Coal-fired unit equipment predictive maintenance method and system based on domestic bim

Abstract

The application discloses a coal-fired unit equipment predictive maintenance method and system based on domestic BIM, relates to the technical field of predictive maintenance, and comprises the following steps: S1, in the domestic BIM platform, integrating multi-source heterogeneous data of a coal-fired unit equipment and a weld in the whole stage of design, construction and operation and maintenance, and realizing data tamper-proof storage through a block chain architecture; S2, quantifying the equipment degradation state and accurately predicting the remaining life; S3, reversely pushing to a design module or a construction simulation module through the domestic BIM platform. The application breaks the data fragmentation barrier of design, construction and operation and maintenance by constructing a full-life-cycle multi-stage data integration system and a cross-stage association rule mining mechanism, solves the pain points of static data storage and update lag of the domestic BIM platform, provides real-time, complete and accurate data support for predictive maintenance, and consolidates the core data basis of intelligent operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of predictive maintenance technology, and in particular to a predictive maintenance method and system for coal-fired power unit equipment based on domestically produced BIM. Background Technology

[0002] As core equipment in energy supply, the operational stability and maintenance effectiveness of coal-fired power units directly affect energy supply security and economic benefits. Due to the complex structure of the unit equipment, harsh operating conditions (high temperature, high pressure, high corrosion), and the fact that the entire life cycle covers multiple stages such as design, construction, and operation and maintenance, with diverse and highly interconnected data types at each stage, traditional maintenance methods often rely on periodic inspections or post-failure repairs. This results in high maintenance costs, delayed fault prediction, and fragmented data across stages, making it difficult to meet the needs of intelligent operation and maintenance.

[0003] With the development of digital technology, BIM (Building Information Modeling) technology has been gradually applied to the field of thermal power engineering. In particular, the maturity of domestically produced BIM technology has provided new technical support for the full life cycle management of coal-fired power units. Among existing technologies, domestically produced BIM technology has achieved basic applications such as visualization of design results, clash detection during construction, digital handover, and correlation query of operation and maintenance data. It can integrate multi-source data such as design drawings, construction records, and equipment ledgers, providing a data foundation for equipment management.

[0004] However, the data integrated by domestic BIM platforms is mostly statically stored, lacking in-depth mining of cross-stage data association rules. The data is difficult to dynamically evolve and update, and cannot provide real-time and accurate data support for predictive maintenance. Moreover, predictive maintenance models mostly rely on data from a single stage (such as only using operation and maintenance stage data), and do not fully integrate key influencing factors from the design and construction stages. Furthermore, the predictive models for weak links such as overall equipment and welds lack targeted optimization, resulting in limited prediction accuracy. Summary of the Invention

[0005] The purpose of this invention is to provide a predictive maintenance method and system for coal-fired power unit equipment based on domestically produced BIM, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a predictive maintenance method and system for coal-fired power unit equipment based on domestically produced BIM, comprising the following steps:

[0007] S1. In the domestic BIM platform, multi-source heterogeneous data of coal-fired unit equipment and welds are integrated throughout the design, construction and operation and maintenance stages. The data is stored tamper-proofly through a blockchain architecture. Based on machine learning models, cross-stage data association rules are mined to drive the dynamic self-evolution of domestic BIM model attributes and data dictionary, forming a self-evolving data chain throughout the entire life cycle.

[0008] S2. Using the self-evolving data chain, the degradation mechanism of coal-fired power unit equipment is integrated with data-driven algorithms to construct and iteratively update predictive maintenance models for the overall equipment and weld seams, quantify the degradation status of the equipment and accurately predict the remaining lifespan.

[0009] S3. When the output of the predictive maintenance model meets the preset feedback triggering conditions, the design defects or construction hazards identified during the maintenance process are automatically pushed back to the design module or construction simulation module through the domestic BIM platform, driving the automatic optimization of design parameters or construction schemes, and the optimization results are synchronously updated to the self-evolving data chain.

