorithms and visual interfaces, aiming to achieve comprehensive real-time monitoring and dynamic early warning of membrane structure stress, displacement, deformation and the environmental conditions in which it is located.

[0109] In terms of data collection, the embodiments of the present invention use devices such as tensiometers, strain gauges, and acceleration sensors to capture key structural state data such as internal tension, cable force, and displacement of the membrane structure in real time. At the same time, environmental monitoring equipment such as anemometers, temperature and humidity sensors are combined to obtain environmental state data such as temperature, humidity, and wind speed of the surrounding environment of the membrane structure. These data provide a solid foundation for subsequent safety assessments and abnormal risk predictions.

[0110] In terms of data processing and analysis, the embodiment of the present invention uses a deep residual network as the core artificial intelligence algorithm to perform deep mining on the acquired structural state sensor monitoring data and environmental state sensor monitoring data. Through X-order monitoring knowledge mining, it is possible to extract the complex patterns and associations hidden in the data, thereby forming a sensor monitoring knowledge relationship network set containing multiple feature dimensions. These relationship network sets not only reveal the relationship between the internal performance parameters of the membrane structure, but also reflect their interaction with external environmental factors.

[0111] In order to more comprehensively understand the performance of the membrane structure, the embodiment of the present invention further integrates the structural state sensor monitoring knowledge relationship network set with the environmental state sensor monitoring knowledge relationship network set to form a first linkage sensor monitoring knowledge relationship network set. This step realizes the organic integration of structural state and environmental state knowledge, providing a more comprehensive perspective for membrane structure performance evaluation.

[0112] In addition, the embodiment of the present invention also introduces a second artificial intelligence algorithm to update the monitoring knowledge. As time goes by and new monitoring data is generated, the original sensor monitoring knowledge relationship network set needs to be continuously adjusted and optimized. Through the second artificial intelligence algorithm, the original knowledge relationship network set can be updated according to the latest monitoring data, thereby obtaining a second linkage sensor monitoring knowledge relationship network set. This step ensures the timeliness and accuracy of the monitoring knowledge.

[0113] Finally, based on the first linkage sensor monitoring knowledge relationship network set and the second linkage sensor monitoring knowledge relationship network set, the embodiment of the present invention performs abnormal risk prediction. By comprehensively analyzing the structural state and environmental state data of the membrane structure, it is possible to predict abnormal conditions or risk events that may occur in the membrane structure, and propose corresponding early warning information and response measures. This step provides strong support for preventing the occurrence of membrane structure safety accidents.

[0114] In summary, the embodiments of the present invention realize real-time monitoring and dynamic early warning of the safety of membrane structures by comprehensively using technical means such as sensor networks, artificial intelligence algorithms, and visual interfaces. This innovative platform not only improves the safety and reliability of membrane structures, but also reduces maintenance costs and manpower consumption, and has good social and economic benefits.

[0115] In another possible application scenario, for a specific membrane structure of a coke oven gas comprehensive utilization project - raw coal shed, the embodiment of the present invention proposes an intelligent monitoring method, which aims to comprehensively and real-time evaluate the performance status of the structure and the impact of its environment on it. The membrane structure has a wide span and high height, and the requirements for its safety are extremely strict. Therefore, it is very important to monitor it in all aspects.

[0116] First, in order to obtain the structural status data of the membrane structure, strain gauges were arranged at the bottom of the periphery to measure the tension of the membrane. Considering the convenience of construction and the accuracy of measurement, a strain gauge was arranged every 20 meters or so, a total of 40, to ensure comprehensive monitoring of the tension of the membrane structure. At the same time, in order to monitor the changes in cable force, vibrating wire strain gauges were selected and arranged at the cable node positions at the bottom of the periphery of the membrane structure at the same spacing, a total of 40.

[0117] In order to gain a deeper understanding of the overall vibration characteristics and modes of the membrane structure and obtain data on displacement changes, 27 acceleration sensors were evenly arranged on the structure, approximately one every 30 meters. The arrangement of these sensors not only covers the main areas of the membrane structure, but also ensures the accuracy and representativeness of the data.

[0118] In addition, considering the impact of environmental factors on the performance of membrane structures, wind speed and wind pressure sensors were arranged on the four sides and top of the structure, 15 each, a total of 30, to monitor wind changes in real time. At the same time, in order to understand the temperature and humidity differences between the inside and outside of the membrane structure, 16 temperature and humidity sensors were arranged, 8 each on the inside and outside.

[0119] After obtaining these data, the deep residual network, an advanced artificial intelligence algorithm, is used to mine the data for X-order monitoring knowledge. By mining the structural state sensor monitoring data, a first sensor monitoring knowledge relationship network set containing X different feature dimensions is obtained; similarly, the mining of the environmental state sensor monitoring data also obtains a second sensor monitoring knowledge relationship network set. These two sets provide rich information on the association between membrane structure performance and environmental factors.

[0120] Next, the two sets are integrated for monitoring knowledge, and the first linkage sensor monitoring knowledge network set of the membrane structure to be monitored is obtained. This step not only realizes the organic integration of structural state and environmental state knowledge, but also provides a more comprehensive perspective for membrane structure performance evaluation.

[0121] As time goes by and new monitoring data is generated, the second artificial intelligence algorithm is used to update the original sensor monitoring knowledge relationship network set to obtain the second linkage sensor monitoring knowledge relationship network set. This step ensures the timeliness and accuracy of the monitoring knowledge, so that the latest performance status of the membrane structure can be grasped in real time.

[0122] Finally, based on the two linked sensor monitoring knowledge network sets, abnormal risk prediction was performed. By comprehensively analyzing the structural state and environmental state data of the membrane structure, it is possible to predict abnormal conditions or risk events that may occur in the membrane structure, such as excessive deformation and rupture of the membrane, so that preventive measures can be taken in time to ensure the safe operation of the membrane structure.

[0123] In summary, the intelligent monitoring method of the embodiment of the present invention realizes comprehensive and real-time monitoring and abnormal risk prediction of the raw coal shed membrane structure of a coke oven gas comprehensive utilization project by comprehensively using multiple sensors and artificial intelligence algorithms, providing strong technical support for ensuring the safety and reliability of the structure.

[0124] In some optional embodiments, the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set are integrated to obtain the first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored, that is, S130, including S131-S134.

[0125] S131. Determine a feature enhancement factor based on a first sensory monitoring knowledge relationship network and a second sensory monitoring knowledge relationship network, wherein the first sensory monitoring knowledge relationship network and the second sensory monitoring knowledge relationship network have consistent feature dimensions, the first sensory monitoring knowledge relationship network is included in the first sensory monitoring knowledge relationship network set, and the second sensory monitoring knowledge relationship network is included in the second sensory monitoring knowledge relationship network set.

[0126] S132. Determine a splicing coefficient by combining the second sensor monitoring knowledge relationship network with the feature enhancement factor to obtain a knowledge splicing coefficient.

[0127] S133. Splicing the knowledge splicing coefficient with the first sensor monitoring knowledge relationship network to obtain a first linkage sensor monitoring knowledge relationship network, wherein the first linkage sensor monitoring knowledge relationship network belongs to the first linkage sensor monitoring knowledge relationship network set.

[0128] S134. Walk the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set in sequence to obtain a first linkage sensor monitoring knowledge relationship network set.

[0129] In some optional embodiments, how to integrate the monitoring knowledge of the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set to obtain the first linkage sensor monitoring knowledge relationship network set (ie, S130) of the membrane structure to be monitored can be further refined into the following S131 to S134.

