Building equipment intelligent diagnosis and energy efficiency optimization method based on digital twinning

Abstract

This invention provides a digital twin-based intelligent diagnosis and energy efficiency optimization method for building equipment, relating to the field of intelligent operation and maintenance and energy-saving technology for building equipment. The method includes: constructing a digital twin that integrates geometric, mechanistic, and data-driven models; synchronizing physical equipment operation and environmental data in real time; calculating the comprehensive operational degradation degree reflecting the equipment's health status in the twin model; performing fault diagnosis and lifespan prediction based on this degradation degree; simulating different control strategies in conjunction with the equipment's health status and calculating dynamic energy efficiency optimization margins; selecting the optimal control strategy with the goal of maximizing this margin and issuing it for execution; and adaptively correcting the digital twin model based on physical feedback data. This invention achieves quantitative assessment of the health status of building equipment and accurate early warning of faults, achieving synergistic optimization of energy efficiency improvement and equipment lifespan extension, and ensuring the continuous reliability of diagnostic and optimization decisions throughout the entire lifecycle through model self-correction.

Description

Technical Field

[0001] This invention relates to the field of technology, and in particular to a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins. Background Technology

[0002] In modern building operations, the stable and efficient operation of critical building equipment such as HVAC systems, elevators, and water pump units is crucial for ensuring a comfortable building environment and achieving energy conservation and emission reduction goals. These devices operate continuously for extended periods, and their performance gradually deteriorates due to component wear, scaling, and aging. This not only leads to increased energy consumption but can also cause sudden malfunctions, resulting in maintenance interruptions and economic losses. Therefore, effective condition monitoring, fault warning, and energy efficiency optimization of building equipment have become urgent needs in the fields of smart buildings and green operation and maintenance.

[0003] Currently, health management of building equipment mainly relies on traditional monitoring and control systems and routine maintenance. A common practice is to install a limited number of sensors on the equipment to monitor key parameters such as temperature, pressure, and current using threshold values. When the measured value exceeds a preset safety limit, the system triggers an alarm, prompting maintenance personnel to investigate on-site. This method suffers from significant lag, only issuing alarms after a fault has occurred or performance has severely degraded, failing to achieve early fault prediction and preventative maintenance. Furthermore, threshold settings are often based on experience, resulting in insufficient sensitivity and accuracy for slowly occurring performance degradation or complex faults involving the interaction of multiple parameters, leading to frequent missed and false alarms.

[0004] In terms of energy efficiency optimization, existing technologies largely rely on energy-saving strategies based on fixed rules or static equipment performance curves. For example, adjusting equipment operating frequency according to a fixed schedule or simple temperature difference feedback. However, the actual operating efficiency of equipment is closely related to its real-time load, environmental conditions, and its own health status. If a water pump with slight wear is still optimized and controlled according to the efficiency curve of a new pump, it will not only fail to achieve the expected energy-saving effect, but may also exacerbate equipment wear and even induce failure. Existing methods lack quantitative assessment of the real-time health status of equipment, resulting in a disconnect between energy efficiency optimization strategies and the actual capacity of the equipment, making it difficult to tap the maximum energy-saving potential while ensuring equipment safety and lifespan.

[0005] In recent years, digital twin technology has provided new ideas for equipment management by constructing virtual mappings of physical entities. However, existing digital twins applied to building equipment mostly focus on 3D visualization and operational data monitoring. Their models are often a combination of geometric models and simple data-driven approaches, lacking in-depth mechanistic models that describe the inherent physical laws of the equipment. This "shallow" digital twin is weak in terms of state extrapolation, root cause analysis of failures, and "hypothetical" simulation optimization. Once the model is established, it remains fixed and cannot evolve with the performance changes of the physical equipment. This leads to an increasing deviation between simulation and reality over time, ultimately rendering simulation-based diagnostics and optimization suggestions unreliable.

[0006] In summary, existing technologies face several prominent bottlenecks when addressing intelligent operation and maintenance of building equipment: fault diagnosis relies on post-event alarms and experience-based judgment, failing to achieve early and accurate predictive maintenance; energy efficiency optimization strategies are static or semi-static, failing to deeply couple with the dynamically changing health status of equipment, posing a risk of poor optimization results or damage to equipment lifespan; even with the introduction of the digital twin concept, the models often lack sufficient depth, physical mechanism support, and self-updating capabilities, making it difficult to support high-confidence closed-loop optimization. Therefore, there is an urgent need for an integrated intelligent method that can deeply integrate equipment mechanisms, perceive equipment health in real time, and make adaptive simulation and optimization decisions accordingly. Summary of the Invention

[0007] To address the technical problems in existing technologies, such as lagging and inaccurate fault diagnosis of building equipment, disconnect between energy efficiency optimization strategies and real-time health status of equipment, and the shallow, rigid, and unadaptive nature of existing digital twin models, which prevent the implementation of predictive maintenance and dynamic energy efficiency closed-loop optimization based on high-reliability simulation, this invention provides a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins.

