Building information model driven full-life-cycle intelligent operation and maintenance management method and system

Abstract

The invention relates to the technical field of building operation and maintenance management, and discloses a whole-life-cycle intelligent operation and maintenance management method and system driven by a building information model. The method comprises the steps of obtaining and preprocessing original data of a building information model to obtain standardized structure data; a preliminary operation and maintenance feature model is constructed based on the space-time diagram convolutional network, and operation and maintenance decision features are generated through meta-learning framework dynamic parameter optimization; an intelligent operation and maintenance strategy set is generated in combination with a multi-objective optimization algorithm, and feasibility is verified through a dynamic time warping algorithm and iterative updating is carried out; and generating a dynamic operation and maintenance management scheme to drive equipment operation and maintenance and update a model state. The invention also relates to construction of a digital twin model based on Internet of Things data, anomaly detection and the like. The method effectively integrates the building operation and maintenance data, realizes multi-target intelligent decision, improves the feasibility of the operation and maintenance strategy, adapts to environmental changes in real time, detects abnormities, improves the building operation and maintenance efficiency, guarantees the safe and stable operation of the building, and is suitable for the whole-life-cycle operation and maintenance management of various buildings.

Description

Technical Field

[0001] The present invention relates to the technical field of building operation and maintenance management, and specifically to a full life cycle intelligent operation and maintenance management method and system driven by a building information model. Background Art

[0002] With the acceleration of urbanization, the construction field has shown a trend of rapid growth in number and diversification of types, and the scale of buildings and system complexity continue to increase. The traditional building operation and maintenance management model has exposed many problems in dealing with these modern buildings and is unable to meet the growing operation and maintenance needs.

[0003] In terms of data management, building operation and maintenance involves massive and complex data, covering building structure, equipment parameters, operating status, maintenance records and other aspects. In the traditional mode, these data are often stored in a scattered manner with different formats, lacking effective integration and standardized processing. For example, different departments may record equipment operation data and maintenance data separately, and the data formats are not unified, which makes it difficult to correlate and analyze the data, and cannot provide a comprehensive and accurate basis for operation and maintenance decisions. It is easy to cause information islands, making it difficult for operation and maintenance personnel to obtain and use data.

[0004] From the perspective of operation and maintenance decision-making, traditional operation and maintenance decisions mainly rely on the experience of operation and maintenance personnel and lack a scientific and systematic decision-making method. This method is highly subjective and it is difficult to comprehensively consider multiple factors such as energy consumption costs, equipment life and safety level. For example, in the formulation of equipment maintenance plans, it may only be based on maintenance experience after equipment failure occurs, without fully considering the actual operating status of the equipment, remaining life and maintenance costs. This is not only likely to lead to over-maintenance or under-maintenance, increase operation and maintenance costs, but may also affect the normal operation of equipment and reduce the safety and reliability of buildings.

[0005] Traditional methods also have obvious deficiencies in the verification and optimization of operation and maintenance strategies. Due to the lack of effective verification methods, it is often difficult to judge the feasibility of newly formulated operation and maintenance strategies in actual applications. Moreover, once problems are found in the strategy, it is difficult to make targeted optimization adjustments. For example, when trying to adopt a new energy-saving strategy, it is impossible to accurately assess the potential impact of the strategy on the life and safety level of the equipment. If it is implemented blindly, it may cause a series of problems that cannot be solved in a timely and effective manner.

[0006] In addition, with the rapid development of emerging technologies such as the Internet of Things, big data, and artificial intelligence, the field of building operation and maintenance management has also ushered in new opportunities and challenges. These technologies have made it possible to achieve intelligent operation and maintenance, but their application in building operation and maintenance is still in the exploratory stage, and a mature and complete intelligent operation and maintenance management system has not yet been formed. The capabilities of existing technologies in data mining and analysis, intelligent decision support, real-time monitoring and dynamic adjustment still need to be improved, and it is impossible to give full play to the advantages of emerging technologies and achieve efficient operation and maintenance management throughout the life cycle of buildings. Summary of the invention

[0007] The purpose of the present invention is to provide a full life cycle intelligent operation and maintenance management method and system driven by a building information model to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a full life cycle intelligent operation and maintenance management method and system driven by a building information model, the method comprising: Acquiring raw data of a building information model, and preprocessing the raw data to obtain standardized structural data; According to the standardized structure data, a preliminary operation and maintenance feature model is constructed based on a spatiotemporal graph convolutional network; Using the preliminary operation and maintenance feature model, dynamic parameter optimization is performed through a meta-learning framework to generate optimized operation and maintenance decision features; According to the optimized operation and maintenance decision characteristics, a set of intelligent operation and maintenance strategies for the entire life cycle is generated in combination with a multi-objective optimization algorithm; Align the intelligent operation and maintenance strategy set with the historical operation and maintenance data in time series, and verify the feasibility of the strategy through a dynamic time warping algorithm; Adjust the constraints of the multi-objective optimization algorithm based on the verification results, and iteratively update the intelligent operation and maintenance strategy set; Generate the final dynamic operation and maintenance management plan based on the updated intelligent operation and maintenance strategy set; Based on the dynamic operation and maintenance management solution, real-time operation and maintenance operations of building equipment are driven, and status data of the building information model is updated synchronously.