[0130] In S131, the system will first determine a feature enhancement factor based on the first sensor monitoring knowledge relationship network and the second sensor monitoring knowledge relationship network. The first sensor monitoring knowledge relationship network and the second sensor monitoring knowledge relationship network here have consistent feature dimensions, which are mined from structural state sensor monitoring data and environmental state sensor monitoring data respectively. The role of the feature enhancement factor is to strengthen those features that have an important impact on the performance of the membrane structure in the subsequent knowledge integration process, so that the integrated knowledge relationship network is more accurate and targeted.

[0131] Next, in S132, the system will determine the splicing coefficient by combining the second sensor monitoring knowledge relationship network and the feature enhancement factor. The splicing coefficient is a key parameter that determines the proportion or method in which knowledge from different sources (i.e., structural state knowledge and environmental state knowledge) should be integrated during the knowledge integration process. By comprehensively considering the information of the second sensor monitoring knowledge relationship network and the feature enhancement factor, the system can calculate an optimal splicing coefficient to ensure that the integrated knowledge relationship network can fully reflect the actual performance status of the membrane structure.

[0132] Then, in S133, the system will splice the calculated knowledge splicing coefficient with the first sensor monitoring knowledge relationship network to obtain the first linkage sensor monitoring knowledge relationship network. In this process, the system will weight each feature in the first sensor monitoring knowledge relationship network according to the splicing coefficient, and merge the weighted result with the second sensor monitoring knowledge relationship network. The first linkage sensor monitoring knowledge relationship network obtained in this way not only contains the information of the structural state, but also integrates the influence of the environmental state, so it can more comprehensively and accurately reflect the performance state of the membrane structure.

[0133] Finally, in S134, the system will traverse all elements in the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set in a certain order, and repeat the operations of S131 to S133 until a complete first linkage sensor monitoring knowledge relationship network set is obtained. Each element in this set is a linkage sensor monitoring knowledge relationship network that integrates structural state and environmental state information, and together they constitute a comprehensive and in-depth understanding of the performance state of the membrane structure to be monitored. In this way, the system can realize real-time, dynamic monitoring and early warning of the membrane structure, providing strong support for ensuring its safety and reliability.

[0134] Under an alternative design idea, the monitoring knowledge is updated according to the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set through a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set, that is, S140, including S141-S147.

[0135] S141. Optimize the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network respectively to obtain a first sensor monitoring optimization knowledge relationship network and a second sensor monitoring optimization knowledge relationship network, wherein the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network have consistent feature dimensions, the third sensor monitoring knowledge relationship network is the sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the fourth sensor monitoring knowledge relationship network is the sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm.

[0136] S142: Integrate monitoring knowledge of the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network to obtain a first linkage sensor monitoring optimization knowledge relationship network.

[0137] S143. Update monitoring knowledge based on the first linkage sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring knowledge relationship network, wherein the second linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the second linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the Xth order of the first artificial intelligence algorithm.

[0138] S144. Optimize the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network respectively to obtain a third sensor monitoring optimization knowledge relationship network and a fourth sensor monitoring optimization knowledge relationship network, wherein the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network have consistent feature dimensions, the fifth sensor monitoring knowledge relationship network is the sensor monitoring knowledge relationship network obtained by performing X-th order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the sixth sensor monitoring knowledge relationship network is the sensor monitoring knowledge relationship network obtained by performing X-th order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm.

[0139] S145. Integrate monitoring knowledge on the third sensor monitoring optimization knowledge relationship network and the fourth sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring optimization knowledge relationship network.

[0140] S146. Update monitoring knowledge based on the second linkage sensor monitoring optimization knowledge relationship network to obtain a third linkage sensor monitoring knowledge relationship network, wherein the third linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the third linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the X-1th order of the first artificial intelligence algorithm.

[0141] S147. Perform X rounds of loop processing to obtain the second linkage sensor monitoring knowledge relationship network set.

[0142] In an alternative design concept, the process of the system executing monitoring knowledge update (ie, S140) can be subdivided into the following S141 to S147.

[0143] First, in S141, the system optimizes the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network respectively. The third sensor monitoring knowledge relationship network here is obtained by the first artificial intelligence algorithm performing X+1-order monitoring knowledge mining on the structural state sensor monitoring data, while the fourth sensor monitoring knowledge relationship network is obtained by the same algorithm performing X+1-order monitoring knowledge mining on the environmental state sensor monitoring data. These two relationship networks have consistent feature dimensions. Through optimization processing, the system obtains the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network. These two optimized relationship networks more accurately reflect the changes in structural state and environmental state.

[0144] Next, in S142, the system integrates the monitoring knowledge of the first sensor monitoring optimization knowledge relationship network with the second sensor monitoring optimization knowledge relationship network to obtain a first linkage sensor monitoring optimization knowledge relationship network. This integration process integrates the information of the structural state and the environmental state, so that the first linkage sensor monitoring optimization knowledge relationship network can more comprehensively reflect the performance state of the membrane structure.

[0145] Then, in S143, the system updates the monitoring knowledge based on the first linkage sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring knowledge relationship network. This update process is to refine and improve the linkage sensor monitoring knowledge relationship network while keeping the characteristic dimension unchanged. The second linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and its characteristic dimension is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the first artificial intelligence algorithm at the Xth order.

[0146] In S144, the system performs network optimization processing again, but this time it is for the fifth sensor monitoring knowledge network and the sixth sensor monitoring knowledge network. These two networks are obtained by the first artificial intelligence algorithm performing X-order monitoring knowledge mining on the structural state sensor monitoring data and the environmental state sensor monitoring data. Through the optimization processing, the system obtains the third sensor monitoring optimization knowledge network and the fourth sensor monitoring optimization knowledge network.

[0147] Subsequently, in S145, the system integrates the third sensor monitoring optimization knowledge relationship network with the fourth sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring optimization knowledge relationship network. This integration process also integrates the information of the structural state and the environmental state, but the difference is that it reflects the X-th order monitoring knowledge.

[0148] Then, in S146, the system updates the monitoring knowledge based on the second linkage sensor monitoring optimization knowledge relationship network to obtain a third linkage sensor monitoring knowledge relationship network. This update process is also carried out under the premise of keeping the characteristic dimension unchanged. The third linkage sensor monitoring knowledge relationship network is also included in the second linkage sensor monitoring knowledge relationship network set, and its characteristic dimension is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the X-1th order of the first artificial intelligence algorithm.

[0149] Finally, in S147, the system optimizes, integrates and updates the sensor monitoring knowledge relationship network from the Xth order to the first order through X rounds of cyclic processing, and finally obtains a complete set of the second linkage sensor monitoring knowledge relationship network. Each element in this set is a linkage sensor monitoring knowledge relationship network that integrates structural state and environmental state information, and together they constitute a comprehensive, in-depth and dynamic understanding of the performance state of the membrane structure to be monitored. In this way, the system can achieve real-time, accurate monitoring and early warning of the membrane structure, providing strong data support for ensuring its safety and reliability.

[0150] As for S141, in order to have a deeper understanding of the performance status of the membrane structure, the system needs to optimize the knowledge relationship network mined from the structural state sensor monitoring data and the environmental state sensor monitoring data. In this process, the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network are two key elements.

