Self-adaptive energy-saving regulation and control method based on equipment topology and digital twinning

Abstract

The invention relates to the field of information technology and energy-saving simulation, in particular to a self-adaptive building energy-saving system based on equipment topology and digital twinning, which comprises the following steps: if specific equipment is added in the system, an edge side can automatically detect the increase of the equipment, and communicates with the equipment through an internet of things protocol to obtain equipment information; the corresponding platform acquires information on the edge side and then mounts the information into the corresponding asset group, a topological structure is established, and a control model updating process is triggered, specifically, the platform acquires equipment parameters, data validity is verified, the equipment parameters are classified and added into a parameter pool, and sensitivity analysis is performed on all the parameters; and performing model optimization based on a reinforcement learning algorithm according to the sensitivity weight to complete model updating. The system can meet the energy-saving requirement in a complex scene, has the self-adaptive capacity in coping with equipment adjustment and scene change, and can reduce the energy consumption and greenhouse gas emission of the system.

Description

Technical Field

[0001] The present invention relates to the fields of information technology and energy-saving simulation, and in particular to an adaptive energy-saving control method based on device topology and digital twins. Background Art

[0002] With the rapid development of information technology and the widespread use of intelligent devices, energy consumption issues in areas such as data centers, industrial systems, and IoT devices are becoming increasingly prominent. According to statistics, the proportion of energy consumption in global data centers to total electricity consumption has increased year by year, and this, coupled with an increase in greenhouse gas emissions, has put significant pressure on the environment. To meet this challenge, the research and application of energy-saving technologies have become an important direction in the current field of information technology. Traditional energy-saving methods are mostly based on static models or preset rules, which make it difficult to adapt to the complex needs brought about by dynamic adjustments to equipment or changes in scenarios. For example, when new equipment is added to the system or the operating environment changes, traditional models often require manual intervention or reconfiguration, and cannot achieve real-time adaptive optimization, resulting in reduced energy efficiency or waste of resources.

[0003] In recent years, digital twin technology and reinforcement learning algorithms have shown great potential in system modeling and optimization. Digital twins can reflect device status in real time and support dynamic simulation through virtual mapping of physical devices; reinforcement learning continuously optimizes strategies through environmental feedback and is suitable for decision-making problems in complex and changing scenarios. However, in existing technologies, digital twins are mostly used for device monitoring and prediction, and rarely combine topology structures and edge computing to achieve energy-saving optimization. At the same time, the parameter selection and model update process of reinforcement learning still face challenges in data verification and computing efficiency. Especially in scenarios with high availability requirements, how to quickly respond to device changes and build adaptive energy-saving models has become a technical problem that needs to be solved urgently. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that traditional energy-saving methods are difficult to adapt to the complex requirements brought about by dynamic adjustment of equipment or scene changes.

[0005] To this end, this paper provides an adaptive energy-saving control method based on device topology and digital twins. This method aims to achieve efficient energy conservation in dynamic environments through automatic edge detection, IoT communication, and reinforcement learning optimization. This method not only reduces energy consumption and greenhouse gas emissions but also improves the system's adaptability to device adjustments and scenario changes, providing an innovative solution for the fields of information technology and energy-saving simulation.

[0006] The technical solution adopted by the present invention to solve its technical problem is:

[0007] An adaptive energy-saving control method based on device topology and digital twins includes the following steps:

[0008] S1: The edge side automatically detects the addition or removal of underlying devices and communicates with the devices through IoT protocols to obtain real-time operating data and physical characteristics of the devices.

[0009] S2: After the corresponding platform obtains device information on the edge side, it mounts the device to the corresponding asset group, builds the device topology structure, and triggers the model update process when the topology changes;

[0010] S3: Perform data processing and feature extraction on device parameters, filter influencing parameters, and build a dynamic parameter pool;

[0011] S4: Based on the parameter pool data, perform sensitivity analysis on the model and make optimization adjustments based on the sensitivity weights;

[0012] S5: Adopt reinforcement learning algorithms to adaptively optimize the model and improve the accuracy and adaptability of energy-saving strategies;

[0013] S6: Replace the original model with the optimized model and perform intelligent energy-saving control based on the model to dynamically adjust the equipment operating status and reduce system energy consumption and greenhouse gas emissions.