[0008] The technical solution provided by this invention is as follows: This invention provides a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins, comprising: S1. Construct a digital twin model of the target building equipment; the digital twin model includes a geometric model consistent with the geometric structure of the physical entity, a mechanism model describing the physical characteristics and operating laws of the equipment, and a data-driven model trained based on historical operating data; S2. Real-time data collection of building equipment operation status and environmental parameters is achieved through a sensor network deployed on the physical entity, and the data is synchronized to the digital twin model. S3. In the digital twin model, based on real-time synchronized data and the mechanism model and / or data-driven model, real-time simulation and deduction of the equipment operating status are performed, and characteristic parameters reflecting the current health status of the equipment are calculated. The characteristic parameters include at least one comprehensive operating deterioration degree. S4. Based on the overall operational degradation, diagnose the potential fault types of the equipment, locate the faulty components, and assess the remaining service life. S5. In the digital twin model, the current environmental parameters, equipment load requirements and the comprehensive operational degradation are combined to simulate the equipment operation process under different control strategies, and the energy efficiency evaluation parameters under each strategy are calculated. The energy efficiency evaluation parameters include at least one dynamic energy efficiency optimization margin. S6. With the goal of maximizing the dynamic energy efficiency optimization margin, select the optimal control strategy from the simulated control strategies. S7. Convert the optimal control strategy into specific control commands and send them to the control system of the physical building equipment to optimize its operating efficiency; S8. After the physical entity executes the control command, continuously collect its actual operation feedback data and compare it with the expected operation data of the digital twin model. Based on the comparison results, adaptively correct the digital twin model.

[0009] The beneficial effects of the technical solution provided by this invention include at least the following: (1) In this invention, a deeply integrated digital twin that combines geometry, mechanism, and data-driven models is constructed, and the characteristic parameter of comprehensive operational degradation is innovatively proposed and calculated, thereby achieving a quantitative, comprehensive, and sensitive assessment of the health status of building equipment. This method not only focuses on the instantaneous deviation of multiple key operating parameters, but also captures the abnormal fluctuation trend of parameters through integral calculation, thus enabling it to keenly perceive early, slowly changing performance degradation signs that are ignored by traditional threshold alarms. Based on this comprehensive index, fault diagnosis, combined with the contribution analysis of a pre-established fault feature library and mechanism model, can accurately identify fault modes from massive amounts of data, precisely locate potential faulty components, and predict remaining service life. This transforms the operation and maintenance mode from post-maintenance and periodic maintenance to condition-based predictive maintenance, significantly improving the initiative and accuracy of operation and maintenance, and effectively avoiding unplanned downtime.

[0010] (2) In this invention, the real-time health status of the equipment, i.e., the overall operational degradation, is creatively introduced as a key input into the energy efficiency optimization decision-making process, and a dynamic energy efficiency optimization margin is defined as an evaluation index. This index, while measuring the energy efficiency improvement potential of different control strategies, simultaneously deducts the potential risk costs caused by the decline in equipment health and simulated load rate. This means that the numerous "hypothetical" simulations and strategy screenings conducted in the digital twin are no longer seeking the theoretically optimal energy efficiency point, but rather finding the safest and most sustainable energy efficiency optimal solution under the current actual health condition of the equipment. This resolves the contradiction between traditional energy efficiency optimization strategies and the actual capacity of the equipment, effectively ensuring the operational safety and service life of the equipment while deeply exploring energy-saving potential, thus achieving synergistic optimization between energy efficiency improvement and equipment longevity.

[0011] (3) In this invention, a complete closed loop from virtual decision-making to physical execution and feedback correction is established. After the optimal control strategy is executed by the physical entity, the system continuously compares the actual operating data with the expected data of the digital twin model. Once a significant deviation is detected, the model self-correction process is automatically initiated. This process can intelligently determine whether the deviation stems from inaccurate mechanistic model or data-driven model degradation, and perform targeted calibration using parameter identification or incremental learning respectively. This mechanism makes the digital twin model no longer a static, unchanging "digital portrait," but a "living entity" that can continuously evolve with the performance degradation of the physical entity, environmental changes, and even updates to the control strategy. It ensures that the digital twin model maintains high fidelity throughout its entire life cycle, providing a continuous and reliable foundation for the state assessment, fault diagnosis, and energy efficiency optimization it supports, and truly realizing a closed loop of intelligent operation and maintenance based on digital twin and sustainable trust. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins, provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the process of constructing a digital twin model of a target building equipment in a digital twin-based intelligent diagnosis and energy efficiency optimization method for building equipment provided in an embodiment of the present invention. Figure 3 A flowchart illustrating step S4 in a digital twin-based intelligent diagnosis and energy efficiency optimization method for building equipment, as provided in an embodiment of the present invention. Figure 4 A flowchart illustrating step S6 in a digital twin-based intelligent diagnosis and energy efficiency optimization method for building equipment, as provided in an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the process of adaptively correcting the digital twin model based on comparison results in a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins, provided in an embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0015] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0016] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0017] In embodiments of the present invention, sometimes the subscript is as follows: It may be mistakenly written as a non-subscript form such as W1. When the distinction is not emphasized, the meaning they express is the same.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] Reference manual attached Figure 1 The diagram illustrates a flowchart of a digital twin-based intelligent diagnosis and energy efficiency optimization method for building equipment, provided by an embodiment of the present invention.

[0020] This invention provides a method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins. The processing flow may include the following steps: S1. Construct a digital twin model of the target building equipment; the digital twin model includes a geometric model consistent with the physical entity's geometry, a mechanism model describing the equipment's physical characteristics and operating rules, and a data-driven model trained based on historical operating data.