[0151] The third sensor monitoring knowledge network is obtained by in-depth analysis of the structural state sensor monitoring data by the first artificial intelligence algorithm. Specifically, the algorithm performs X+1-order monitoring knowledge mining on the structural state sensor monitoring data. This means that it explores possible complex patterns, correlations and trends in the data that may not have been discovered in the previous mining stage. Through this mining, it is possible to more accurately understand how the structural state affects the overall performance of the membrane structure.

[0152] At the same time, the fourth sensor monitoring knowledge relationship network is obtained by processing the environmental state sensor monitoring data through the same first artificial intelligence algorithm. Unlike structural state data, environmental state data provides valuable information about the environmental conditions of the membrane structure. Through the X+1th order monitoring knowledge mining, the subtle connection between environmental conditions and membrane structure performance can be captured.

[0153] It is worth noting that the third sensor monitoring knowledge network and the fourth sensor monitoring knowledge network have consistent characteristic dimensions. This means that when performing analysis and optimization, the system can uniformly process the information in these two networks to ensure that the data of structural status and environmental status are compared and integrated on the same dimension.

[0154] The process of network optimization involves many aspects, including removing redundant information, strengthening important features, correcting erroneous associations, etc. Through these optimization measures, the system can improve the accuracy and reliability of the knowledge network, thereby more accurately reflecting the performance status of the membrane structure.

[0155] Finally, the optimized third sensor monitoring knowledge network is transformed into the first sensor monitoring optimized knowledge network, and the fourth sensor monitoring knowledge network is transformed into the second sensor monitoring optimized knowledge network. These two optimized networks not only retain the key information in the original data, but also make the subsequent analysis and decision-making process more efficient and accurate by removing noise and redundant data.

[0156] As for S142, when comprehensively evaluating the performance status of membrane structures, the system needs to effectively integrate knowledge from different data sources. In this process, the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network play a key role.

[0157] The first sensor monitoring optimization knowledge network is optimized based on the structural state sensor monitoring data. It deeply reflects the relationship between various elements inside the membrane structure and their impact on the performance state. This network contains rich structural information and is an indispensable part of understanding the performance state of the membrane structure.

[0158] At the same time, the second sensor monitoring optimization knowledge network is optimized based on the environmental state sensor monitoring data. It reveals how environmental conditions interact with membrane structures and how this interaction affects the performance of membrane structures. This network provides a deep insight into the dynamic interaction between membrane structures and the external environment.

[0159] In order to integrate the information in these two knowledge networks, the system conducts monitoring knowledge integration. In this process, the system not only integrates the data in the two networks, but also ensures the accuracy and consistency of the information through a series of algorithms and technical means. This integration is not a simple data superposition, but a targeted information fusion based on a deep understanding of the internal logic of the two knowledge networks.

[0160] Through monitoring knowledge integration, the system obtains the first linkage sensor monitoring optimization knowledge relationship network. This new relationship network not only contains information on structural status and environmental status, but also organically combines them through specific algorithms and logic. In this way, the first linkage sensor monitoring optimization knowledge relationship network can provide a more comprehensive and in-depth perspective to examine the performance status of membrane structures. It not only reveals the impact of structural elements and environmental elements on the performance status, but also shows the complex interactions and joint influences between them.

[0161] In short, by integrating the monitoring knowledge of the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network, the system successfully constructed a more comprehensive and in-depth knowledge relationship network - the first linkage sensor monitoring optimization knowledge relationship network. This new relationship network provides a strong data support and analysis basis for the subsequent membrane structure performance status evaluation.

[0162] As for S143, after obtaining the first linkage sensor monitoring optimization knowledge relationship network, the system did not stop there, but further updated its monitoring knowledge to obtain more accurate and comprehensive membrane structure performance status information. This update process is based on the existing linkage sensor monitoring knowledge relationship network, combined with new data and analysis results, to refine and improve the knowledge.

[0163] Specifically, the system will identify possible new associations, trends or patterns based on the information in the knowledge relationship network optimized by the first linkage sensor monitoring, combined with other data sources and analysis models. These new discoveries may come from changes in structural or environmental states, or from new ways of interaction between them.

[0164] In the process of monitoring knowledge update, the system will pay special attention to those elements and relationships that are closely related to the performance status of membrane structures. These elements and relationships may include stress distribution, temperature changes, wind speed effects, etc., which are all key factors affecting the performance status of membrane structures. By deeply analyzing and updating these factors, the system can more accurately grasp the performance status of membrane structures and their changing trends.

[0165] After the monitoring knowledge is updated, the system obtains the second linkage sensor monitoring knowledge relationship network. This new relationship network retains the effective information in the first linkage sensor monitoring optimization knowledge relationship network while incorporating new knowledge and discoveries. Therefore, it can more comprehensively reflect the performance status of the membrane structure and its interaction with the external environment.

[0166] It is worth mentioning that the characteristic dimensions of the second linkage sensor monitoring knowledge relationship network are the same as those of the linkage sensor monitoring knowledge relationship network generated by the first artificial intelligence algorithm at the Xth stage. This means that in the process of updating knowledge, the system maintains the consistency and comparability of the data, ensuring that the new relationship network can be effectively compared and analyzed with the previous results.

[0167] In short, by updating the monitoring knowledge based on the first linkage sensor monitoring optimization knowledge network, the system successfully built a more accurate and comprehensive knowledge network - the second linkage sensor monitoring knowledge network. This new network provides a strong data support and analysis basis for the subsequent membrane structure performance status evaluation and early warning.

[0168] As for S144, in the process of continuous monitoring and evaluation of the performance status of the membrane structure, the system needs to continuously mine new knowledge networks from the structural state sensor monitoring data and the environmental state sensor monitoring data, and optimize these networks to improve the accuracy and reliability of the evaluation. This time, the system focuses on the fifth sensor monitoring knowledge network and the sixth sensor monitoring knowledge network.

[0169] The fifth sensor monitoring knowledge network is obtained by mining the X-order monitoring knowledge of the structural state sensor monitoring data through the first artificial intelligence algorithm. At this stage, the algorithm will deeply explore the inherent laws and correlations in the structural state data to reveal the interaction between the internal elements of the membrane structure and its impact on the performance state. This network provides an important perspective for understanding the performance state of the membrane structure from a structural perspective.

[0170] At the same time, the sixth sensor monitoring knowledge relationship network is obtained by mining the Xth-order monitoring knowledge of environmental state sensor monitoring data through the same artificial intelligence algorithm. The environmental state data contains rich information about the environmental conditions of the membrane structure, such as temperature, humidity, wind speed, etc. By mining this data, we can better understand how environmental conditions affect the performance of the membrane structure and the changing trend of this influence under different environmental conditions.

[0171] It is worth noting that the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network have consistent characteristic dimensions. This means that when optimizing the relationship network, the system can adopt a unified processing method to ensure that the data of structural state and environmental state are compared and integrated on the same dimension, thereby improving the accuracy and reliability of the optimization results.

[0172] The process of optimizing the relationship network is a complex and delicate process, which involves removing redundant information, strengthening important features, correcting erroneous associations, etc. Through these optimization measures, the system can improve the accuracy and reliability of the knowledge relationship network, making it more in line with the actual situation, and providing more accurate data support for subsequent analysis and decision-making.

[0173] After optimization, the fifth sensor monitoring knowledge network is transformed into the third sensor monitoring optimization knowledge network, while the sixth sensor monitoring knowledge network is transformed into the fourth sensor monitoring optimization knowledge network. These two new optimized knowledge networks not only retain the key information in the original data, but also make the subsequent analysis and decision-making process more efficient and accurate by removing noise and redundant data. At the same time, they also provide the system with a richer and deeper knowledge base, laying a solid foundation for the subsequent membrane structure performance status evaluation and early warning.