[0014] Furthermore, in step S1, a connection is established with the underlying device through the Internet of Things protocol, and the device status change event is continuously monitored. When the device status change is triggered, the edge side actively sends an information collection request to the device, and obtains the physical properties, energy consumption data, operating status and related parameters of the device and performs a preliminary screening of the parameters. The device set is defined as E = {e1, e2, ..., e n}, when device e is detected n+1 Join or device m When removed: E′=E∪{e n+1} or E′=E\{e m}, the obtained device parameter set Θ={θ1,θ2,...,θ n}, the device parameter set of the i-th device is θ i ={S i ,P i ,A i ,O i}, where S i is the operating status of the i-th device, P i is the energy consumption data of the i-th device, A i is the physical property of the i-th device, O i are other parameters of the i-th device.

[0015] Furthermore, step S2 specifically includes:

[0016] S2-1 Establishing the device topology: Finding the asset information V in the device set = {v1, v2, ..., v n}, based on the physical connection relationship, communication protocol, device function and location relationship of the device E={(v i ,v j )|exists a physical or logical connection}, determining the asset group to which the device belongs, and updating the device logical topology information V;

[0017] S2-2 Update virtual device model: When the topology changes, the virtual device model is updated and the device changes are mapped to the virtual device layer V′=V∪{v n+1} or V′=V\{v m The virtual device layer regenerates the digital twin model of the device on the platform based on the topology information.

[0018] Furthermore, in step S3, the collected device data is standardized, including normalization, denoising and outlier detection, and an adaptive denoising filter is introduced: Where: P(ti) represents the original sampling data of history, the value at the ti-th moment; w i (α) represents the adaptive weight function associated with data position i; α is a parameter that controls the weight distribution.

[0019] Furthermore, the adaptive weight function satisfies W t The floor value of the window function, that is, the window size, W t Adjust parameters for adaptive window, Where: W base is the minimum window size, k1 is used to control the window increment, and λ controls the sensitivity of the window to the rate of change. That is, the rate of change of the signal.

[0020] Furthermore, in step S3, key parameter features are extracted based on the time series analysis method; according to the equipment operation rules and historical data trends, Lasso regression is used to perform feature selection to screen out key influencing factors, and the parameter pool is dynamically updated.

[0021] Furthermore, in step S4, the Pearson correlation coefficient correlation analysis method is used to evaluate the degree of influence of each parameter on the energy-saving model; the Sobol index is used to assign sensitivity weights to parameters with higher influence, the model input variables are adjusted based on the sensitivity weights, and the parameter combination is adjusted using Bayesian optimization to improve the adaptability and prediction accuracy of the model under different load conditions.

[0022] Furthermore, in step S5, a state-action-reward mechanism is defined to build a reinforcement learning environment. The Deep Q Network (DQN) deep reinforcement learning algorithm is used to intelligently optimize the energy-saving strategy and learn the optimal energy-saving strategy under different device configurations: Where γ is the discount factor and α is the learning rate. Online learning is performed under different working conditions to adjust the decision boundary of the reinforcement learning model to enhance the dynamic adaptability of the model. The Q value is updated using the following formula:

[0023] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0024] A computer device comprising:

[0025] a memory having a computer program stored thereon;

[0026] A processor is used to execute the computer program in the memory to implement the steps of the above method.