[0009] Preferably, the preprocessing of the raw data includes: Extract geometric properties, material properties and equipment association data from building information models; Fill missing values ​​and remove outliers on the extracted data to generate a structured data table; The structured data table is converted into tensor data of uniform dimension as standardized structured data through a standardized mapping algorithm.

[0010] Preferably, the construction of a preliminary operation and maintenance feature model based on a spatiotemporal graph convolutional network includes: The standardized structure data is input into the spatial convolution layer of the spatiotemporal graph convolutional network to capture the topological correlation features between devices; The temporal dependency of historical operation and maintenance data is modeled through the time convolution layer to generate spatiotemporal fusion features; The attention mechanism is used to assign weights to the spatiotemporal fusion features, and a preliminary operation and maintenance feature model containing key operation and maintenance nodes is output.

[0011] Preferably, the dynamic parameter optimization through a meta-learning framework includes: Define a set of meta-learning tasks, each of which corresponds to parameter adjustment requirements in different operation and maintenance scenarios; A model-agnostic meta-learning algorithm is used to quickly update the local parameters of the spatiotemporal graph convolutional network on the support set; The generalization performance after parameter update is evaluated through the query set, and the optimized operation and maintenance decision features that are suitable for multiple scenarios are screened out.

[0012] Preferably, the intelligent operation and maintenance strategy set for the entire life cycle generated by combining a multi-objective optimization algorithm includes: Construct a multi-objective optimization function that includes energy cost, equipment life, and safety level; A non-dominated sorting genetic algorithm is used to solve the optimization function and generate a Pareto optimal strategy candidate set; Based on the fuzzy comprehensive evaluation method, the intelligent operation and maintenance strategy set that meets the current working conditions is selected from the candidate set.

[0013] Preferably, the feasibility of the strategy verified by the dynamic time warping algorithm includes: Align the predicted operation sequence in the intelligent operation and maintenance strategy set with the historical success case sequence in the time dimension; Calculate the minimum path distance between aligned sequences as a quantitative indicator of strategy feasibility; Strategies that deviate from the threshold range are screened out and marked based on quantitative indicators.

[0014] Preferably, the iterative update of the intelligent operation and maintenance strategy set includes: Adaptive penalty terms are introduced to dynamically weight the constraints of multi-objective optimization algorithms; Adjust the search direction of the optimization algorithm through the covariance matrix adaptive evolution strategy; Infeasible strategies are eliminated based on the marking results to generate an updated intelligent operation and maintenance strategy set.

[0015] Preferably, after generating the final dynamic operation and maintenance management solution, the method further includes: Build a real-time status mapping model of digital twins based on IoT sensor data; The dynamic operation and maintenance management solution is fine-tuned online through the differential evolution algorithm to adapt to real-time environmental changes; Synchronize the fine-tuned plan to the operation and maintenance log database of the building information model.

[0016] Preferably, the method further includes the step of performing abnormality detection on the dynamic operation and maintenance management solution: Use the isolation forest algorithm to identify abnormal nodes in the operation sequence; Analyze the associated impact scope of abnormal nodes through the knowledge graph reasoning engine; Generate exception handling plans and trigger the early warning module of the building information model.

[0017] Preferably, the present invention further includes a full life cycle intelligent operation and maintenance management system driven by a building information model, the system comprising: Data preprocessing module: used to obtain the original data of the building information model and preprocess the original data to obtain standardized structural data; Preliminary model building module: constructing a preliminary operation and maintenance feature model based on the spatiotemporal graph convolutional network according to the standardized structure data; Dynamic optimization module: using the preliminary operation and maintenance feature model, dynamic parameter optimization is performed through a meta-learning framework to generate optimized operation and maintenance decision features; Strategy generation module: based on the optimized operation and maintenance decision features, a set of intelligent operation and maintenance strategies for the entire life cycle is generated in combination with a multi-objective optimization algorithm; Strategy verification module: align the intelligent operation and maintenance strategy set with the historical operation and maintenance data in time series, and verify the feasibility of the strategy through a dynamic time warping algorithm; Iterative update module: adjusts the constraints of the multi-objective optimization algorithm based on the verification results, and iteratively updates the intelligent operation and maintenance strategy set; Solution output module: Generates the final dynamic operation and maintenance management solution based on the updated intelligent operation and maintenance strategy set; Operation and maintenance driving module: drives the real-time operation and maintenance of building equipment based on the dynamic operation and maintenance management solution, and synchronously updates the status data of the building information model.