[0174] As for S145, when comprehensively evaluating the performance status of the membrane structure, the system needs to effectively integrate the knowledge from different data sources. This time, the focus of the integration is on the third sensor monitoring optimization knowledge relationship network and the fourth sensor monitoring optimization knowledge relationship network.

[0175] The third sensor monitoring optimization knowledge network is optimized based on the structural state sensor monitoring data. It reveals the complex connections between the elements inside the membrane structure, which are crucial for understanding the overall performance and stability of the membrane structure. This network provides a structural perspective to help gain insight into the response and behavior of the membrane structure under different conditions.

[0176] At the same time, the fourth sensor monitoring optimization knowledge network is optimized based on environmental state sensor monitoring data. It captures the subtle relationship between environmental conditions and membrane structure performance, and shows how external factors affect the stability and durability of membrane structures. This network provides an environmental factor perspective, enabling a better understanding of the performance of membrane structures in specific environments.

[0177] In order to integrate the information in these two knowledge networks, the system conducts monitoring knowledge integration. This process involves not only the integration of data, but also the use of a series of algorithms and technologies to ensure the accuracy and consistency of information. The goal of the integration is to create a more comprehensive and integrated knowledge network that can simultaneously reflect the impact of structural and environmental conditions on the performance of membrane structures.

[0178] Through monitoring knowledge integration, the system obtains the second linkage sensor monitoring optimization knowledge relationship network. This new relationship network integrates the key information in the third sensor monitoring optimization knowledge relationship network and the fourth sensor monitoring optimization knowledge relationship network, forming a more complete and in-depth knowledge base of membrane structure performance status. It not only includes the influence of structural elements and environmental elements on performance status, but also reveals their interaction and joint influence.

[0179] The generation of the second linkage sensor monitoring optimization knowledge network provides strong support for the comprehensive evaluation of the performance status of the membrane structure. It enables the system to more accurately predict the behavior of the membrane structure under different conditions, discover potential problems in a timely manner, and take corresponding measures to ensure the safety and stability of the membrane structure. At the same time, this new knowledge network also provides an important reference and basis for subsequent monitoring and early warning work.

[0180] As for S146, after obtaining the second linkage sensor monitoring optimization knowledge relationship network, in order to maintain continuous and accurate tracking of the performance status of the membrane structure, the system needs to continuously update the knowledge. This time, the update process is based on the second linkage sensor monitoring optimization knowledge relationship network, aiming to obtain more accurate and comprehensive performance status information.

[0181] Specifically, the system will deeply analyze the data and information in the second linkage sensor monitoring optimization knowledge relationship network to identify new trends, patterns or associations. These new discoveries may come from changes in structural states, fluctuations in environmental conditions, or new ways of interaction between them. In order to capture these subtle but important changes, the system will combine multiple data sources and advanced analytical models to make comprehensive judgments.

[0182] In the process of monitoring knowledge updates, the system pays special attention to those factors and relationships that are closely related to the performance status of membrane structures. These factors may include changes in stress distribution, signs of material aging, cumulative effects of environmental factors, etc. By deeply analyzing and updating these factors, the system can more accurately grasp the current performance status of membrane structures and predict their future change trends.

[0183] After a series of complex calculations and processing, the system obtains the third linkage sensor monitoring knowledge relationship network. This new knowledge relationship network inherits the effective information of the second linkage sensor monitoring optimization knowledge relationship network and incorporates new monitoring data and knowledge discovery. Therefore, it can more comprehensively reflect the performance status of the membrane structure and its interaction with the external environment.

[0184] It is worth mentioning that the characteristic dimension of the third linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the first artificial intelligence algorithm at the X-1th order. This design ensures the consistency and comparability of data in the knowledge update process, so that the new relationship network can be effectively connected and integrated with the previous results. In this way, the system can not only track the changes in the performance status of the membrane structure in real time, but also draw experience from historical data to provide more solid data support for future early warning and decision-making.

[0185] As for S147, in the process of membrane structure performance status monitoring and evaluation, in order to build a comprehensive and accurate second linkage sensor monitoring knowledge network set, the system needs to perform multiple rounds of loop processing. The purpose of this set is to gather linkage sensor monitoring knowledge at different stages and under different conditions, so as to provide a more macro and in-depth performance status analysis perspective.

[0186] Each round of cyclic processing includes in-depth mining and analysis of structural state sensor monitoring data and environmental state sensor monitoring data. Through the first artificial intelligence algorithm, the system continuously extracts useful information from these data, identifies key features and patterns, and then builds a linkage sensor monitoring knowledge relationship network that reflects the performance status of the membrane structure.

[0187] During the cyclic processing, the system will adjust and optimize the algorithm parameters according to the results of the previous round to ensure the accuracy and effectiveness of subsequent mining. At the same time, the knowledge relationship network obtained in each round of processing will be carefully reviewed and evaluated to ensure its quality and reliability. Only the knowledge relationship network that has been strictly screened and verified will be included in the second linkage sensor monitoring knowledge relationship network collection.

[0188] As the cycle progresses, the second linkage sensor monitoring knowledge network collection gradually becomes richer. It contains knowledge that reflects the performance status of the membrane structure from all angles and levels, forming a multi-dimensional and multi-level knowledge system. This collection not only provides comprehensive performance status information for the system, but also provides strong support for subsequent monitoring, early warning and decision-making.

[0189] After X rounds of cyclic processing, the system finally obtains a complete and accurate second linkage sensor monitoring knowledge network set. This set is the crystallization of the system's wisdom and experience. It brings together massive data and in-depth analysis, providing a solid guarantee for the long-term safe and stable operation of the membrane structure.

[0190] In some other optional embodiments, the method also includes S210-S270.

[0191] S210, obtaining an algorithm debugging sample set and an artificial intelligence algorithm to be debugged, wherein the algorithm debugging sample set includes active sensor monitoring data samples and environmental state sensor monitoring data.

[0192] S220: Perform X-order monitoring knowledge mining on the active sensor monitoring data sample based on the first artificial intelligence algorithm to obtain X sensor monitoring knowledge relationship networks of the active sensor monitoring data sample.

[0193] S230: Perform X-order monitoring knowledge mining on the environmental state sensor monitoring data based on the first artificial intelligence algorithm to obtain X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data.

[0194] S240: Perform monitoring knowledge integration on the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data to obtain X linkage sensor monitoring knowledge relationship network samples.

[0195] S250: Transmit the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data into the artificial intelligence algorithm to be debugged to update the monitoring knowledge, so as to obtain X updated sensor monitoring knowledge relationship networks.

[0196] S260: Perform abnormal risk prediction based on the X updated sensor monitoring knowledge relationship networks and the X linked sensor monitoring knowledge relationship network examples to obtain a debugging error variable.

[0197] S270. Debug the artificial intelligence algorithm to be debugged based on the debugging error variable to obtain the second artificial intelligence algorithm.

[0198] In some other more specific embodiments, the method is further refined into S210 to S270, and how these steps are implemented in the system will be explained in detail one by one below.

[0199] First, the system needs to obtain the algorithm debugging sample set and the artificial intelligence algorithm to be debugged. The algorithm debugging sample set here includes active sensor monitoring data samples and environmental state sensor monitoring data. Active sensor monitoring data samples refer to those data that can directly reflect the performance status of the membrane structure, such as stress, strain, etc.; while environmental state sensor monitoring data refers to environmental factor data that affect the performance of the membrane structure, such as temperature, humidity, wind speed, etc. These data samples will be used for subsequent knowledge mining and algorithm debugging.