[0010] Preferably, step S1 specifically includes:

[0011] S11. Collect basic equipment data and weld-specific data according to the design, construction and operation and maintenance stages to form a data closed loop of static recording and dynamic monitoring, and bind it to the domestic BIM model through coding.

[0012] S12. Deploy multi-entity blockchain nodes, adopt sharded storage and hash index mode, store large-capacity data on distributed servers, and ensure the integrity and traceability of core data information through smart contracts.

[0013] Preferably, the method for forming the self-evolving data chain in step S1 includes:

[0014] S13. Extract the spatial, process, temporal, and decay characteristics of the equipment and weld seams to form a three-dimensional feature vector;

[0015] S14. Mining cross-stage data association rules through hybrid machine learning models, supporting online model updates;

[0016] S15. Ensure data quality through data missing completion and anomaly filtering, and dynamically update BIM model attributes and platform data dictionary based on association rules.

[0017] Preferably, the method for constructing the overall predictive maintenance model of the equipment in step S2 includes:

[0018] S21. Adopt a fusion framework of mechanism model and data-driven model, establish physical equations based on equipment degradation mechanism, train data-driven model with self-evolving data, dynamically adjust the weights of the two types of models, and focus on mechanism model in the early stage of operation and maintenance.

[0019] S22. By merging the model, the degradation rate and performance decline trend of equipment components are output and visualized using a domestically produced BIM platform.

[0020] Preferably, the method for constructing the weld-specific predictive maintenance model in step S2 includes:

[0021] S23. Construct a geometric twin and calibrate it using actual measurement data. Embed a temperature-stress-corrosion multi-field coupling simulation module and dynamically adjust the simulation boundary conditions.

[0022] S24. Based on the Paris formula in fracture mechanics, corrosion state parameters and fatigue accumulation parameters in the weld maintenance stage are introduced for correction, and a mapping relationship between weld defects and equipment failures is established.

[0023] Preferably, the specific method for predicting the remaining life of the weld in step S2 includes: using probabilistic simulation and reasoning methods, calculating the crack propagation rate based on the improved Paris formula, predicting the probability distribution of crack length and risk level, and outputting the remaining life range and warning threshold.

[0024] Preferably, step S2, after predicting the remaining lifetime, further includes:

[0025] S25. Establish a two-dimensional risk-cost assessment model to calculate the maintenance priority of equipment and welds;

[0026] S26. Generate maintenance operation paths, tool selection, and personnel planning based on domestically produced BIM models, verify the feasibility of the scheme through simulation and generate guidance documents.

[0027] Preferably, step S3 specifically includes:

[0028] S31. Set three types of triggering conditions: numerical threshold, trend threshold, and frequency threshold. Cross-stage feedback will be triggered when any one of the conditions is met.

[0029] S32. Push the design defects to the design module, automatically adjust the design parameters and generate optimized drawings, and update the platform standard model library after approval;

[0030] S33. Push potential construction hazards to the construction simulation module, adjust the construction simulation parameters and process instructions, and push them to the construction unit.

[0031] Preferably, the adaptation and assurance methods for the domestically produced BIM platform in steps S1-S3 include:

[0032] S01. Extract the geometric and attribute information of multi-format models through the format conversion interface and convert them into the platform's native format;

[0033] S02. Adopting an edge computing and cloud-coordinated architecture, real-time and non-real-time tasks are deployed separately to ensure the continuous and stable operation of the system;

[0034] S03. Encrypt cross-system data transmission through encryption algorithms and rely on blockchain to record data access behavior to achieve full traceability.

[0035] A predictive maintenance system for coal-fired power plant equipment based on domestically developed BIM includes a data acquisition and storage module, a data self-evolution module, a predictive model construction module, a closed-loop feedback optimization module, and a platform adaptation and support module. The modules work together to achieve predictive maintenance.

[0036] The data acquisition and evidence storage module is used to classify and collect multi-source data from equipment and weld seams throughout the entire process, and to achieve tamper-proof evidence storage through blockchain multi-nodes.