[0018] Compared with the prior art, the present invention has the following beneficial effects: In data processing, by obtaining the original data of the building information model and preprocessing it, the geometric properties, material properties and equipment association data are extracted, missing values ​​are filled, outliers are removed and converted into uniform dimensional tensor data. This makes the originally chaotic data standardized and orderly, providing a solid and reliable foundation for subsequent analysis. For example, in a large commercial complex, after processing, the data of many different types of equipment can be easily analyzed for association, and the operation and maintenance personnel can quickly grasp the connection between the various equipment and the overall operating status, which greatly improves the efficiency of data utilization.

[0019] In the process of building a preliminary operation and maintenance feature model, the application of spatiotemporal graph convolutional networks can effectively capture the topological correlation characteristics between devices and the temporal dependencies of historical operation and maintenance data. This helps to discover the potential laws in the operation of building equipment. For example, by analyzing the operation data of each device in the air-conditioning system in different seasons and time periods, its operation mode and potential failure risks can be accurately grasped, providing strong support for early prevention of failures.

[0020] Dynamic parameter optimization under the meta-learning framework enables the model to quickly adapt to different operation and maintenance scenarios. In hospitals, schools and other buildings with different functions, the model can adjust parameters in a targeted manner according to their different emphasis on equipment reliability, safety and energy consumption, generate operation and maintenance decision features that are more in line with the actual situation, and improve the accuracy and effectiveness of operation and maintenance decisions.

[0021] The multi-objective optimization algorithm combined with the fuzzy comprehensive evaluation method generates an intelligent operation and maintenance strategy set, which fully considers key factors such as energy consumption cost, equipment life and safety level. Taking office buildings as an example, this method can be used to formulate an operation and maintenance strategy that effectively reduces energy consumption costs while ensuring the safe and stable operation of equipment and extending the life of equipment, thus achieving a balanced optimization of multiple objectives and saving a lot of costs for building operations.

[0022] The strategy verification and iterative update mechanism ensures the feasibility and continuous optimization of the operation and maintenance strategy. The dynamic time warping algorithm verifies the strategy, can promptly discover potential problems, and combines the adaptive penalty term and the covariance matrix adaptive evolution strategy to adjust the optimization algorithm and eliminate infeasible strategies. This allows the operation and maintenance strategy to be continuously improved in practical applications, and improves the reliability and stability of operation and maintenance management.

[0023] The real-time state mapping model of the digital twin and the differential evolution algorithm enable real-time fine-tuning of the operation and maintenance plan. With the help of IoT sensor data, the operating status of building equipment can be reflected in real time. For example, the intelligent lighting system adjusts the brightness in real time according to the changes in indoor and outdoor light, achieving energy saving while meeting lighting needs. The anomaly detection mechanism uses the isolation forest algorithm and the knowledge graph reasoning engine to quickly identify anomalies and analyze the scope of impact, generate processing plans and warnings, and ensure the safe operation of the building.

[0024] In general, the present invention realizes intelligent and refined operation and maintenance management of the entire life cycle of the building, improves operation and maintenance efficiency, reduces operation and maintenance costs, enhances the safety and reliability of the building, brings innovative solutions to the field of building operation and maintenance management, and has broad application prospects and significant social and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A working principle diagram of the full life cycle intelligent operation and maintenance management method driven by a building information model according to the present invention; Figure 2 Flowchart for the construction of a preliminary O&M feature model based on spatiotemporal convolution; Figure 3 A flowchart for feasibility verification of intelligent operation and maintenance strategy; Figure 4 Schematic diagram of the architecture of the full life cycle intelligent operation and maintenance management system driven by building information modeling. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] See also Figure 1-Figure 4 The present invention provides a full life cycle intelligent operation and maintenance management method and system driven by a building information model, and the specific implementation steps are as follows: Obtain and preprocess raw data: Obtain raw data from the building information model, which contains various types of building information, such as building structure, equipment parameters, material properties, etc. Then preprocess the raw data to convert it into standardized structural data that is convenient for subsequent analysis and processing.