[0200] Next, the system conducts X-order monitoring knowledge mining on the active sensor monitoring data samples based on the first artificial intelligence algorithm. The purpose of this step is to extract useful information and features from the active sensor monitoring data and build a sensor monitoring knowledge relationship network that reflects the performance status of the membrane structure. X-order monitoring knowledge mining means that the system will conduct multi-level in-depth analysis and mining of the data to obtain more comprehensive and refined knowledge.

[0201] At the same time, the system will also perform X-order monitoring knowledge mining on the environmental state sensor monitoring data based on the same first artificial intelligence algorithm. The purpose of this step is to extract information and features related to membrane structure performance from environmental state data and construct a sensor monitoring knowledge relationship network that reflects the impact of environmental factors on membrane structure performance.

[0202] After obtaining the sensor monitoring knowledge network of active sensor monitoring data samples and environmental state sensor monitoring data, the system will integrate them into monitoring knowledge. The purpose of this step is to effectively integrate knowledge from different data sources to form a more comprehensive and integrated linkage sensor monitoring knowledge network sample. These samples will serve as reference standards for subsequent algorithm debugging.

[0203] Then, the system will pass the sensor monitoring knowledge relationship network of active sensor monitoring data samples and environmental state sensor monitoring data to the artificial intelligence algorithm to be debugged for monitoring knowledge update. The purpose of this step is to use the algorithm to be debugged to further process and optimize these knowledge relationship networks to obtain updated sensor monitoring knowledge relationship networks. These updated relationship networks will reflect the algorithm's processing capabilities and effects on the current data.

[0204] Next, the system performs anomaly risk prediction based on the updated sensor monitoring knowledge network and the linked sensor monitoring knowledge network sample. The purpose of this step is to predict the anomalies and risks that may occur when the algorithm processes similar data in the future by comparing and analyzing the differences and consistency between the two networks. The output of this step is the debugging error variable, which reflects the performance and accuracy of the algorithm in its current state.

[0205] Finally, the system debugs the AI ​​algorithm to be debugged based on the debugging error variable. The purpose of this step is to make targeted adjustments and optimizations to the algorithm based on the feedback information obtained in the previous step to improve its processing performance and accuracy. After this step, the system will obtain the second AI algorithm, which is an optimized and improved version that can better adapt to and handle the tasks of membrane structure performance status monitoring and evaluation.

[0206] In some exemplary embodiments, the environmental state sensing monitoring data includes temperature and humidity state sensing monitoring data and ice and wind load sensing monitoring data of a membrane structure building sample, and the active sensing monitoring data sample is stress state sensing monitoring data of a membrane structure building sample.

[0207] In some exemplary embodiments, the environmental state sensing monitoring data processed by the system specifically includes the temperature and humidity state sensing monitoring data and ice and wind load sensing monitoring data of the membrane structure building sample. These data are essential for evaluating the performance of the membrane structure building under different environmental conditions. At the same time, the active sensing monitoring data samples focus on the stress state of the membrane structure building, which is a key indicator for evaluating its structural integrity and stability.

[0208] Specifically, the temperature and humidity state sensing monitoring data can reflect the performance changes of membrane structure buildings under different temperature and humidity conditions. For example, high temperature and high humidity environment may cause the membrane material to expand and deform, thereby affecting its overall stability. By continuously monitoring this data, the system can promptly detect potential performance problems and take corresponding measures for prevention and maintenance.

[0209] The ice and wind load sensing monitoring data focuses on the load of membrane structures under the action of ice, snow and wind. These data can directly reflect the load-bearing capacity and wind resistance of membrane structures. Under extreme weather conditions, such as blizzards or strong winds, membrane structures may be subjected to huge load pressure. By monitoring these data in real time, the system can evaluate the safety of membrane structures and issue early warning signals when necessary.

[0210] At the same time, as part of active sensing monitoring, stress state sensing monitoring data can directly reflect the stress distribution inside the membrane structure building. These data are essential for evaluating the structural integrity and stability of the membrane structure building. By monitoring the changes in stress state, the system can promptly detect potential stress concentration or damage areas and take corresponding measures to repair and reinforce them.

[0211] In summary, through the comprehensive application of these environmental conditions and active sensing monitoring data, the system can comprehensively evaluate the performance status of membrane structure buildings under different conditions. This not only helps to discover potential safety hazards in a timely manner, but also provides strong data support for subsequent maintenance and management.

[0212] In some further preferred embodiments, the method further includes S310 - S320.

[0213] S310. Obtain at least two basic normalized distribution subnets of the first artificial intelligence algorithm, wherein the at least two basic normalized distribution subnets have a one-to-one matching relationship with the active sensor monitoring data samples and the environmental state sensor monitoring data.

[0214] S320: Perform supervisory debugging on the at least two basic normalized distribution subnets based on the debugging error variable to obtain at least two target normalized distribution subnets, wherein the at least two target normalized distribution subnets have different algorithm model variables.

[0215] In some further preferred embodiments, the method is further refined to include S310 and S320. The following will explain in detail how these steps are implemented in the system one by one.

[0216] First, the system needs to obtain at least two basic normalized subnets of the first artificial intelligence algorithm. These basic normalized subnets are an important part of the first artificial intelligence algorithm, and they are responsible for processing the input active sensor monitoring data samples and environmental state sensor monitoring data. The "basic normalized subnets" mentioned here refer to those subnets that can convert input data into a normal distribution form. The purpose of doing this is to make the data more in line with the processing requirements of the algorithm and improve the accuracy and stability of the algorithm. At the same time, these basic normalized subnets have a one-to-one matching relationship with active sensor monitoring data samples and environmental state sensor monitoring data, which means that each subnet is specifically responsible for processing a specific data type.

[0217] Next, the system supervises and debugs at least two basic normalized subnets based on the debugging error variables. The purpose of this step is to make targeted adjustments and optimizations to these basic normalized subnets based on the debugging error variables obtained in the previous steps to improve their processing performance and accuracy. Supervised debugging is a commonly used algorithm optimization method that adjusts and improves the algorithm by comparing the difference between the actual output of the algorithm and the expected output. Here, the system adjusts the parameters and structure of the basic normalized subnets according to the size and direction of the debugging error variables so that their outputs are closer to the expected normal distribution form. After this step, the system will obtain at least two target normalized subnets, which are optimized and improved versions with different algorithm model variables, and can better adapt to and handle the tasks of membrane structure performance status monitoring and evaluation. These target normalized subnets will be used as an important part of the second artificial intelligence algorithm for subsequent monitoring knowledge updates and abnormal risk prediction tasks.

[0218] In some optional embodiments, the first artificial intelligence algorithm is used to perform X-order monitoring knowledge mining on the structural state sensor monitoring data to obtain a first sensor monitoring knowledge relationship network set, and the first artificial intelligence algorithm is used to perform X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data to obtain a second sensor monitoring knowledge relationship network set, that is, S120, including: performing monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm and processing it through the corresponding target normal distribution subnet to obtain the first sensor monitoring knowledge relationship network set; performing monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm and processing it through the corresponding target normal distribution subnet to obtain the second sensor monitoring knowledge relationship network set.

[0219] In some optional embodiments, when the system performs the tasks of performing X-order monitoring knowledge mining on structural state sensor monitoring data through a first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and performing X-order monitoring knowledge mining on at least one environmental state sensor monitoring data to obtain a second sensor monitoring knowledge relationship network set (i.e., step S120), this process actually includes two key processing links.