[0037] The data self-evolution module is used to extract multi-dimensional data features, mine association rules through hybrid models, drive the dynamic update of BIM models and data dictionaries, and form a self-evolutionary data chain.

[0038] The prediction model building module is used to build prediction models for the overall equipment and weld seams, quantify the degradation status and predict the remaining life, and generate maintenance priorities and customized solutions.

[0039] The closed-loop feedback optimization module is used to set feedback trigger conditions, push back defects or hidden dangers and drive design and construction optimization, and synchronously update the self-evolving data chain.

[0040] The platform adaptation and protection module is used to achieve model format compatibility, stable operation under high load, and data security management.

[0041] The technical effects and advantages of this invention are as follows:

[0042] (1) This invention breaks down the data silos between design, construction and operation and maintenance by constructing a multi-stage data integration system for the entire life cycle of coal-fired units and a cross-stage association rule mining mechanism, solves the pain points of static storage and lagging updates of data in domestic BIM platforms, provides real-time, complete and accurate data support for predictive maintenance, and consolidates the core data foundation for intelligent operation and maintenance.

[0043] (2) This invention integrates multi-source data throughout the entire life cycle and combines digital twin and multi-field coupling simulation technology to optimize the predictive maintenance model for weak links of the unit such as welds. It overcomes the shortcomings of traditional models that rely on single-stage data and lack specificity, significantly improves the quantitative accuracy of equipment degradation status and the accuracy of remaining life prediction, and realizes early prediction and accurate warning of faults.

[0044] (3) This invention generates accurate life prediction and customized maintenance solutions, and relies on the virtual inspection and equipment disassembly simulation functions of the domestic BIM platform to replace the passive mode of traditional periodic maintenance and post-fault maintenance, reduce the losses caused by over-maintenance and unplanned downtime, reduce maintenance time, material consumption and other comprehensive maintenance costs, and improve operation and maintenance efficiency and economic benefits.

[0045] (4) By establishing a cross-stage closed-loop feedback optimization mechanism, the present invention pushes back the design defects and construction hazards identified during the maintenance process to the early stage, drives the continuous optimization of design parameters and construction schemes, reduces the recurrence of similar problems, improves the operational stability and reliability of coal-fired units, and ensures the safety of energy supply. Attached Figure Description

[0046] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention, but do not constitute a limitation thereof. In the drawings:

[0047] Figure 1 This is a schematic flowchart of the maintenance method of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] This invention provides, for example Figure 1 The method for predictive maintenance of coal-fired power unit equipment based on domestically produced BIM, as shown, is characterized by the following steps:

[0050] S1. In the domestic BIM platform, multi-source heterogeneous data of coal-fired unit equipment and welds are integrated throughout the design, construction and operation and maintenance stages. The data is stored tamper-proofly through a blockchain architecture. Based on machine learning models, cross-stage data association rules are mined to drive the dynamic self-evolution of domestic BIM model attributes and data dictionary, forming a self-evolving data chain throughout the entire life cycle.

[0051] S2. By utilizing a self-evolving data chain, integrating the degradation mechanism of coal-fired power unit equipment with data-driven algorithms, a predictive maintenance model for the overall equipment and weld seams is constructed and iteratively updated to quantify the degradation status of the equipment and accurately predict its remaining lifespan.

[0052] S3. When the output of the predictive maintenance model meets the preset feedback triggering conditions, the design defects or construction hazards identified during the maintenance process are automatically pushed back to the design module or construction simulation module through the domestic BIM platform, driving the automatic optimization of design parameters or construction schemes, and the optimization results are synchronously updated to the self-evolving data chain.

[0053] Specifically, step S1, which involves the construction and self-evolution of lifecycle data, includes:

[0054] S11. Multi-source data classification and collection:

[0055] Equipment basic data acquisition: During the design phase, static data such as equipment material, design parameters, KKS code, and pipeline layout dimensions are collected; during the construction phase, process data such as installation deviation, concrete strength, and construction process records are collected; during the operation and maintenance phase, real-time operating data such as temperature, pressure, and vibration are accessed through the plant-level monitoring information system, and degradation data such as equipment defects, maintenance records, and fault information are collected simultaneously.