[0021] Step S1 involves constructing a digital twin model of the target building equipment. This digital twin model is a multi-dimensional, multi-physics, and multi-scale virtual mapping entity, consisting of three core parts. The first part is a geometric model, which accurately reproduces the spatial structure, shape, size, and assembly relationships of the physical equipment. The second part is a mechanistic model, built upon fundamental physical laws such as mass conservation, energy conservation, and momentum conservation, as well as the equipment's unique working principles. This model can describe the internal state changes and energy conversion processes of the equipment in the form of mathematical equations. The third part is a data-driven model, trained using machine learning on massive amounts of historical operational data. This model can capture and learn complex characteristics and degradation patterns during equipment operation that are difficult to describe using explicit physical equations. These three models are integrated and coupled within a unified digital space to form a comprehensive digital twin capable of real-time mapping, interactive analysis, and dynamic prediction.

[0022] S2. Real-time data collection of building equipment operation status and environmental parameters is achieved through a sensor network deployed on physical entities, and the data is synchronized to the digital twin model.

[0023] Step S2 involves collecting real-time operational status data and environmental parameter data of building equipment through a sensor network deployed on the physical entity, and synchronizing the data to the digital twin model. The sensor network is deployed at key locations according to the monitoring needs of the equipment, and the collected operational status data directly reflects the working condition of the equipment. Simultaneously, environmental sensors collect relevant parameters of the environment in which the equipment is located. All collected data is aggregated and preliminarily processed through IoT gateways or edge computing devices, and then transmitted in real-time via wired or wireless communication networks in a low-latency, highly reliable manner, and synchronously injected into the digital twin model, ensuring that the virtual model and the physical entity remain synchronized at the data level.

[0024] S3. In the digital twin model, based on real-time synchronized data and mechanism models and / or data-driven models, real-time simulation and deduction of equipment operating status are performed, and characteristic parameters reflecting the current health status of the equipment are calculated. The characteristic parameters include at least one comprehensive operating deterioration degree.

[0025] Step S3 involves performing real-time simulation and deduction of the equipment's operating status within the digital twin model, based on real-time synchronized data and a mechanistic model and / or a data-driven model, to calculate characteristic parameters reflecting the equipment's current health status. After receiving real-time data, the digital twin model drives the mechanistic model to calculate the theoretical state under the current operating conditions, while simultaneously utilizing the data-driven model for state identification and anomaly detection. By deeply fusing and comparing real-time data with model simulation data, characteristic parameters characterizing the overall performance degradation and health status of the equipment can be calculated. These characteristic parameters include at least a comprehensive operational degradation degree, a quantitative indicator that integrates multi-parameter deviations and abnormal fluctuation time-series characteristics, comprehensively and sensitively reflecting the degree to which the equipment deviates from its healthy baseline state.

[0026] S4. Based on the overall operational degradation, diagnose the potential fault types of the equipment, locate the faulty components, and assess the remaining service life.

[0027] Step S4 involves diagnosing potential fault types, locating faulty components, and assessing remaining service life based on the overall operational degradation degree. The diagnostic process is triggered when the overall operational degradation degree exceeds the normal threshold. First, based on the magnitude and trend of the degradation degree, a pre-defined fault mode knowledge base is used to preliminarily determine the possible fault types of the equipment. Then, utilizing the simulation capabilities of the mechanistic model in the digital twin model, the contribution of each submodule or component parameter to the total overall operational degradation degree is analyzed to accurately locate the component or subsystem most likely to experience an anomaly. Finally, combining the time-series prediction function of the data-driven model, based on current and historical degradation degree data, the development trend is extrapolated to predict the remaining time before equipment performance degrades to an unacceptable level or functional failure occurs.

[0028] S5. In the digital twin model, the equipment operation process under different control strategies is simulated by combining the current environmental parameters, equipment load requirements and overall operational degradation. The energy efficiency evaluation parameters under each strategy are calculated. The energy efficiency evaluation parameters include at least one dynamic energy efficiency optimization margin.

[0029] Step S5 involves simulating the equipment operation process under different control strategies within the digital twin model, taking into account current environmental parameters, equipment load requirements, and overall operational degradation. Energy efficiency assessment parameters are then calculated for each strategy. While ensuring the equipment meets current process load requirements, the digital twin model uses the currently synchronized environmental parameters and the calculated equipment health state (overall operational degradation) as boundary conditions. Subsequently, a series of possible equipment control strategies are automatically generated or invoked in the virtual space, such as adjusting operating frequency, changing valve opening, and optimizing start-stop logic. For each strategy, the digital twin model performs a rapid, full-lifecycle simulation of the operation process and calculates the overall energy efficiency performance of the equipment system under that strategy. The energy efficiency assessment parameters include at least one dynamic energy efficiency optimization margin, which considers not only the potential for energy efficiency improvement but also the constraint of the current equipment health state on achieving that potential.

[0030] S6. With the goal of maximizing the dynamic energy efficiency margin, the optimal control strategy is selected from the simulated control strategies.