[0220] First, the system will use the first artificial intelligence algorithm to conduct in-depth monitoring knowledge mining on the structural state sensor monitoring data. The structural state sensor monitoring data here refers to those data that directly reflect the internal state of the membrane structure, such as stress, strain, displacement, etc. The purpose of mining is to extract useful information and features from these raw data to form a knowledge relationship network that can describe the performance state of the membrane structure. However, in this process, the system does not directly mine the data, but will first process it through the corresponding target normal distribution subnet. These target normal distribution subnets are pre-optimized and debugged. Their function is to convert the input structural state sensor monitoring data into a form that is more in line with the normal distribution, thereby improving the data processing efficiency and the accuracy of the algorithm. After being processed by the target normal distribution subnet, the system will then mine it through the first artificial intelligence algorithm, and finally obtain the first sensor monitoring knowledge relationship network set.

[0221] Secondly, the system will also use the same first artificial intelligence algorithm to perform monitoring knowledge mining on at least one environmental state sensor monitoring data. Environmental state sensor monitoring data refers to data on external environmental factors that affect the performance of the membrane structure, such as temperature, humidity, wind speed, wind direction, etc. These data also need to be processed by the target normal distribution subnet to be converted into a form that better meets the algorithm processing requirements. Then, the system uses the first artificial intelligence algorithm to mine the environmental state sensor monitoring data, extract information and features related to the performance of the membrane structure, and form a second sensor monitoring knowledge relationship network set.

[0222] In summary, through the processing of these steps, the system can more accurately extract useful information and features from the raw data, forming a comprehensive and detailed set of sensor monitoring knowledge relationship networks, providing strong data support for subsequent performance status evaluation and early warning.

[0223] In some optional embodiments, the method further includes S160.

[0224] S160: Generate a maintenance strategy for the membrane structure to be monitored based on the abnormal risk prediction viewpoint.

[0225] In some optional embodiments, the method of the system further includes S160, that is, generating a maintenance strategy for the membrane structure to be monitored based on the abnormal risk prediction point of view. This step is based on the monitoring and evaluation results of the performance status of the membrane structure in the previous step, especially the prediction of abnormal risks, to formulate specific maintenance measures and strategies.

[0226] First of all, it should be made clear that the abnormal risk prediction viewpoint is the judgment of potential abnormalities or risks made by the system through continuous monitoring and data analysis of the membrane structure. These judgments may be based on factors such as the changing trends of various sensor monitoring data, comparison with historical data, and hidden patterns or associations mined by artificial intelligence algorithms.

[0227] In S160, the system will formulate maintenance strategies based on these abnormal risk prediction viewpoints. Specifically, the system may consider the following aspects:

[0228] Nature and severity of risk: The system first evaluates the specific nature and possible severity of the consequences of the predicted abnormal risk. For example, some risks may only involve local areas of the membrane structure, while others may affect the stability of the entire structure;

[0229] Priority of maintenance measures: Based on the nature and severity of the risk, the system will determine the priority of different maintenance measures. For risks that may lead to serious consequences, the system will recommend emergency measures for intervention, while for lower-level risks, more moderate maintenance strategies may be adopted;

[0230] Allocation of maintenance resources: The system will also consider the resources available for maintenance, including human, material, and time factors. When formulating maintenance strategies, the system will strive to achieve the best maintenance results under limited resource conditions;

[0231] Specific implementation steps of maintenance measures: Finally, the system will generate a detailed maintenance strategy, including specific maintenance measures to be taken, implementation schedule, required resources, expected results, etc. These strategies can be operating instructions presented in the form of documents or automated maintenance instructions directly integrated into the membrane structure monitoring system.

[0232] By implementing S160, the system can provide comprehensive and specific guidance for the maintenance of membrane structures, thereby helping managers to promptly identify and resolve potential safety hazards and ensure the long-term stable operation of membrane structures.

[0233] In some independent embodiments, the maintenance strategy for the membrane structure to be monitored generated according to the abnormal risk prediction viewpoint as described in S160 includes: generating a vibration state maintenance plan sequence according to the abnormal risk prediction viewpoint, wherein the vibration state maintenance plan sequence includes Y vibration state maintenance plans with time series connection, and Y is an integer greater than or equal to 1; acquiring a deformation repair maintenance plan sequence according to the vibration state maintenance plan sequence, wherein the deformation repair maintenance plan sequence includes Y deformation repair maintenance plans with time series connection; based on the vibration state maintenance plan sequence, the first semantic recognition included in the maintenance strategy output network is used to generate a maintenance strategy output network. A vibration state semantic moment sequence is obtained through a separate branch, wherein the vibration state semantic moment sequence includes Y vibration state semantic moments; based on the deformation repair and maintenance plan sequence, a deformation repair semantic moment sequence is obtained through a second semantic recognition branch included in the maintenance strategy output network, wherein the deformation repair semantic moment sequence includes Y deformation repair semantic moments; based on the vibration state semantic moment sequence and the deformation repair semantic moment sequence, a feasibility weight corresponding to the vibration state maintenance plan sequence is obtained through a feasibility recognition branch included in the maintenance strategy output network; and a strategy execution order of the vibration state maintenance plan sequence is determined according to the feasibility weight.

[0234] In some independent embodiments, the system executes step S160, that is, generating a maintenance strategy for the membrane structure to be monitored based on the abnormal risk prediction point of view, and the process involves multiple subdivision steps and technical applications to ensure that the generated maintenance strategy is both effective and feasible.

[0235] First, the system generates a vibration state maintenance plan sequence based on the abnormal risk prediction point of view. This sequence contains Y vibration state maintenance plans with time sequence connection, where Y is an integer greater than or equal to 1. The time sequence connection means that these maintenance plans need to be executed in a specific order. The vibration state maintenance plan is a specific measure formulated for the abnormal risks that may occur in the membrane structure under the vibration state, such as adjusting the tension distribution of the membrane structure, strengthening key nodes, etc.

[0236] Next, the system will obtain a deformation repair maintenance plan sequence based on the vibration state maintenance plan sequence. This sequence also includes Y deformation repair maintenance plans with time sequence connection. The deformation repair maintenance plan is a specific measure formulated to address the abnormal risks that may occur when the membrane structure is deformed, such as repairing local damage, adjusting the supporting structure, etc. It should be noted that the deformation repair maintenance plan sequence and the vibration state maintenance plan sequence are one-to-one corresponding in time sequence, that is, each vibration state maintenance plan has a corresponding deformation repair maintenance plan.

[0237] Then, the system uses the first semantic recognition branch in the maintenance strategy output network to process the vibration state maintenance plan sequence to obtain the vibration state semantic moment sequence. This sequence includes Y vibration state semantic moments, each of which contains the semantic information of the corresponding maintenance plan, such as maintenance goals, execution steps, etc. This information is represented in the form of a matrix to facilitate subsequent calculation and processing.

[0238] Similarly, the system uses the second semantic recognition branch in the maintenance strategy output network to process the deformation repair maintenance plan sequence to obtain a deformation repair semantic moment sequence. This sequence also includes Y deformation repair semantic moments, each of which contains the semantic information of the corresponding maintenance plan.