[0028] Constructing a preliminary operation and maintenance feature model: Based on the obtained standardized structure data, a preliminary operation and maintenance feature model is constructed with the help of a spatiotemporal graph convolutional network. The spatiotemporal graph convolutional network can effectively capture the topological association characteristics between building equipment and the temporal dependency of historical operation and maintenance data, thereby generating a preliminary model containing key operation and maintenance nodes.

[0029] Dynamic parameter optimization to generate operation and maintenance decision features: Using the preliminary operation and maintenance feature model, dynamic parameter optimization is performed through a meta-learning framework. The meta-learning framework can quickly adjust and optimize model parameters according to the requirements of different operation and maintenance scenarios, and then generate optimized operation and maintenance decision features to make them more adaptable to complex and changing operation and maintenance environments.

[0030] Generate intelligent operation and maintenance strategy set: Combined with the optimized operation and maintenance decision-making characteristics, a multi-objective optimization algorithm is used to generate an intelligent operation and maintenance strategy set for the entire life cycle. The multi-objective optimization algorithm comprehensively considers multiple goals such as energy consumption cost, equipment life and safety level, and the generated strategy set can achieve a balance between multiple goals.

[0031] Verify the feasibility of the strategy: Align the generated intelligent operation and maintenance strategy set with the historical operation and maintenance data in time series, and verify the feasibility of the strategy through the dynamic time warping algorithm. The dynamic time warping algorithm can measure the similarity between the predicted operation sequence and the historical success case sequence, and judge the feasibility of the strategy in actual application.

[0032] Iteratively update the intelligent operation and maintenance strategy set: According to the verification results, adjust the constraints of the multi-objective optimization algorithm and iteratively update the intelligent operation and maintenance strategy set. By introducing adaptive penalty items and adjusting the search direction of the optimization algorithm, infeasible strategies are eliminated to make the strategy set more reasonable and effective.

[0033] Generate dynamic operation and maintenance management plan: Based on the updated intelligent operation and maintenance strategy set, generate the final dynamic operation and maintenance management plan. This plan takes into account various factors and provides specific guidance for the operation and maintenance of building equipment.

[0034] Drive maintenance operations and update model status: Drive real-time maintenance operations of building equipment based on dynamic maintenance management solutions, and simultaneously update the status data of the building information model to ensure that the building information model is consistent with the actual maintenance situation, providing accurate data support for subsequent maintenance management.

[0035] The implementation of the present invention is further described below in conjunction with Examples 1 to 6. Example

[0036] After obtaining the raw data of the building information model, preprocessing is particularly critical. The raw data is usually complex and diverse, and may contain information of various formats and types. First, extract geometric properties from the building information model, which includes information such as the shape, size, and spatial layout of the building. For example, data such as the building's floor height, room size, and wall thickness are essential for understanding the physical structure of the building. At the same time, extract material properties, such as the type, strength, and durability of building materials, such as the grade of concrete and the model of steel. These properties affect the performance and maintenance requirements of the building. It is also necessary to extract equipment association data to clarify the connection method and control relationship between each device, such as the connection relationship between the host and the terminal device in the air-conditioning system, and the relationship between lamps and switches and distribution boxes in the lighting system.

[0037] After the data is extracted, there may be incomplete or abnormal data. There are many methods to fill missing values. If it is continuous data, the mean and median can be used to fill the missing values; if it is discrete data, it can be filled according to the value with the highest frequency. For example, in the data recording the operating temperature of building equipment, if there are individual missing values, the average value of other normal data can be calculated to fill them. For outlier removal, it is judged by setting a reasonable threshold range. For example, the operating power of building equipment fluctuates within a certain range under normal circumstances. If there is a maximum or minimum value beyond this range, it can be judged as an outlier and removed. After these operations, a structured data table is generated, and various types of data are presented in a clear and standardized table format to facilitate subsequent processing.

[0038] Finally, the structured data table is converted into tensor data of uniform dimension through the standardized mapping algorithm. Assume that the data dimensions in the structured data table are inconsistent. Some columns represent the equipment running time, and the data range may be 0-1000 hours, while some columns represent the equipment temperature, and the data range may be 0-100 degrees Celsius. The standardized mapping algorithm will uniformly map these data of different dimensions to a specific dimensional range, such as mapping all data to between 0-1, making it standardized structured data, which is convenient for subsequent input into the spatiotemporal graph convolutional network for processing. The processed data can be better understood and utilized by the model, providing a reliable data foundation for subsequent operation and maintenance management. Example

[0039] The spatiotemporal graph convolutional network plays a core role in building the preliminary operation and maintenance feature model. The standardized structured data obtained after preprocessing is input into the spatial convolution layer of the spatiotemporal graph convolutional network. The role of the spatial convolutional layer is to capture the topological association characteristics between devices. For example, in a large commercial building, there are complex connections and interactions between various devices such as air-conditioning systems, power systems, and lighting systems. Through convolution operations, the spatial convolutional layer can analyze the spatial adjacency, connection strength and other characteristics of these devices. Taking the air-conditioning system as an example, it can identify the connection relationship between the air-conditioning hosts on different floors and the terminal devices on each floor, as well as the degree of closeness of their spatial layout, thereby extracting the spatial topological characteristics between the devices.