[0056] Weld-specific data acquisition: During the construction phase, welding robots equipped with high-frequency data acquisition modules collect welding current, voltage, speed, and arc temperature. Combined with digital non-destructive testing, defect features such as weld porosity, cracks, and slag inclusions are extracted. The data is then double-bound to the domestic BIM model through weld number and KKS code. During the operation and maintenance phase, high-temperature resistant miniature fiber optic sensors are deployed around key welds such as the four main boiler pipes and the main steam pipe to collect stress, strain, and temperature data. Combined with regular digital phased array testing and metal monitoring system data, a closed loop of static detection and dynamic monitoring data is formed.

[0057] S12. Blockchain Data Storage: Blockchain nodes are deployed in design institutes, construction units, operation and maintenance parties, equipment manufacturers, and owners respectively. A sharded storage and hash index mode is adopted to store large-capacity model files and video data on the distributed server of the domestic BIM platform. Only the data summary, timestamp, data source, and verification code are uploaded to the chain. The integrity of the data is verified through smart contracts and the entire process is traceable.

[0058] The method for forming the self-evolving data chain in step S1 includes:

[0059] S13. Multi-dimensional feature extraction: For core equipment such as boilers, steam turbines, and main transformers of coal-fired power units, as well as welds, the system automatically extracts spatial features from the design stage, process features from the construction stage, and temporal and decay features from the operation and maintenance stage, and combines them to form a three-dimensional feature vector of spatial-temporal-decay.

[0060] S14. Hybrid Algorithm Model Training: A hybrid model of LSTM, attention mechanism and online gradient descent (OGD) is adopted. The attention mechanism is used to assign weights to key related data such as welding defects and installation deviations. The OGD algorithm is used to realize online model updates. When new operation and maintenance data is added, only local parameters of the model are updated to ensure the real-time self-evolution of data.

[0061] S15. Data Quality and Dynamic Model Updates: Generative Adversarial Networks (GANs) are introduced to fill in missing data values. The 3σ criterion is used to verify and filter abnormal data in relation to equipment status. A data association rule-model attribute mapping table is established. When the hybrid algorithm identifies a new association rule, the key parameters are automatically written into the corresponding component of the domestic BIM model and the platform data dictionary is updated. Association rules that are repeatedly verified are upgraded to mandatory verification rules, which trigger automatic model warnings in the subsequent unit design stage.

[0062] The method for constructing the overall predictive maintenance model for equipment in step S2 includes:

[0063] S21. Fusion Model Framework Construction: A fusion framework combining mechanistic and data-driven models is adopted. Physical equations are established based on the degradation mechanisms of corrosion, wear, and fatigue in coal-fired power unit equipment. The equation coefficients are optimized through self-evolving data fitting. A gradient boosting tree (XGBoost) model is trained based on self-evolving data, and the weights of the two types of models are dynamically adjusted. In the early stage of operation and maintenance (e.g., when the cumulative operating time is less than 20% of the total design life), the mechanistic model is mainly used for prediction, and the data-driven model provides auxiliary correction. As operation and maintenance data accumulates, the weight of the data-driven model is gradually increased. In the later stage of operation and maintenance (e.g., when the cumulative operating time exceeds 60% of the total design life), the data-driven model is mainly used for prediction, and the mechanistic model is used to verify the rationality of the prediction results.

[0064] S22. Degradation Status Quantification: By integrating the model output, the degradation rate of each component of the equipment and the degradation trend of key performance parameters are displayed. Combined with the three-dimensional visualization function of the domestic BIM platform, the overall degradation status distribution of the equipment is displayed intuitively.

[0065] The method for constructing the weld-specific predictive maintenance model in step S2 includes:

[0066] S23. Digital twin modeling of weld seams:

[0067] Geometric twin construction: Based on the domestic BIM platform, the theoretical dimensions of the weld are imported to establish a basic model. During the construction phase, the actual geometric data is obtained through laser scanning (accuracy ±0.1mm) to correct deviations. During the maintenance phase, the geometric state of internal defects of the weld is updated through industrial CT scanning. It supports macro-micro two-level scaling display.