[0031] Step S6 aims to maximize the dynamic energy efficiency margin by selecting the optimal control strategy from the simulated control strategies. For all simulated control strategies, the system ranks them using the calculated dynamic energy efficiency margin as the core evaluation index. The selection process prioritizes maximizing this margin, aiming to find the control scheme that achieves the best potential for energy efficiency improvement under the current equipment health condition. Simultaneously, the selection process must ensure that candidate strategies meet all hard constraints for safe equipment operation and the specific requirements of the production process. Finally, the system selects the control strategy that maximizes the dynamic energy efficiency margin from the feasible strategy set, determining it as the optimal control strategy to be executed at the current moment.

[0032] S7. Convert the optimal control strategy into specific control commands and send them to the control system of the physical building equipment to optimize its operating efficiency.

[0033] Step S7 involves converting the optimal control strategy into specific control commands and sending them to the control system of the physical building equipment to optimize its operational energy efficiency. The optimal control strategy generated by the digital twin system is a high-level logic and parameter setting. This step requires parsing and converting this strategy into low-level control commands that the physical equipment's field controllers or actuators can directly recognize and execute. These commands are typically in the form of standard control signals and are sent to the target equipment's control system via industrial communication networks or control buses. After receiving the commands, the control system drives the relevant actuators to perform actions, thereby implementing the strategy optimized in the digital space on the physical entity and achieving real-time optimization of operational energy efficiency.

[0034] S8. After the physical entity executes the control command, continuously collect its actual operation feedback data and compare it with the expected operation data of the digital twin model. Based on the comparison results, adaptively correct the digital twin model.

[0035] Step S8 involves continuously collecting actual operational feedback data from the physical entity after it executes control commands, comparing it with the expected operational data of the digital twin model, and adaptively correcting the digital twin model based on the comparison results. After the optimal control strategy is implemented, the system continuously collects actual operational feedback data from the device under new control commands through a sensor network. This feedback data is input into the digital twin model and compared and analyzed in real time with the expected operational data generated by the model when the control command was issued. If a significant and persistent deviation is found in the comparison, it indicates that the prediction accuracy of the digital twin model has decreased. At this time, the system will automatically activate the model correction mechanism, analyze the source of the deviation, and perform targeted calibration and updates to the parameters of the mechanistic model or the structure of the data-driven model, so that the digital twin model can continuously evolve with the state evolution of the physical entity or changes in the environment, maintaining the accuracy of its prediction and optimization capabilities.

[0036] In one possible implementation, such as Figure 2 As shown, step S1, constructing a digital twin model of the target building equipment, specifically includes: S101. Based on the device's CAD drawings or 3D scanned point cloud data, construct its high-fidelity geometric model; S102. Based on the equipment's design parameters, physical laws, and control logic, construct a mechanism model describing its energy conversion, mass transfer, and dynamic characteristics; S103. Collect historical operating data of the equipment under normal operation, multiple fault modes and different loads. After preprocessing the historical operating data, train a data-driven model for state recognition and trend prediction. S104. The geometric model, mechanistic model and data-driven model are fused and integrated under a unified spatiotemporal reference to form an interactive, computable and updatable digital twin model.

[0037] The process of constructing a digital twin model of the target building equipment in step S1 includes the following steps.

[0038] The implementation process of S101 is as follows: export the three-dimensional model file of the equipment using a computer-aided design system, or scan the field equipment using a three-dimensional laser scanner and generate point cloud data; then use three-dimensional modeling software to process these basic data and build a high-fidelity geometric model that includes the outer shell, internal key components and connecting pipes or lines. This model supports scaling, rotation and sectioning for viewing.

[0039] The implementation process of S102 is as follows: Based on the design manual and technical drawings provided by the equipment manufacturer, key parameters such as rated power, efficiency curve, heat exchange area, and pipeline resistance characteristics are extracted; based on these parameters, the principles of thermodynamics, fluid mechanics, and electrical machinery are applied to build a mechanism model in the simulation software environment that can be described by differential equations, algebraic equations, or state equations. This model can receive control signals and boundary conditions and output various state variables.

[0040] The implementation process of S103 is as follows: export historical operating data for at least one full year from the building equipment management system, or collect data specifically by temporarily deploying sensors; the data needs to cover typical operating conditions in summer and winter, different load rates, and the time periods before and after known failures; after cleaning the data using data preprocessing tools, select feature variables, and use machine learning platforms to train algorithm models such as random forests, gradient boosting trees, or one-dimensional convolutional neural networks to establish a mapping relationship from operating data to health status or failure modes.

[0041] The implementation process of S104 is as follows: Using a professional digital twin development platform or industrial IoT platform, the geometric model files, mechanism model scripts, and trained data-driven models are imported and registered; a data bus is configured in the platform, the association mapping rules between component objects in the geometric model and state variables in the mechanism model, as well as the input and output of the data-driven model, are defined, and a unified simulation clock step and data exchange protocol are set, thereby completing the model integration and forming a composite model that can interact with external data and services through API interfaces.

[0042] In one possible implementation, in step S2, the operating status data includes at least voltage, current, power, speed, temperature, pressure, flow rate, and vibration amplitude; the environmental parameter data includes at least ambient temperature and humidity.