[0040] Next, the data enters the temporal convolution layer. The temporal convolution layer mainly models the temporal dependencies of historical operation and maintenance data. Historical operation and maintenance data contains information such as the operating status and maintenance records of the equipment at different time points. Through the temporal convolution layer, it is possible to discover the changing patterns of the equipment's operating status over time, as well as the impact of maintenance operations on the subsequent operation of the equipment. For example, by analyzing the fault records and maintenance times of an elevator over a period of time, the temporal convolution layer can learn the time interval pattern of elevator faults and the changing trend of the equipment's operating stability after each maintenance.

[0041] After processing by the spatial convolution layer and the temporal convolution layer, spatiotemporal fusion features are generated. In order to highlight key information, the attention mechanism is used to assign weights to the spatiotemporal fusion features. The attention mechanism assigns different weights to different features according to their importance. In the building operation and maintenance scenario, the attention mechanism will assign higher weights to the operating status features of some key equipment, such as fire protection system equipment, elevators, etc., and lower weights to some relatively minor features, such as individual state changes of ordinary lighting equipment. In this way, a preliminary operation and maintenance feature model containing key operation and maintenance nodes is output, which can more accurately reflect the key information in building operation and maintenance and provide strong support for subsequent decision-making. Example

[0042] The meta-learning framework plays an important role in dynamic parameter optimization. First, define the meta-learning task set, where each task corresponds to the parameter adjustment requirements under different operation and maintenance scenarios. Different operation and maintenance scenarios may involve different building types, equipment operating conditions, environmental conditions, etc. For example, in hospital buildings, the reliability and safety of equipment are extremely high, which requires setting tasks for hospital operation and maintenance scenarios in the meta-learning task set, focusing on the operating stability and maintenance timeliness of medical equipment. The corresponding parameter adjustment requirements may focus on optimizing the accuracy of equipment failure prediction and the rationality of maintenance plans. In ordinary office buildings, more attention is paid to the control of energy consumption costs. The tasks for office buildings in the meta-learning task set will set parameter adjustment requirements around reducing energy consumption.

[0043] A model-agnostic meta-learning algorithm is used to quickly update the local parameters of the spatiotemporal graph convolutional network on the support set. The support set is a representative part of the historical operation and maintenance data. Taking the historical operation and maintenance data of a specific building as an example, a part of the data is used as the support set. The model-agnostic meta-learning algorithm can quickly try different parameter adjustment methods on these data to find the parameter update direction that can improve the model performance. For example, adjust the size of the convolution kernel, learning rate and other parameters in the spatiotemporal graph convolutional network, observe the performance of the model on the support set, and select the optimal parameter update strategy.

[0044] The generalization performance after parameter update is evaluated through the query set. The query set also comes from historical operation and maintenance data, but unlike the support set, it is used to evaluate the performance of the model on new data. The model with updated parameters is applied to the query set to calculate the various performance indicators of the model, such as prediction accuracy, error rate, etc. By comparing the performance of the models after different parameter updates on the query set, the optimized operation and maintenance decision features that are suitable for multiple scenarios are selected. The operation and maintenance decision features obtained in this way can have good performance in different operation and maintenance scenarios, improving the adaptability and reliability of the model. Example

[0045] When generating a set of intelligent operation and maintenance strategies for the entire life cycle, combining a multi-objective optimization algorithm is a key step. First, a multi-objective optimization function including energy consumption cost, equipment life and safety level is constructed. Energy consumption cost is an important indicator in building operation and maintenance, which is related to factors such as equipment operation time and power. Assume that the energy consumption cost function is ,in represents the energy cost, Indicates the number of devices. Indicates The power of the device, Indicates The operating time of a device. The device life is related to the frequency of use and maintenance of the device. For example, the device life function can be expressed as , is the actual life of the device, is the initial design life of the equipment, is the number of maintenance times, It is The frequency of equipment use during maintenance. It is The impact coefficient of maintenance on equipment life. The safety level takes into account the importance of the equipment, the possible harm caused by failure, and other factors. Indicates the security level, which is a value calculated based on various factors.