[0027] The beneficial effect of the present invention is that, by combining the device topology structure with digital twin technology, the present invention realizes the function of automatically detecting device changes on the edge side and dynamically updating the model. Compared with traditional static energy-saving models or solutions requiring manual intervention, it has higher adaptability and real-time performance. At the same time, based on the sensitivity analysis and parameter optimization of the reinforcement learning algorithm, it can accurately reduce energy consumption and greenhouse gas emissions for complex scenarios, overcoming the limitations of existing technologies in insufficient energy efficiency and low computing efficiency in dynamic environments, thereby significantly improving the energy-saving effect and high availability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below with reference to the accompanying drawings and examples.

[0029] Figure 1 This is a flow chart of the adaptive energy-saving control method based on device topology and digital twins in the present invention;

[0030] Figure 2 This is a flowchart of monitoring device status in step S1 of the present invention;

[0031] Figure 3 This is a flowchart of the process of triggering the virtual device model update in step S2 of the present invention;

[0032] Figure 4 This is a flow chart of constructing a dynamic parameter pool in step S3 of the present invention;

[0033] Figure 5 This is a flowchart of the sensitivity analysis of the virtual device model in step S4 of the present invention;

[0034] Figure 6 This is a flowchart of the adaptive optimization of the virtual device model in step S5 of the present invention;

[0035] Figure 7 This is a flowchart of the virtual device model executing the intelligent energy-saving control device in step S6 of the present invention. DETAILED DESCRIPTION

[0036] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0037] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, features defined as "first" or "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "multiple" means two or more.

[0038] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0039] Example 1

[0040] An adaptive energy-saving control method based on device topology and digital twins includes the following steps:

[0041] S1, the edge side automatically detects the increase or decrease of underlying devices and communicates with the devices through the IoT protocol to obtain the real-time operation data and physical characteristics information of the devices.

[0042] S1-1: Establish a connection with the underlying device through an IoT protocol (such as MQTT, CoAP, HTTP, or other applicable protocols), continuously monitor device status change events, and introduce adaptive thresholds during the monitoring process:

[0043]

[0044] Among them, P i For device e i Power, S t is the real-time load of the building, S max is the building full power load, T prev is the threshold at the previous moment, and α, β, γ are weight coefficients.

[0045] When the building load is high ( When the building load is low ( Small), lower the threshold, improve the detection sensitivity, ensure that changes in small devices (such as air conditioning and lighting at night) are captured, and further increase energy saving space.

[0046] S1-2: When the adaptive threshold is reached, the device status change is triggered. The edge side actively sends an information collection request to the device and obtains the device's physical properties, energy consumption data, operating status, and related parameters.

[0047] Define the device set as E = {e1, e2, ..., e n}, when device e is detected n+1 When added: E′=E∪{e n+1}, when device e is detected m When moving out: E′=E\{e m The acquired device parameter set Θ={θ1,θ2,...,θ n}, the device parameter set of the i-th device is θ i ={S i ,P i ,A i ,O i}, where S i is the operating status of the i-th device, P i is the energy consumption data of the i-th device, A i is the physical property of the i-th device, O i are other parameters of the i-th device.

[0048] S1-3: The acquired device data is preliminarily screened using boundary values, such as negative numbers, zero values, etc. Invalid, redundant, or non-compliant data is removed, and all parameters are encapsulated into a device set before uploading the data to the platform.

[0049] S2, after the corresponding platform obtains the device information on the edge side, it mounts the device to the corresponding asset group, builds the device topology structure, and triggers the model update process.

[0050] S2-1: Find the asset information in the device set V = {v1, v2, ..., v n}, based on the physical connection relationship, communication protocol, device function and location relationship of the device E={(v i ,v j )|exists a physical or logical connection}, determining the asset group to which the device belongs, and updating the device logical topology information V;

[0051] S2-2: When the topology changes, the virtual device model is updated and the device changes are mapped to the virtual device layer V′=V∪{v n+1} or V′=V\{v m The virtual device layer regenerates the device's digital twin model on the platform based on the topology information. This simultaneously triggers the control model update process, reports the update information to the computing server, and records device change events to ensure the real-time and integrity of the topology information.