[0239] After obtaining the vibration state semantic moment sequence and the deformation repair semantic moment sequence, the system will use the feasibility identification branch in the maintenance strategy output network to conduct a comprehensive analysis of the two sequences to obtain the feasibility weights corresponding to the vibration state maintenance plan sequence. The feasibility weight reflects the feasibility and effect expectation of each maintenance plan in actual execution, and is an important basis for determining the execution order of the maintenance strategy.

[0240] Finally, the system will determine the strategy execution order of the vibration state maintenance plan sequence according to the feasibility weight. Generally, the maintenance plan with a higher feasibility weight will be executed first to ensure the safety and stability of the membrane structure. At the same time, the system will flexibly adjust and optimize the maintenance strategy according to actual needs and resource conditions.

[0241] In some independent embodiments, the feasibility weight corresponding to the vibration state maintenance plan sequence is obtained through the feasibility identification branch included in the maintenance strategy output network based on the vibration state semantic moment sequence and the deformation repair semantic moment sequence, including: based on the vibration state semantic moment sequence, Y first quantitative monitoring element arrays are obtained through the first feature focusing model included in the maintenance strategy output network, wherein each first quantitative monitoring element array corresponds to a vibration state semantic moment; based on the deformation repair semantic moment sequence, Y second quantitative monitoring element arrays are obtained through the second feature focusing model included in the maintenance strategy output network, wherein each second quantitative monitoring element array corresponds to a deformation repair semantic moment; the Y first quantitative monitoring element arrays and the Y second quantitative monitoring element arrays are spliced ​​to obtain Y target quantitative monitoring element arrays, wherein each target quantitative monitoring element array includes a first quantitative monitoring element array and a second quantitative monitoring element array; based on the Y target quantitative monitoring element arrays, the feasibility weight corresponding to the vibration state maintenance plan sequence is obtained through the feasibility identification branch included in the maintenance strategy output network.

[0242] In some independent embodiments, the system performs specific steps to determine the feasibility weights of the vibration state maintenance plan sequence. This process involves multiple stages of data processing and model application to ensure that the final weights are both accurate and reliable.

[0243] First, based on the vibration state semantic moment sequence, the system uses the first feature focusing model in the maintenance strategy output network to obtain Y first quantitative monitoring element arrays. Each first quantitative monitoring element array corresponds to a feature representation of a vibration state semantic moment. These features may include key information such as vibration frequency, amplitude, duration, etc., which are quantized into numerical form for subsequent processing. The role of the first feature focusing model is to extract key features related to the maintenance strategy from the semantic moment and convert them into numerical form.

[0244] Next, the system also uses the second feature focus model in the maintenance strategy output network to obtain Y second quantitative monitoring element arrays based on the deformation repair semantic moment sequence. Similar to the first quantitative monitoring element array, each second quantitative monitoring element array also corresponds to a feature representation of a deformation repair semantic moment, but the features here may focus more on aspects such as deformation degree and repair difficulty. The second feature focus model is also responsible for extracting key features from the semantic moment and quantizing them into numerical values.

[0245] Then, the system concatenates the Y first quantitative monitoring element arrays and the Y second quantitative monitoring element arrays. The purpose of concatenation is to combine the feature information of vibration state and deformation repair to form a more comprehensive feature set. In this way, each target quantitative monitoring element array contains the feature information of a vibration state semantic moment and a deformation repair semantic moment.

[0246] Finally, based on the Y target quantitative monitoring element array, the system uses the feasibility identification branch in the maintenance strategy output network to obtain the feasibility weight corresponding to the vibration state maintenance plan sequence. The role of the feasibility identification branch is to evaluate the feasibility of each maintenance plan based on these comprehensive feature information and assign a weight value to it. This weight value reflects the feasibility and effect expectation of the maintenance plan in actual application, and is an important basis for determining the execution order of the maintenance strategy.

[0247] Through the above steps, the system can comprehensively consider the characteristic information of vibration state and deformation repair, and assign an accurate feasibility weight to each maintenance plan, thereby providing strong support for the subsequent execution of maintenance strategies.

[0248] In some independent embodiments, based on the vibration state semantic moment sequence, Y first quantitative monitoring element arrays are obtained through the first feature focusing model included in the maintenance strategy output network, including: for each vibration state semantic moment in the vibration state semantic moment sequence, a first local sampling semantic moment is obtained through the local sampling layer included in the first feature focusing model, wherein the first feature focusing model belongs to the maintenance strategy output network; for each vibration state semantic moment in the vibration state semantic moment sequence, a first global semantic moment is obtained through the global semantic layer included in the first feature focusing model; for each vibration state semantic moment in the vibration state semantic moment sequence, based on the first local sampling semantic moment and the first global semantic moment, a first residual semantic moment is obtained through the residual layer included in the first feature focusing model; for each vibration state semantic moment in the vibration state semantic moment sequence, based on the first residual semantic moment and the vibration state semantic moment, a first quantitative monitoring element array is obtained through the first global semantic layer included in the first feature focusing model.

[0249] In some independent embodiments, the system processes the vibration state semantic moment sequence through specific steps and models to obtain the first quantitative monitoring element array. These arrays are important basis for determining the feasibility of the maintenance plan in the future. The following is a detailed explanation.

[0250] First, for each vibration state semantic moment in the vibration state semantic moment sequence, the system uses the local sampling layer in the first feature focusing model to obtain the first local sampled semantic moment. The role of the local sampling layer is to extract and sample the local features of the semantic moment to capture the detailed information. This information is crucial for understanding the local vibration state and maintenance requirements of the membrane structure.

[0251] Next, the system will also obtain the first global semantic moment for each vibration state semantic moment through the global semantic layer in the first feature focusing model. The global semantic layer is responsible for capturing the overall features and contextual information in the semantic moment, which helps the system to understand the vibration state of the membrane structure more comprehensively.

[0252] Then, the system combines the first local sampled semantic moment and the first global semantic moment to obtain the first residual semantic moment through the residual layer in the first feature focusing model. The residual layer is designed to learn the residual relationship between the input and output, that is, the changing part that the model needs to learn. This mechanism helps the model better capture and represent the complex features in the vibration state semantic moment.

[0253] Finally, the system will obtain the first quantitative monitoring factor array based on the first residual semantic moment and the original vibration state semantic moment through the global semantic layer in the first feature focusing model. This array is a deep feature representation of the vibration state semantic moment, which contains key quantitative information related to the maintenance strategy.

[0254] Through the above steps, the system can extract rich feature information for each vibration state semantic moment and quantify this information into a first quantitative monitoring element array in numerical form. These arrays provide important data support for the subsequent determination of the feasibility of the maintenance plan.

[0255] Figure 2 FIG. 2 is a schematic diagram of a membrane structure abnormality prediction system 200 provided in an embodiment of the present invention. Figure 2 The membrane structure abnormality prediction system 200 shown includes a processor 210, and the processor 210 can call and run a computer program from a memory to implement the method in the embodiment of the present invention.

[0256] Alternatively, if Figure 2 As shown, the membrane structure abnormality prediction system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present invention.

[0257] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0258] Alternatively, if Figure 2 As shown, the membrane structure abnormality prediction system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0259] Optionally, the membrane structure abnormality prediction system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present invention, which will not be repeated here for the sake of brevity.

[0260] It should be understood that the processor of the embodiment of the present invention may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by the hardware integrated logic circuit in the processor or the instruction in the form of software. The above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor can be combined to perform. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0261] It can be understood that the memory in the embodiment of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0262] It should be understood that the above-mentioned memory is exemplary but not restrictive. For example, the memory in the embodiment of the present invention may also be static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synch link DRAM, SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DR RAM), etc. That is to say, the memory in the embodiment of the present invention is intended to include but not limited to these and any other suitable types of memory.