[0043] The operational status data collected in step S2 is obtained in the following ways: voltage and current signals are acquired using clamp-on ammeters or voltage transformers; power signals can be directly measured by a power meter or calculated from voltage and current; rotational speed is acquired using an encoder or tachogenerator; temperature is measured using a PT100 or thermocouple sensor; pressure is acquired using a piezoresistive or piezoelectric pressure transmitter; flow rate is measured using an electromagnetic or ultrasonic flow meter; vibration amplitude is acquired using an accelerometer. Ambient temperature and humidity are acquired using an integrated temperature and humidity sensor. All sensor signals undergo analog-to-digital conversion and preliminary formatting via a field-installed data acquisition module, and are then transmitted to an edge computing gateway or directly uploaded to a cloud server via industrial Ethernet or wireless LoRa network. Finally, they are consumed by the data access service of the digital twin model, completing the synchronization.

[0044] In one possible implementation, such as Figure 3As shown, step S4 specifically includes: S401. A fault feature library is established in advance, which stores the mapping relationship between different fault types and the comprehensive operational deterioration degree of different numerical ranges. S402. Match the calculated comprehensive operational degradation degree with the fault feature database to determine the possible fault types of the equipment; S403. Based on the mechanism model, locate the component or subsystem that contributes the most to the overall operational degradation in the digital twin model and identify it as a suspected faulty component. S404. Based on a data-driven model, predict the time when it will reach a preset failure threshold based on current and historical comprehensive operational degradation data, and use this as the remaining service life.

[0045] The specific diagnostic process for step S4 is as follows.

[0046] The method for establishing the fault feature library in S401 is as follows: collect various fault maintenance records that have occurred in the history of this type of equipment or similar equipment, extract the comprehensive operational degradation value sequence within a certain period of time before each fault occurs, analyze its mean, variance, trend and other characteristics, and form the feature vector and threshold range of each fault type; or actively inject fault parameters such as component performance degradation, blockage, and leakage into the mechanism model of the digital twin model, simulate the operation and record the generated degradation curve, thereby constructing a fault mode library.

[0047] The matching process for S402 is as follows: calculate the moving average and recent gradient of the current overall operational degradation, perform similarity calculation with multiple records in the fault feature library, and use methods such as Euclidean distance or cosine similarity to select the top few fault types with the highest similarity as candidate outputs, and attach confidence probabilities.

[0048] The process of locating S403 is as follows: In the mechanism model of the digital twin model, the current measured boundary conditions and control system are used as inputs for simulation to obtain the theoretical state values ​​of each component or subsystem; by comparing the difference between the theoretical state values ​​and the actual synchronous data, or by calculating the partial derivative of each component parameter with respect to the comprehensive operational degradation calculation formula, its contribution is quantified; the unit with a significantly higher contribution than other components is marked as a suspected faulty component.

[0049] The prediction process for S404 is as follows: the overall operational degradation is taken as a time series and input into a pre-trained long short-term memory network or Prophet prediction model; the model predicts the trend of degradation over a future period based on historical patterns; the prediction curve is compared with a preset failure threshold line, and the time point at which the two curves intersect is calculated. The difference between this time point and the current time is the estimated remaining service life.

[0050] In one possible implementation, in step S3, the overall operational degradation index (DOI) is calculated using the following formula: Wherein, DOI represents the overall operational degradation, which is a dimensionless indicator; N represents the total number of key operational parameters monitored. This represents the actual measured value of the i-th key operating parameter at time t; This represents the standard value of the i-th key operating parameter under rated operating conditions; The sensitivity coefficient representing the deviation of the i-th parameter is determined by the degree of influence of this parameter on the overall performance of the equipment. This represents the time-cumulative effect coefficient of abnormal parameter fluctuations; Indicates the preset review time window length; Indicates the i-th parameter at time The absolute value of the abnormal rate of change is used. When the rate of change exceeds its normal fluctuation threshold, the actual value is taken; otherwise, 0 is taken.

[0051] In step S3, the calculation of the overall operational degradation is performed by a dedicated computing service module deployed in the digital twin platform. This module first reads N key parameters from the real-time database. Current value and corresponding standard value For each parameter, its normalized instantaneous deviation squared term is calculated. Simultaneously, the module retrieves the parameter's past performance from a time-series database. The historical data sequence within the time window is processed by numerical differentiation to calculate the rate of change at each time point. This rate is then compared with the normal fluctuation threshold of the parameter pre-stored in the knowledge base. Points with abnormal rate of change exceeding the threshold are selected and subjected to absolute value integration. Subsequently, the sensitivity coefficients of each parameter predefined in the knowledge base are invoked. and a unified cumulative effect coefficient The weighted summation calculation is then completed. Finally, the intermediate results of all parameters are arithmetically averaged and the square root is taken. The final result is written into the real-time database as the comprehensive operational degradation value at the current moment, triggering the subsequent diagnostic process.

[0052] In one possible implementation, in step S5, the dynamic energy efficiency optimization margin $DEOM$ is calculated using the following formula: Wherein, DEOM represents the dynamic energy efficiency optimization margin, which is a dimensionless indicator; This represents the energy efficiency ratio of the equipment system obtained by simulating operation using control strategy S in the digital twin model; This represents the actual energy efficiency ratio of a physical device at the current moment; The DOI represents the constraint factor on the realization of energy efficiency potential based on the equipment's health status; the DOI represents the overall operational degradation. This is the default maximum allowed value for DOI; This represents the actual load on the device during simulated operation under control strategy S; This indicates the rated load of the equipment.