[0052] S3, performs data processing and feature extraction on device parameters, filters influencing parameters, and builds a dynamic parameter pool.

[0053] S3-1: Standardize the collected device data, including normalization, denoising, and outlier detection, to ensure data quality;

[0054] Normalization uses the Min-Max Scaling method to convert the data to the range of [0,1] or [-1,1]

[0055] Improve the Moving Average Filter (MAF) and introduce the Adaptive Noise Filtering (ANF):

[0056]

[0057] Where: P(ti) represents the original sampling data of history, the value at the ti-th moment, w i (α) represents the adaptive weight function associated with data position i, To control the parameters of weight distribution, reflecting the noise intensity or environmental volatility, Represents σ tThe floor value of the window function is used to determine the number of data to be included, that is, the window size. The outlier detection uses the adaptive Z-score method to dynamically adjust the threshold:

[0058] The filter introduces adaptive window adjustment parameters based on the original W base is the minimum window size, k1 is used to control the window increment, and λ controls the sensitivity of the window to the rate of change. That is, the signal change rate. By increasing the adaptive parameter W t In scenarios where signals change quickly (such as equipment start and stop), R t Increase, W t Decrease, the window automatically decreases, in the moving average filter w i (α) The number of sums is reduced to quickly respond to changes and avoid signal loss; in scenarios where the signal is stable (such as stable operation), R t Decrease, W t As the window increases, it automatically increases. In the moving average filter, w i (α) The number of sums calculated increases, more data points are calculated, and the data is smoother, which enhances the filtering effect.

[0059] For example, when the signal changes very quickly, R t Approaches positive infinity, then W t ≈W base , indicating the maximum window size, is used to enhance the smoothing effect. When the signal changes very slowly, R t Approaching 0, W t ≈W base +k1 indicates the smallest window, which is used to respond quickly to changes.

[0060] Outlier detection uses an adaptive Z-score method to dynamically adjust the threshold:

[0061]

[0062] μ, σ are the mean and standard deviation of the data segment respectively, It is the instantaneous rate of change of the signal, and is used to enhance the monitoring of abnormal data when the signal changes suddenly.

[0063] S3-2: Extract key parameter features based on time series analysis methods (such as Fourier transform), including but not limited to load characteristics, energy consumption fluctuations, and environmental factors;

[0064] The Fourier transform formula is as follows:

[0065] F(ω)=∫x(t)e -jωtdt

[0066] Among them, F(ω) is the frequency domain characteristic signal, which represents the intensity of each frequency component in the signal, and x(t) represents the original signal of a physical quantity (such as power, temperature, current) of the equipment operation over time.

[0067] S3-3: Based on the equipment operation rules and historical data trends, Lasso regression is used for feature selection to screen out key influencing factors, and the parameter pool is dynamically updated to provide high-value input for subsequent model optimization.

[0068] The feature selection formula for Lasso regression is as follows:

[0069] min(∑(y i -X i β) 2 +λ∑|β j |)

[0070] Among them, y i represents the target output of the i-th sample (such as unit energy consumption); X i represents the feature vector of the i-th sample; β is the regression coefficient of the model, which is used to represent the influence weight of each feature; λ is the L1 regularization factor, which is used to constrain the complexity of the model.

[0071] S4, based on the parameter pool data, conduct sensitivity analysis on the model and optimize and adjust it according to the sensitivity weight.

[0072] S4-1: Use the Pearson correlation coefficient correlation analysis method to evaluate the impact of each parameter on the energy-saving model;

[0073] The formula for calculating the Pearson correlation coefficient is as follows:

[0074]

[0075] Among them, X i is the value of the i-th characteristic variable, Y i is the i-th response value, and r is used to measure the linear correlation between the feature and the response.

[0076] S4-2: Use the Sobol index to assign sensitivity weights to the parameters with higher influence, ensuring that the optimization process focuses on the core variables;

[0077] The Sobol index is calculated as follows:

[0078]

[0079] Among them, V i Represents variable X i The variance contribution of V totalrepresents the total variance of the variable, S i Used to represent the variable X i Contribution to the results.