[0068] Physical twin simulation: A temperature-stress-corrosion multi-field coupled simulation module is embedded in the domestic BIM platform. The sensor data collected during the operation and maintenance phase and the flue gas flow characteristics are input to simulate the temperature distribution, stress concentration area and corrosion rate distribution of welds and surrounding components. The simulation boundary conditions are dynamically adjusted in combination with the unit operating parameters.

[0069] S24. Construction of an improved crack propagation model: Based on the Paris formula of fracture mechanics, corrosion state parameters and fatigue accumulation parameters of the weld operation and maintenance stage are introduced for correction; the corrosion state parameters are taken from the weld corrosion depth and corrosion area data recorded by the metal monitoring system, and the fatigue accumulation parameters are calculated by statistically analyzing the number of unit start-ups and shutdowns, load fluctuation amplitude and duration, and directly establishing the mapping relationship between weld defects and equipment leakage, cracking and other faults.

[0070] The specific method for predicting the remaining life of the weld in step S2 includes: using Monte Carlo simulation and Bayesian inference, calculating the crack propagation rate of the weld based on the improved Paris formula, predicting the probability distribution of crack length in different time periods, quantifying the risk level of crack exceeding the standard, and outputting the remaining life range of the weld and the fault warning threshold.

[0071] Step S2, following the prediction of remaining lifetime, also includes the generation of maintenance strategies:

[0072] S25. Maintenance Priority Ranking: Establish a two-dimensional risk-cost assessment model. The risk dimension covers the importance of equipment and welds, the scope of failure impact, and remaining life. The cost dimension includes maintenance man-hours, material costs, and downtime losses. The maintenance priority is calculated using the Analytic Hierarchy Process (AHP).

[0073] S26. Customized maintenance plan generation: Based on the domestic BIM model, the system automatically generates maintenance operation paths, tool selection, and personnel positioning plans. The system also performs pre-runs of the maintenance process through construction simulation to verify the safety and feasibility of the plan and simultaneously generates maintenance process guidelines.

[0074] Specifically, the cross-stage closed-loop feedback and maintenance optimization in step S3 includes:

[0075] S31. Feedback trigger condition setting: Set three types of trigger conditions: numerical threshold (the cumulative maintenance cost exceeds the preset proportion of the total equipment price), trend threshold (the equipment degradation rate is significantly accelerated due to similar design or construction problems), and frequency threshold (the same type of hidden danger occurs repeatedly under the same design or construction conditions to reach a preset number of times). Cross-stage feedback is triggered when any of the conditions is met.

[0076] S32. Design Phase Feedback Optimization: Through the secondary development interface of the domestic BIM platform, design defects associated with maintenance records are pushed to the design module. The parameter optimization engine is called to automatically adjust design parameters such as pipeline spacing and maintenance space size, generate optimized drawing versions, and update the platform's standard model library after approval.

[0077] S33. Construction Phase Feedback Optimization: Potential construction hazards such as welding quality and installation accuracy discovered during maintenance are pushed to the construction simulation module, automatically adjusting the pre-simulation parameters for critical and major construction projects and the construction process guidelines, and pushing them to the construction unit through the visualization technology briefing function.

[0078] Among them, the adaptation and assurance methods for the domestically produced BIM platform in steps S1-S3 include:

[0079] S01, Model Format Compatibility Processing: Embeds a unified model format conversion interface to automatically extract geometric and attribute information from multiple model formats such as pmodel, rvt, rvm, and dgn, and convert them into the native format of the domestic BIM platform to ensure data consistency.

[0080] S02. High-load operation guarantee: Adopting an edge computing and cloud collaborative architecture, the real-time data acquisition and fault early warning calculation are deployed on the local edge nodes of the power plant, while non-real-time tasks such as data storage, model training, and cross-stage feedback are deployed on the cloud-based domestic BIM platform, which is compatible with the domestic Kylin operating system V10 and above, ensuring the system operates stably 24 / 7.