[0263] Based on the above, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the above method when running.

[0264] The above are only embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations.

Claims

1. A membrane structure abnormality prediction method based on artificial intelligence, characterized in that: Applied to a membrane structure abnormality prediction system, the method comprises: Acquire structural state sensor monitoring data of the membrane structure to be monitored and at least one environmental state sensor monitoring data; Performing X-order monitoring knowledge mining on the structural state sensor monitoring data through a first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data through the first artificial intelligence algorithm to obtain a second sensor monitoring knowledge relationship network set, wherein the first artificial intelligence algorithm is a deep residual network, the first sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, and the second sensor monitoring knowledge relationship network set includes sensor monitoring knowledge relationship networks of X different feature dimensions, where X is a positive integer; Integrating the monitoring knowledge of the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set to obtain a first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored; updating monitoring knowledge according to the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set by a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set; Based on the first linkage sensor monitoring knowledge relationship network set and the second linkage sensor monitoring knowledge relationship network set, abnormal risk prediction is performed to obtain an abnormal risk prediction viewpoint of the membrane structure to be monitored; The first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set are integrated to obtain a first linkage sensor monitoring knowledge relationship network set of the membrane structure to be monitored, including: Determining a feature enhancement factor according to a first sensor monitoring knowledge relationship network and a second sensor monitoring knowledge relationship network, wherein the first sensor monitoring knowledge relationship network and the second sensor monitoring knowledge relationship network have consistent feature dimensions, the first sensor monitoring knowledge relationship network is included in the first sensor monitoring knowledge relationship network set, and the second sensor monitoring knowledge relationship network is included in the second sensor monitoring knowledge relationship network set; Determine the splicing coefficient by combining the second sensor monitoring knowledge relationship network with the feature enhancement factor to obtain the knowledge splicing coefficient; Splicing the knowledge splicing coefficient with the first sensor monitoring knowledge relationship network to obtain a first linkage sensor monitoring knowledge relationship network, wherein the first linkage sensor monitoring knowledge relationship network belongs to the first linkage sensor monitoring knowledge relationship network set; Walk the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set in order to obtain a first linkage sensor monitoring knowledge relationship network set; The monitoring knowledge is updated according to the first sensor monitoring knowledge relationship network set and the second sensor monitoring knowledge relationship network set by a second artificial intelligence algorithm to obtain a second linkage sensor monitoring knowledge relationship network set including: Respectively optimizing the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network to obtain a first sensor monitoring optimization knowledge relationship network and a second sensor monitoring optimization knowledge relationship network, wherein the third sensor monitoring knowledge relationship network and the fourth sensor monitoring knowledge relationship network have consistent feature dimensions, the third sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the fourth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X+1-th order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm; Integrating monitoring knowledge on the first sensor monitoring optimization knowledge relationship network and the second sensor monitoring optimization knowledge relationship network to obtain a first linkage sensor monitoring optimization knowledge relationship network; performing monitoring knowledge updating according to the first linkage sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring knowledge relationship network, wherein the second linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the second linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the first artificial intelligence algorithm at the Xth order; Respectively optimizing the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network to obtain a third sensor monitoring optimization knowledge relationship network and a fourth sensor monitoring optimization knowledge relationship network, wherein the fifth sensor monitoring knowledge relationship network and the sixth sensor monitoring knowledge relationship network have consistent feature dimensions, the fifth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X-order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm, and the sixth sensor monitoring knowledge relationship network is a sensor monitoring knowledge relationship network obtained by performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm; Integrating monitoring knowledge on the third sensor monitoring optimization knowledge relationship network and the fourth sensor monitoring optimization knowledge relationship network to obtain a second linkage sensor monitoring optimization knowledge relationship network; performing monitoring knowledge updating according to the second linkage sensor monitoring optimization knowledge relationship network to obtain a third linkage sensor monitoring knowledge relationship network, wherein the third linkage sensor monitoring knowledge relationship network is included in the second linkage sensor monitoring knowledge relationship network set, and the characteristic dimension of the third linkage sensor monitoring knowledge relationship network is the same as the characteristic dimension of the linkage sensor monitoring knowledge relationship network generated by the X-1th order of the first artificial intelligence algorithm; Through X rounds of cyclic processing, the second linkage sensor monitoring knowledge relationship network set is obtained.

2. The method according to claim 1, characterized in that The method further comprises: Acquire an algorithm debugging sample set and an artificial intelligence algorithm to be debugged, wherein the algorithm debugging sample set includes active sensor monitoring data samples and environmental state sensor monitoring data; Performing X-order monitoring knowledge mining on the active sensor monitoring data sample based on the first artificial intelligence algorithm to obtain X sensor monitoring knowledge relationship networks of the active sensor monitoring data sample; Based on the first artificial intelligence algorithm, the environmental state sensor monitoring data is subjected to X-order monitoring knowledge mining to obtain X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data; Performing monitoring knowledge integration on the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data to obtain X linkage sensor monitoring knowledge relationship network samples; Transmitting the X sensor monitoring knowledge relationship networks of the active sensor monitoring data samples and the X sensor monitoring knowledge relationship networks of the environmental state sensor monitoring data into the artificial intelligence algorithm to be debugged to update the monitoring knowledge, so as to obtain X updated sensor monitoring knowledge relationship networks; Performing abnormal risk prediction based on the X updated sensor monitoring knowledge relationship networks and the X linked sensor monitoring knowledge relationship network examples to obtain a debugging error variable; The artificial intelligence algorithm to be debugged is debugged based on the debugging error variable to obtain the second artificial intelligence algorithm.

3. The method according to claim 2, characterized in that The environmental state sensing monitoring data includes temperature and humidity state sensing monitoring data and ice and wind load sensing monitoring data of the membrane structure building sample, and the active sensing monitoring data sample is stress state sensing monitoring data of the membrane structure building sample.

4. The method according to claim 2, characterized in that: The method further comprises: Obtain at least two basic normalized subnets of the first artificial intelligence algorithm, wherein the at least two basic normalized subnets have a one-to-one matching relationship with the active sensor monitoring data sample and the environmental state sensor monitoring data; The at least two basic normalized distribution subnetworks are supervised and debugged based on the debugging error variable to obtain at least two target normalized distribution subnetworks, wherein the at least two target normalized distribution subnetworks have different algorithm model variables.

5. The method according to claim 4, characterized in that The step of performing X-order monitoring knowledge mining on the structural state sensor monitoring data by the first artificial intelligence algorithm to obtain a first sensor monitoring knowledge relationship network set, and performing X-order monitoring knowledge mining on the at least one environmental state sensor monitoring data by the first artificial intelligence algorithm to obtain a second sensor monitoring knowledge relationship network set includes: Performing monitoring knowledge mining on the structural state sensor monitoring data through the first artificial intelligence algorithm and processing through the corresponding target normal distribution subnet to obtain the first sensor monitoring knowledge relationship network set; The monitoring knowledge mining of the at least one environmental status sensor monitoring data is performed through the first artificial intelligence algorithm and processed through the corresponding target normal distribution subnet to obtain the second sensor monitoring knowledge relationship network set.

6. The method according to claim 1, characterized in that The method further comprises: A maintenance strategy for the membrane structure to be monitored is generated based on the abnormal risk prediction viewpoint.

7. A membrane structure abnormality prediction system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program implements the method according to any one of claims 1 to 6 when running.

Images