[0080] S4-3: Adjust model input variables based on sensitivity weights and use Bayesian optimization to adjust parameter combinations to improve the model's adaptability and prediction accuracy under different load conditions.

[0081] Bayesian optimization objective function, where θ is the parameter combination and f(θ) is the objective function, such as unit energy consumption:

[0082]

[0083] S5 uses reinforcement learning algorithms to adaptively optimize the model to improve the accuracy and adaptability of energy-saving strategies.

[0084] S5-1: Define the State-Action-Reward (SAR) mechanism and build a reinforcement learning environment;

[0085] S5-2: Use the Deep Q Network (DQN) deep reinforcement learning algorithm to intelligently optimize energy-saving strategies and learn the optimal energy-saving strategies under different device configurations;

[0086]

[0087] Among them, γ is the discount factor, α is the learning rate, and s′ is the new state.

[0088] S5-3: Perform online learning under different working conditions and adjust the decision boundary of the reinforcement learning model to enhance the dynamic adaptability of the model. The Q value is updated using the following formula:

[0089]

[0090] S6, replaces the original model with the optimized model, and performs intelligent energy-saving control based on the model to dynamically adjust the equipment operation status and reduce system energy consumption and greenhouse gas emissions.

[0091] S6-1: The cloud computing server regularly updates the control model. After the update is completed, the new model is pushed to the platform to replace the original model.

[0092] S6-2: The platform calculates the optimal operating parameters P of the equipment in real time based on the control model opt =argmin∑(P i +λ ij P j ), dynamically optimize energy consumption distribution, and issue control instructions C={c1,c2,...,c n}To edge devices such as gateways.

[0093] S6-3: After receiving the adjustment signal, the edge device sends it to the corresponding downstream device to adjust the device working state S′=f(C,S,T response ).

[0094] S6-4: The edge device collects device-side operating parameter information.

[0095] S6-5: The platform collects parameter information reported by the edge device, processes the data, and reports it to the cloud computing server.

[0096] S6-6: The cloud computing server obtains the data reported by the platform side and updates the control model based on the new data.

[0097] This patented system leverages edge-side automatic detection and IoT communication technologies, combined with platform-side model optimization and control strategies, to achieve dynamic monitoring and intelligent adjustment throughout the data collection and processing process of an intelligent energy-saving control system. By automatically detecting underlying device changes, it acquires real-time device operating parameters and physical characteristics, clarifies device data collection requirements and processing methods, constructs a device asset topology, and automatically triggers model updates based on these changes. The system standardizes the collected data, extracts features, and filters parameters, dynamically establishes a parameter pool, and continuously optimizes model inputs through sensitivity analysis and weight assignment. Subsequently, a reinforcement learning algorithm is used to adaptively optimize and iteratively update the model, improving its accuracy and responsiveness under varying loads and operating conditions. Ultimately, the optimized model is used for intelligent energy-saving control, dynamically adjusting device operating status and reducing overall system energy consumption and carbon emissions. Simultaneously, a closed-loop update mechanism is formed between the cloud and platform to continuously optimize control strategies based on the latest collected data. This ensures continuous optimization and dynamic adaptation of device operating status and energy consumption control, enhancing overall energy efficiency and the level of intelligent system management.

[0098] Example 2

[0099] An embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement an adaptive energy-saving control method based on device topology and digital twins as provided in the above method embodiment.