[0053] In step S5, when calculating the dynamic energy efficiency margin, the specific calculation process is completed online within the strategy simulation loop of the digital twin model. For each simulated control strategy S, the digital twin model first runs the simulation under the set environmental parameters and load requirements, records the total energy consumption and effective output during the simulation process, and calculates the simulation energy efficiency ratio. Simultaneously, the actual energy efficiency ratio of the current physical devices is read from the real-time database. Subsequently, the relative energy efficiency improvement rate is calculated and logarithmically transformed to obtain the benefit term. Simultaneously, the latest DOI value of the overall operational degradation and its preset maximum value are retrieved from the database. and the average load during simulation operation and equipment rated load The penalty term is calculated according to the formula, where the constraint factor is... It is a constant preset based on the equipment type and operation and maintenance strategy. Finally, the benefit term is subtracted from the penalty term to obtain the dynamic energy efficiency margin (DEOM) value under strategy S, which is stored together with the unique identifier of this strategy for subsequent strategy selection.

[0054] In one possible implementation, such as Figure 4 As shown, step S6 specifically includes: S601. In the digital twin model, set the safety boundary constraints and process requirement constraints for equipment operation; S602. Among the set of control strategies that satisfy all constraints, select the control strategy that maximizes the dynamic energy efficiency optimization margin as the optimal control strategy. S603. If multiple strategies exist that allow the dynamic energy efficiency optimization margin to reach the same maximum value, then the strategy that minimizes the load on the key components of the equipment shall be selected as the optimal control strategy.

[0055] The specific constraints and decision-making logic for selecting the optimal control strategy in step S6 are as follows.

[0056] In S601, safety boundary constraints directly obtain specific numerical limits from equipment safety operating procedures or manufacturer safety data sheets, and embed them into the screening algorithm in the form of inequalities; process requirement constraints receive current requirement settings from the superior building energy management system or building automation system.

[0057] In S602, the screening algorithm first constructs a list containing all simulation strategies, then iterates through the list to check whether all state variables of the simulation output of each strategy have not touched the safety boundary and whether the process output fully meets the requirements; the strategies that pass the check are put into the feasible strategy set, and finally the record with the largest dynamic energy efficiency optimization margin is found in the set.

[0058] In S603, when multiple strategies have the same margin value and are all at their maximum value, the system will query the simulation results of these strategies for detailed data on the load rate, working stress or temperature of pre-specified key components (such as compressors and main motors), compare these data, and select the strategy with the lowest load rate or the lowest working stress as the final optimal solution.

[0059] In one possible implementation, step S7, converting the optimal control strategy into specific control instructions, includes converting the set parameters in the strategy into signals recognizable by the device controller. These signals include analog signals, digital signals, or data packets conforming to a specific communication protocol.

[0060] In step S7, the conversion and distribution of control commands are completed through a control command distribution service deployed at the edge or in the cloud. This service receives the optimal strategy description from the digital twin decision engine, which typically includes target setpoints (such as frequency setpoints or opening percentages) or mode commands (such as start / stop commands). The command distribution service has pre-built communication protocol driver libraries that match various brands and models of underlying controllers. Based on the identifier of the target device, the service selects the corresponding protocol driver and converts the general strategy description into command frames required by the specific controller, conforming to its register address specifications and data format requirements. For example, for controllers supporting Modbus TCP, it converts them into TCP packets containing function codes, register addresses, and setpoint data; for drivers requiring analog input, it outputs the corresponding 4-20mA analog signal through the connected digital output module. The converted commands are then distributed in real time to the control unit of the target device via the industrial network.

[0061] In one possible implementation, such as Figure 5 As shown, step S8, which involves adaptively correcting the digital twin model based on the comparison results, specifically includes: S801. Calculate the residual between the actual operational feedback data and the model's expected data; S802. When the residual exceeds the preset threshold, determine whether the deviation mainly comes from the inaccuracy of the mechanism model or the inaccuracy of the data-driven model. S803. If the mechanism model is inaccurate, adjust the specific physical parameters or relationships in the mechanism model. S804. If the data-driven model is inaccurate, add new running data to the training set to incrementally learn and update the data-driven model.

[0062] The adaptive correction process of the model in step S8 runs in an automated closed-loop manner.

[0063] In S801, the system continuously monitors new feedback data packets from physical devices and prediction data packets made by the digital twin model at the corresponding time points. After aligning them by timestamp, the system calculates the differences for core comparable variables such as total system power and key point temperature, forming a residual sequence.

[0064] In S802, the system monitors the moving average and standard deviation of the residual sequence. When the average continuously exceeds threshold A or the standard deviation continuously exceeds threshold B, a deviation source diagnosis is triggered. The diagnostic methods include: performing correlation analysis between the residuals and operating conditions (load rate, ambient temperature). If a strong correlation is observed, the system tends to be inaccurate in the mechanistic model; if the residuals exhibit randomness and are unrelated to operating conditions, or are highly correlated with the activation state of certain neurons within the data-driven model, the system tends to be inaccurate in the data-driven model.

[0065] In S803, if the mechanism model is determined to be inaccurate, the system calls the parameter identification algorithm. Using the historical actual data of the most recent period as the observation value and the mechanism model as the adjustable model, the system uses optimization algorithms such as least squares method and genetic algorithm to find the optimal model parameter correction value and automatically update the corresponding parameters in the mechanism model file.