[0046] The non-dominated sorting genetic algorithm is used to solve the optimization function and generate a Pareto optimal strategy candidate set. The non-dominated sorting genetic algorithm searches the solution space by simulating the biological evolution process. It regards different operation and maintenance strategies as individuals in the population and continuously evolves the population through operations such as selection, crossover, and mutation. In this process, the algorithm will evaluate the pros and cons of each individual according to the multi-objective optimization function, and screen out individuals that cannot be dominated by other individuals (that is, no other individual is better than it in all objectives) to form a Pareto optimal strategy candidate set. For example, at a certain moment, there are multiple operation and maintenance strategies, some of which perform well in reducing energy consumption costs, but the equipment life may be affected to a certain extent; while other strategies may pay more attention to equipment life and safety level, but the energy consumption cost is relatively high. The non-dominated sorting genetic algorithm will include all these strategies with different focuses into the candidate set.

[0047] Based on the fuzzy comprehensive evaluation method, the intelligent operation and maintenance strategy set that meets the current working conditions is selected from the candidate set. The fuzzy comprehensive evaluation method takes into account the fuzziness and uncertainty of multiple evaluation factors. First, determine the evaluation factor set, namely energy consumption cost, equipment life and safety level. Then establish a comment set, such as "excellent", "good", "medium", "poor", etc. Determine the membership of each factor to the comment set through expert scoring or other methods, and construct a fuzzy relationship matrix. Then determine the weight vector according to the importance of each factor, and finally obtain the comprehensive evaluation result of each candidate strategy through fuzzy synthesis operation, and select the strategy set with the best evaluation result as the intelligent operation and maintenance strategy set that meets the current working conditions. For example, under the current working conditions where the operating load of building equipment is high, the fuzzy comprehensive evaluation method may be more inclined to choose a strategy set that focuses on equipment life and safety level, while having a certain control over energy consumption costs. Example

[0048] There are closely connected steps in verifying the feasibility of the strategy and iteratively updating the intelligent operation and maintenance strategy set. First, the feasibility of the strategy is verified through the dynamic time warping algorithm. The predicted operation sequence in the intelligent operation and maintenance strategy set is aligned with the historical success case sequence in time dimension. For example, when predicting the maintenance operation sequence of a large ventilation equipment, the predicted maintenance time, maintenance content and other operation sequences are aligned with the operation sequences of similar equipment in history when they were successfully maintained. Since there may be certain differences in time in actual operation, the dynamic time warping algorithm can find the best time matching relationship between the two sequences.

[0049] The minimum path distance between the aligned sequences is calculated as a quantitative indicator of the feasibility of the strategy. This minimum path distance reflects the similarity between the predicted operation sequence and the historical success case sequence. Assume that the minimum path distance is , The smaller the value, the more similar the predicted operation sequence is to the historical success case sequence, and the higher the feasibility of the strategy; conversely, The larger the value, the lower the feasibility of the strategy. Filter out strategies that deviate from the threshold range based on quantitative indicators and mark them. For example, set a threshold ,when , the strategy is considered to deviate from the normal range and marked as a potentially infeasible strategy.

[0050] Iterate and update the intelligent operation and maintenance strategy set based on the verification results. Introduce adaptive penalty terms to dynamically weight the constraints of the multi-objective optimization algorithm. The adaptive penalty terms are adjusted according to the feasibility of the strategy. For those strategies marked as potentially infeasible, increase the penalty intensity in the constraints of the multi-objective optimization algorithm to make them more difficult to be selected in subsequent iterations. For example, for the energy cost target, if a strategy is found to have too high energy consumption and is not feasible during verification, increase the penalty weight of the strategy on the constraints of the energy cost target to reduce its priority in subsequent optimizations.

[0051] The search direction of the optimization algorithm is adjusted through the covariance matrix adaptive evolution strategy. The covariance matrix adaptive evolution strategy can dynamically adjust the search direction according to the distribution of the current population. If it is found that most of the strategies in the currently generated strategy set are biased towards a certain goal (such as excessive emphasis on energy consumption costs and neglect of equipment life), the strategy will adjust the search direction so that the subsequently generated strategies can consider multiple goals more balanced. Infeasible strategies are eliminated in combination with the marking results to generate an updated intelligent operation and maintenance strategy set. Those strategies that have been verified to be infeasible are eliminated from the strategy set, and feasible strategies are retained. New strategies are generated through the adjusted optimization algorithm to continuously improve the intelligent operation and maintenance strategy set to make it more in line with actual operation and maintenance needs. Example

[0052] After generating the final dynamic operation and maintenance management plan, there is still a series of important follow-up work. Construct a real-time state mapping model of the digital twin based on IoT sensor data. IoT sensor devices are spread all over the building and can collect real-time operating data of building equipment, such as temperature, pressure, speed, etc. These data are used to construct a real-time state mapping model of the digital twin, which can reflect the real operating status of building equipment in real time. For example, the running speed of the elevator, the temperature in the car, the door switch status and other data are collected by IoT sensors, and this information is updated in real time in the digital twin model, so that managers can intuitively understand the operation of the elevator, just like seeing a real elevator in the virtual world.