[0081] S03. Data security management: The domestic SM4 encryption algorithm is used to encrypt data transmitted across systems. All data access behaviors are recorded through blockchain to achieve full traceability of data operations and prevent data tampering and unauthorized access.

[0082] A predictive maintenance system for coal-fired power plant equipment based on domestically developed BIM includes:

[0083] The data acquisition and storage module, the data self-evolution module, the prediction model construction module, the closed-loop feedback optimization module, and the platform adaptation and guarantee module work together to implement a predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM.

[0084] The data acquisition and storage module is used to collect multi-source heterogeneous data of coal-fired unit equipment and welds throughout the design, construction and operation and maintenance stages. It uses a blockchain architecture to deploy multiple nodes to achieve tamper-proof data storage. The multi-source heterogeneous data includes equipment basic static data, construction process dynamic data, operation and maintenance degradation data and weld-specific inspection and monitoring data.

[0085] The data self-evolution module is used to extract multi-dimensional data features and build a hybrid algorithm model, mine cross-stage data association rules, drive the dynamic update of the attributes and data dictionary of the domestic BIM model, form a self-evolving data chain throughout the entire life cycle, and also has the functions of data missing value completion and abnormal data filtering.

[0086] The prediction model building module is used to build and iteratively update the overall equipment prediction model and the weld-specific prediction model based on the self-evolving data chain, which integrates the degradation mechanism of coal-fired power unit equipment with data-driven algorithms. It can quantify the degradation status of the equipment and accurately predict the remaining life, and can also generate maintenance priority ranking and customized maintenance plans.

[0087] The closed-loop feedback optimization module is used to set feedback trigger conditions. When the prediction result meets the trigger conditions, it automatically pushes the identified design defects or construction hazards back to the design module or construction simulation module, drives the automatic optimization of design parameters or construction schemes, and synchronizes the optimization results to the self-evolving data chain.

[0088] The platform adaptation and protection module is used to achieve compatible conversion of multiple format models. It ensures stable operation of the system under high load through edge computing and cloud collaborative architecture, and adopts domestic encryption algorithms and data access traceability mechanisms to achieve data security management.

[0089] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A predictive maintenance method for coal-fired power unit equipment based on domestically developed BIM, characterized in that, Includes the following steps: S1. In the domestic BIM platform, multi-source heterogeneous data of coal-fired unit equipment and welds are integrated throughout the design, construction and operation and maintenance stages. The data is stored tamper-proofly through a blockchain architecture. Based on machine learning models, cross-stage data association rules are mined to drive the dynamic self-evolution of domestic BIM model attributes and data dictionary, forming a self-evolving data chain throughout the entire life cycle. S2. Using the self-evolving data chain, the degradation mechanism of coal-fired power unit equipment is integrated with data-driven algorithms to construct and iteratively update predictive maintenance models for the overall equipment and weld seams, quantify the degradation status of the equipment and accurately predict the remaining lifespan. S3. When the output of the predictive maintenance model meets the preset feedback triggering conditions, the design defects or construction hazards identified during the maintenance process are automatically pushed back to the design module or construction simulation module through the domestic BIM platform, driving the automatic optimization of design parameters or construction schemes, and the optimization results are synchronously updated to the self-evolving data chain.

2. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 1, characterized in that, Step S1 specifically includes: S11. Collect basic equipment data and weld-specific data according to the design, construction and operation and maintenance stages to form a data closed loop of static recording and dynamic monitoring, and bind it to the domestic BIM model through coding. S12. Deploy multi-entity blockchain nodes, adopt sharded storage and hash index mode, store large-capacity data on distributed servers, and ensure the integrity and traceability of core data information through smart contracts.

3. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 1, characterized in that, The method for forming the self-evolving data chain in step S1 includes: S13. Extract the spatial, process, temporal, and decay characteristics of the equipment and weld seams to form a three-dimensional feature vector; S14. Mining cross-stage data association rules through hybrid machine learning models, supporting online model updates; S15. Ensure data quality through data missing completion and anomaly filtering, and dynamically update BIM model attributes and platform data dictionary based on association rules.

4. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 3, characterized in that, The method for constructing the overall predictive maintenance model for equipment in step S2 includes: S21. Adopt a fusion framework of mechanism model and data-driven model, establish physical equations based on equipment degradation mechanism, train data-driven model with self-evolving data, dynamically adjust the weights of the two types of models, and focus on mechanism model in the early stage of operation and maintenance. S22. By merging the model, the degradation rate and performance decline trend of equipment components are output and visualized using a domestically produced BIM platform.

5. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 3, characterized in that, The method for constructing the weld-specific predictive maintenance model in step S2 includes: S23. Construct a geometric twin and calibrate it using actual measurement data. Embed a temperature-stress-corrosion multi-field coupling simulation module and dynamically adjust the simulation boundary conditions. S24. Based on the Paris formula in fracture mechanics, corrosion state parameters and fatigue accumulation parameters in the weld maintenance stage are introduced for correction, and a mapping relationship between weld defects and equipment failures is established.

6. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 5, characterized in that, The specific method for predicting the remaining life of the weld in step S2 includes: using probabilistic simulation and reasoning methods, calculating the crack propagation rate based on the improved Paris formula, predicting the probability distribution of crack length and risk level, and outputting the remaining life range and warning threshold.

7. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 5, characterized in that, Step S2, following the prediction of remaining lifetime, also includes: S25. Establish a two-dimensional risk-cost assessment model to calculate the maintenance priority of equipment and welds; S26. Based on the domestically produced BIM model, generate maintenance operation paths, tool selection and personnel planning, verify the feasibility of the solution through simulation and generate guidance documents.

8. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 1, characterized in that, Step S3 specifically includes: S31. Set three types of triggering conditions: numerical threshold, trend threshold, and frequency threshold. Cross-stage feedback will be triggered when any one of the conditions is met. S32. Push the design defects to the design module, automatically adjust the design parameters and generate optimized drawings, and update the platform standard model library after approval; S33. Push potential construction hazards to the construction simulation module, adjust the construction simulation parameters and process instructions, and push them to the construction unit.

9. The predictive maintenance method for coal-fired power unit equipment based on domestically produced BIM according to claim 1, characterized in that, Steps S1-S3, the adaptation and assurance methods for the domestically produced BIM platform include: S01. Extract the geometric and attribute information of multi-format models through the format conversion interface and convert them into the platform's native format; S02. Adopting an edge computing and cloud-coordinated architecture, real-time and non-real-time tasks are deployed separately to ensure the continuous and stable operation of the system; S03. Encrypt cross-system data transmission through encryption algorithms and rely on blockchain to record data access behavior to achieve full traceability.

10. A predictive maintenance system for coal-fired power unit equipment based on domestically developed BIM, characterized in that, It includes a data acquisition and storage module, a data self-evolution module, a prediction model construction module, a closed-loop feedback optimization module, and a platform adaptation and guarantee module. The modules work together to achieve the method described in any one of claims 1-9. The data acquisition and evidence storage module is used to classify and collect multi-source data from equipment and weld seams throughout the entire process, and to achieve tamper-proof evidence storage through blockchain multi-nodes. The data self-evolution module is used to extract multi-dimensional data features, mine association rules through hybrid models, drive the dynamic update of BIM models and data dictionaries, and form a self-evolutionary data chain. The prediction model building module is used to build prediction models for the overall equipment and weld seams, quantify the degradation status and predict the remaining life, and generate maintenance priorities and customized solutions. The closed-loop feedback optimization module is used to set feedback trigger conditions, push back defects or hidden dangers and drive design and construction optimization, and synchronously update the self-evolving data chain. The platform adaptation and protection module is used to achieve model format compatibility, stable operation under high load, and data security management.

Images