[0066] In S804, if the data-driven model is determined to be inaccurate, the system adds newly collected, labeled data blocks to the model's circular buffer training set, triggering the online fine-tuning training process. This process uses a small learning rate to prevent catastrophic forgetting, incrementally updates the model in the background, and seamlessly switches to the new model version after the update is complete.

[0067] In one possible implementation, the data collected in steps S2 and S8 must undergo a preprocessing process including data cleaning, outlier removal, time series alignment, and normalization before synchronization or comparison.

[0068] The data preprocessing in steps S2 and S8 is handled by a separate data preprocessing microservice. This service receives the raw data stream from the sensor network. The data cleaning step filters out invalid data packets according to predefined rules (such as numerical range reasonableness and non-empty checks). Outlier removal uses statistical methods, such as calculating the short-term moving average and standard deviation for each parameter sequence, and considering data points exceeding three times the standard deviation as outliers and replacing them with linear interpolation. The time series alignment step uses a high-precision time clock to timestamp each data packet and resamples and interpolates each parameter sequence at fixed time intervals (e.g., 1 second) to ensure all data points are strictly aligned on the time axis. The normalization step uses the maximum and minimum values ​​of each parameter's historical data, or a pre-defined engineering range, to linearly scale the data to the [0, 1] interval. The processed and normalized data is published to the message bus for consumption by the digital twin model or for comparative analysis.

[0069] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: (1) In this invention, a deeply integrated digital twin that combines geometry, mechanism, and data-driven models is constructed, and the characteristic parameter of comprehensive operational degradation is innovatively proposed and calculated, thereby achieving a quantitative, comprehensive, and sensitive assessment of the health status of building equipment. This method not only focuses on the instantaneous deviation of multiple key operating parameters, but also captures the abnormal fluctuation trend of parameters through integral calculation, thus enabling it to keenly perceive early, slowly changing performance degradation signs that are ignored by traditional threshold alarms. Based on this comprehensive index, fault diagnosis, combined with the contribution analysis of a pre-established fault feature library and mechanism model, can accurately identify fault modes from massive amounts of data, precisely locate potential faulty components, and predict remaining service life. This transforms the operation and maintenance mode from post-maintenance and periodic maintenance to condition-based predictive maintenance, significantly improving the initiative and accuracy of operation and maintenance, and effectively avoiding unplanned downtime.

[0070] (2) In this invention, the real-time health status of the equipment, i.e., the overall operational degradation, is creatively introduced as a key input into the energy efficiency optimization decision-making process, and a dynamic energy efficiency optimization margin is defined as an evaluation index. This index, while measuring the energy efficiency improvement potential of different control strategies, simultaneously deducts the potential risk costs caused by the decline in equipment health and simulated load rate. This means that the numerous "hypothetical" simulations and strategy screenings conducted in the digital twin are no longer seeking the theoretically optimal energy efficiency point, but rather finding the safest and most sustainable energy efficiency optimal solution under the current actual health condition of the equipment. This resolves the contradiction between traditional energy efficiency optimization strategies and the actual capacity of the equipment, effectively ensuring the operational safety and service life of the equipment while deeply exploring energy-saving potential, thus achieving synergistic optimization between energy efficiency improvement and equipment longevity.

[0071] (3) In this invention, a complete closed loop from virtual decision-making to physical execution and feedback correction is established. After the optimal control strategy is executed by the physical entity, the system continuously compares the actual operating data with the expected data of the digital twin model. Once a significant deviation is detected, the model self-correction process is automatically initiated. This process can intelligently determine whether the deviation stems from inaccurate mechanistic model or data-driven model degradation, and perform targeted calibration using parameter identification or incremental learning respectively. This mechanism makes the digital twin model no longer a static, unchanging "digital portrait," but a "living entity" that can continuously evolve with the performance degradation of the physical entity, environmental changes, and even updates to the control strategy. It ensures that the digital twin model maintains high fidelity throughout its entire life cycle, providing a continuous and reliable foundation for the state assessment, fault diagnosis, and energy efficiency optimization it supports, and truly realizing a closed loop of intelligent operation and maintenance based on digital twin and sustainable trust.

[0072] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0073] The following points need to be explained: (1) The accompanying drawings of the embodiments of the present invention only involve the structures involved in the embodiments of the present invention. Other structures can refer to the general design.

[0074] (2) For clarity, the thickness of layers or regions is enlarged or reduced in the drawings used to describe embodiments of the invention, i.e., these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.

[0075] (3) Where there is no conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other to obtain new embodiments.