[0053] The differential evolution algorithm is used to fine-tune the dynamic operation and maintenance management plan online to adapt to real-time environmental changes. The building operating environment is constantly changing. For example, changes in outdoor temperature and humidity will affect the operating efficiency of building equipment. The differential evolution algorithm adjusts the dynamic operation and maintenance management plan based on real-time collected environmental data and equipment operation data. For example, when the outdoor temperature rises, the load of the air-conditioning system may increase. The differential evolution algorithm will adjust the operating parameters of the air-conditioning system according to this change, such as adjusting the cooling capacity, fan speed, etc., to ensure indoor comfort while reducing energy consumption.

[0054] The fine-tuned solution is synchronized to the operation and maintenance log database of the building information model. The operation and maintenance log database records all operation and maintenance operations and status changes of building equipment. The fine-tuned solution can ensure that the information in the database is consistent with the actual operation and maintenance situation. In this way, accurate historical data support can be provided during subsequent data analysis and troubleshooting.

[0055] In addition, it also includes the step of anomaly detection for the dynamic operation and maintenance management scheme. The isolation forest algorithm is used to identify abnormal nodes in the operation and maintenance operation sequence. The isolation forest algorithm determines the degree of isolation of data points by constructing a tree structure. In the operation and maintenance operation sequence, normal operation behaviors have certain patterns and rules, while abnormal operations are often quite different from normal patterns. The isolation forest algorithm can quickly identify these abnormal operation nodes. For example, in the startup operation sequence of the equipment, if there are abnormal situations such as too long startup time or too large startup current, the isolation forest algorithm can identify these operations as abnormal nodes.

[0056] The knowledge graph reasoning engine is used to analyze the associated impact range of abnormal nodes. The knowledge graph contains various relationships between building equipment, such as the composition relationship, connection relationship, control relationship, etc. of the equipment. After identifying the abnormal node, the knowledge graph reasoning engine can analyze other equipment and systems that may be affected by the abnormal node based on these relationships. For example, if a water pump fails, the knowledge graph reasoning engine can infer the range of water flow abnormalities that may be caused by the failure, as well as other equipment that may be affected, such as heat exchangers, cooling towers, etc., by analyzing the connection relationship between the water pump and pipes, valves, and other equipment.

[0057] Generate an exception handling plan and trigger the early warning module of the building information model. According to the associated impact range of the abnormal node, formulate a corresponding exception handling plan. For example, if an abnormality is found in the power system of a certain area, it may cause power outages for some equipment. The plan may include timely switching to backup power supplies, notifying maintenance personnel to carry out emergency repairs, and other measures. At the same time, the early warning module of the building information model is triggered to remind management personnel to handle abnormal situations in a timely manner through sound and light alarms, SMS notifications, etc., to ensure the safe and stable operation of the building.

[0058] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0059] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A full life cycle intelligent operation and maintenance management method driven by a building information model, characterized in that: The method comprises: Acquiring raw data of a building information model, and preprocessing the raw data to obtain standardized structural data; According to the standardized structure data, a preliminary operation and maintenance feature model is constructed based on a spatiotemporal graph convolutional network; Using the preliminary operation and maintenance feature model, dynamic parameter optimization is performed through a meta-learning framework to generate optimized operation and maintenance decision features; According to the optimized operation and maintenance decision characteristics, a set of intelligent operation and maintenance strategies for the entire life cycle is generated in combination with a multi-objective optimization algorithm; Align the intelligent operation and maintenance strategy set with the historical operation and maintenance data in time series, and verify the feasibility of the strategy through a dynamic time warping algorithm; Adjust the constraints of the multi-objective optimization algorithm based on the verification results, and iteratively update the intelligent operation and maintenance strategy set; Generate the final dynamic operation and maintenance management plan based on the updated intelligent operation and maintenance strategy set; Based on the dynamic operation and maintenance management solution, real-time operation and maintenance operations of building equipment are driven, and status data of the building information model is updated synchronously.

2. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 1, characterized in that: The preprocessing of the raw data comprises: Extract geometric properties, material properties and equipment association data from building information models; Fill missing values ​​and remove outliers on the extracted data to generate a structured data table; The structured data table is converted into tensor data of uniform dimension as standardized structured data through a standardized mapping algorithm.

3. The full life cycle intelligent operation and maintenance management method driven by building information model as claimed in claim 2, characterized in that: The construction of a preliminary operation and maintenance feature model based on the spatiotemporal graph convolutional network includes: The standardized structure data is input into the spatial convolution layer of the spatiotemporal graph convolutional network to capture the topological correlation features between devices; The temporal dependency of historical operation and maintenance data is modeled through the time convolution layer to generate spatiotemporal fusion features; The attention mechanism is used to assign weights to the spatiotemporal fusion features, and a preliminary operation and maintenance feature model containing key operation and maintenance nodes is output.

4. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 3 is characterized in that: The dynamic parameter optimization through the meta-learning framework includes: Define a set of meta-learning tasks, each of which corresponds to parameter adjustment requirements in different operation and maintenance scenarios; A model-agnostic meta-learning algorithm is used to quickly update the local parameters of the spatiotemporal graph convolutional network on the support set; The generalization performance after parameter update is evaluated through the query set, and the optimized operation and maintenance decision features that are suitable for multiple scenarios are screened out.

5. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 4, characterized in that: The intelligent operation and maintenance strategy set for the entire life cycle generated by combining the multi-objective optimization algorithm includes: Construct a multi-objective optimization function that includes energy cost, equipment life, and safety level; A non-dominated sorting genetic algorithm is used to solve the optimization function and generate a Pareto optimal strategy candidate set; Based on the fuzzy comprehensive evaluation method, the intelligent operation and maintenance strategy set that meets the current working conditions is selected from the candidate set.

6. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 5, characterized in that: The feasibility of the strategy verified by the dynamic time warping algorithm includes: Align the predicted operation sequence in the intelligent operation and maintenance strategy set with the historical success case sequence in the time dimension; Calculate the minimum path distance between aligned sequences as a quantitative indicator of strategy feasibility; Strategies that deviate from the threshold range are screened out and marked based on quantitative indicators.

7. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 6, characterized in that: The iterative update intelligent operation and maintenance strategy set includes: Adaptive penalty terms are introduced to dynamically weight the constraints of multi-objective optimization algorithms; Adjust the search direction of the optimization algorithm through the covariance matrix adaptive evolution strategy; Infeasible strategies are eliminated based on the marking results to generate an updated intelligent operation and maintenance strategy set.

8. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 1, characterized in that: After generating the final dynamic operation and maintenance management plan, the following steps are also included: Build a real-time status mapping model of digital twins based on IoT sensor data; The dynamic operation and maintenance management solution is fine-tuned online through the differential evolution algorithm to adapt to real-time environmental changes; Synchronize the fine-tuned plan to the operation and maintenance log database of the building information model.

9. The full life cycle intelligent operation and maintenance management method driven by building information model according to claim 1, characterized in that: It also includes the steps of anomaly detection for dynamic operation and maintenance management solutions: Use the isolation forest algorithm to identify abnormal nodes in the operation sequence; Analyze the associated impact scope of abnormal nodes through the knowledge graph reasoning engine; Generate exception handling plans and trigger the early warning module of the building information model.

10. A full life cycle intelligent operation and maintenance management system driven by a building information model, characterized in that: include: Data preprocessing module: used to obtain the original data of the building information model and preprocess the original data to obtain standardized structural data; Preliminary model building module: constructing a preliminary operation and maintenance feature model based on the spatiotemporal graph convolutional network according to the standardized structure data; Dynamic optimization module: using the preliminary operation and maintenance feature model, dynamic parameter optimization is performed through a meta-learning framework to generate optimized operation and maintenance decision features; Strategy generation module: based on the optimized operation and maintenance decision features, a set of intelligent operation and maintenance strategies for the entire life cycle is generated in combination with a multi-objective optimization algorithm; Strategy verification module: align the intelligent operation and maintenance strategy set with the historical operation and maintenance data in time series, and verify the feasibility of the strategy through a dynamic time warping algorithm; Iterative update module: adjusts the constraints of the multi-objective optimization algorithm based on the verification results, and iteratively updates the intelligent operation and maintenance strategy set; Solution output module: Generates the final dynamic operation and maintenance management solution based on the updated intelligent operation and maintenance strategy set; Operation and maintenance driving module: drives the real-time operation and maintenance of building equipment based on the dynamic operation and maintenance management solution, and synchronously updates the status data of the building information model.

Images