[0076] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins, characterized in that, include: S1. Construct a digital twin model of the target building equipment; The digital twin model includes a geometric model consistent with the geometric structure of the physical entity, a mechanistic model describing the physical characteristics and operating rules of the equipment, and a data-driven model trained based on historical operating data. S2. Real-time data collection of building equipment operation status and environmental parameters is achieved through a sensor network deployed on the physical entity, and the data is synchronized to the digital twin model. S3. In the digital twin model, based on real-time synchronized data and the mechanism model and / or data-driven model, real-time simulation and deduction of the equipment operating status are performed, and characteristic parameters reflecting the current health status of the equipment are calculated. The characteristic parameters include at least one comprehensive operating deterioration degree. S4. Based on the overall operational degradation, diagnose the potential fault types of the equipment, locate the faulty components, and assess the remaining service life. S5. In the digital twin model, the current environmental parameters, equipment load requirements and the comprehensive operational degradation are combined to simulate the equipment operation process under different control strategies, and the energy efficiency evaluation parameters under each strategy are calculated. The energy efficiency evaluation parameters include at least one dynamic energy efficiency optimization margin. S6. With the goal of maximizing the dynamic energy efficiency optimization margin, select the optimal control strategy from the simulated control strategies. S7. Convert the optimal control strategy into specific control commands and send them to the control system of the physical building equipment to optimize its operating efficiency; S8. After the physical entity executes the control command, continuously collect its actual operation feedback data and compare it with the expected operation data of the digital twin model. Based on the comparison results, adaptively correct the digital twin model.

2. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, In step S1, constructing a digital twin model of the target building equipment specifically includes: S101. Based on the device's CAD drawings or 3D scanned point cloud data, construct its high-fidelity geometric model; S102. Based on the equipment's design parameters, physical laws, and control logic, construct a mechanism model describing its energy conversion, mass transfer, and dynamic characteristics; S103. Collect historical operating data of the equipment under normal operation, multiple fault modes and different loads, and after preprocessing the historical operating data, train a data-driven model for state recognition and trend prediction. S104. The geometric model, mechanism model and data-driven model are fused and integrated under a unified spatiotemporal reference to form an interactive, computable and updatable digital twin model.

3. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, In step S2, the operating status data includes at least voltage, current, power, speed, temperature, pressure, flow rate, and vibration amplitude; the environmental parameter data includes at least ambient temperature and humidity.

4. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, Step S4 specifically includes: S401. A fault feature library is established in advance, which stores the mapping relationship between different fault types and the comprehensive operational deterioration degree of different numerical ranges. S402. Match the calculated comprehensive operational degradation degree with the fault feature library to determine the possible fault types of the equipment; S403. Based on the aforementioned mechanism model, locate the component or subsystem that contributes the most to the overall operational degradation in the digital twin model, and identify it as a suspected faulty component. S404. Based on the data-driven model, predict the time when it will reach the preset failure threshold according to the current and historical comprehensive operational degradation data, and use this as the remaining service life.

5. A method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1 or 3, characterized in that, In step S3, the overall operational degradation index (DOI) is calculated using the following formula: Wherein, DOI represents the overall operational degradation, which is a dimensionless indicator; N represents the total number of key operational parameters monitored. This represents the actual measured value of the i-th key operating parameter at time t; This represents the standard value of the i-th key operating parameter under rated operating conditions; The sensitivity coefficient representing the deviation of the i-th parameter is determined by the degree of influence of this parameter on the overall performance of the equipment. This represents the time-cumulative effect coefficient of abnormal parameter fluctuations; Indicates the preset review time window length; Indicates the i-th parameter at time The absolute value of the abnormal rate of change is used. When the rate of change exceeds its normal fluctuation threshold, the actual value is taken; otherwise, 0 is taken.

6. A method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1 or 5, characterized in that, include: In step S5, the dynamic energy efficiency optimization margin $DEOM$ is calculated using the following formula: Wherein, DEOM represents the dynamic energy efficiency optimization margin, which is a dimensionless indicator; This represents the energy efficiency ratio of the equipment system obtained by simulating operation using control strategy S in the digital twin model; This represents the actual energy efficiency ratio of a physical device at the current moment; The DOI represents the constraint factor on the realization of energy efficiency potential based on the equipment's health status; the DOI represents the overall operational degradation. This is the default maximum allowed value for DOI; This represents the actual load on the device during simulated operation under control strategy S; This indicates the rated load of the equipment.

7. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, Step S6 specifically includes: S601. In the digital twin model, set the safety boundary constraints and process requirement constraints for equipment operation; S602. Among the set of control strategies that satisfy all constraints, select the control strategy that maximizes the dynamic energy efficiency optimization margin as the optimal control strategy. S603. If multiple strategies exist that allow the dynamic energy efficiency optimization margin to reach the same maximum value, then the strategy that minimizes the load on the key components of the equipment shall be selected as the optimal control strategy.

8. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, In step S7, converting the optimal control strategy into specific control instructions includes: converting the set parameters in the strategy into signals that the device controller can recognize, including analog signals, digital signals, or data packets conforming to a specific communication protocol.

9. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, In step S8, the adaptive correction of the digital twin model based on the comparison results specifically includes: S801. Calculate the residual between the actual operational feedback data and the model's expected data; S802. When the residual exceeds the preset threshold, determine whether the deviation mainly comes from the inaccuracy of the mechanism model or the inaccuracy of the data-driven model. S803. If the mechanism model is inaccurate, adjust the specific physical parameters or relationships in the mechanism model. S804. If the data-driven model is inaccurate, add new running data to the training set to incrementally learn and update the data-driven model.

10. The method for intelligent diagnosis and energy efficiency optimization of building equipment based on digital twins according to claim 1, characterized in that, Before synchronization or comparison, the data collected in steps S2 and S8 must undergo preprocessing processes including data cleaning, outlier removal, time series alignment, and normalization.

Images