[0100] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be integrated into any of the other components of a computer device (or mobile device) in whole or in part. As described in the embodiments of the present application, the data processing circuitry functions as a processor control (e.g., selection of a variable resistor terminal path connected to an interface). The memory may be used to store software programs and modules of application software, such as the program instruction / data storage device corresponding to a rolling bearing fault intelligent diagnosis method in the embodiments of the present application. The processor executes the software programs and modules stored in the memory to perform various functional applications and data processing, thereby implementing one of the aforementioned methods. The memory may include high-speed random access memory (RAM) and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, which may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wireless network provided by a communication provider of the computer device. In one example, the transmission device includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices via a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly. The display can be, for example, a touch screen liquid crystal display (LCD), which can enable a user to interact with the user interface of the computer device (or mobile device).

[0101] Example 3

[0102] The embodiment of the present application also provides a computer-readable storage medium, which can be set in a server to store at least one instruction or at least one program related to an adaptive energy-saving control method based on device topology and digital twins for implementing a method embodiment, and the at least one instruction or the at least one program is loaded and executed by the processor to implement an adaptive energy-saving control method based on device topology and digital twins provided in the above method embodiment. Optionally, in this embodiment, the above-mentioned storage medium may be located in at least one network server among multiple network servers in a computer network. Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0103] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical spirit of this invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. An adaptive energy-saving control method based on device topology and digital twin, characterized in that: The following steps are included: S1: The edge side automatically detects the addition or removal of underlying devices and communicates with the devices through IoT protocols to obtain real-time operating data and physical characteristics of the devices. S2: After the corresponding platform obtains device information on the edge side, it mounts the device to the corresponding asset group, builds the device topology structure, and triggers the model update process when the topology changes; S3: Perform data processing and feature extraction on device parameters, filter influencing parameters, and build a dynamic parameter pool; S4: Based on the parameter pool data, perform sensitivity analysis on the model and make optimization adjustments based on the sensitivity weights; S5: Adopt reinforcement learning algorithms to adaptively optimize the model and improve the accuracy and adaptability of energy-saving strategies; S6: Replace the original model with the optimized model and perform intelligent energy-saving control based on the model to dynamically adjust the equipment operating status and reduce system energy consumption and greenhouse gas emissions.

2. The adaptive energy-saving control method based on device topology and digital twin according to claim 1 is characterized in that: In step S1, a connection is established with the underlying device through the Internet of Things protocol, and the device status change event is continuously monitored. When the device status change is triggered, the edge side actively sends an information collection request to the device, obtains the device's physical properties, energy consumption data, operating status and related parameters, and performs a preliminary screening of the parameters. The device set is defined as E = {e1, e2, ..., e n }, when device e is detected n+1 Join or device m When removed: E ′ =E∪{e n+1 } or E ′ =E\{e m }, the obtained device parameter set Θ={θ1,θ2,...,θ n }, the device parameter set of the i-th device is θ i ={S i ,P i ,A i ,O i }, where S i is the operating status of the i-th device, P i is the energy consumption data of the i-th device, A i is the physical property of the i-th device, O i are other parameters of the i-th device.

3. The adaptive energy-saving control method based on device topology and digital twin according to claim 1 is characterized in that: Step S2 specifically includes: S2-1 Establishing the device topology: Finding the asset information V in the device set = {v1, v2, ..., v n }, based on the physical connection relationship, communication protocol, device function and location relationship of the device E={(v i ,v j )|exists a physical or logical connection}, determining the asset group to which the device belongs, and updating the device logical topology information V; S2-2 Update virtual device model: When the topology changes, trigger the update of the virtual device model and map the device changes to the virtual device layer V of the digital twin platform. ′ =V∪{v n+1 } or V ′ =V\{v m The virtual device layer regenerates the digital twin model of the device on the platform based on the topology information.

4. The adaptive energy-saving control method based on device topology and digital twin according to claim 1 is characterized in that: In step S3, the collected device data is standardized, including normalization, denoising and outlier detection, and an adaptive denoising filter is introduced: Where: P(ti) represents the original sampling data of history, the value at the ti-th moment; w i (α) represents the adaptive weight function associated with data position i; α is a parameter that controls the weight distribution.

5. The adaptive energy-saving control method based on device topology and digital twin according to claim 4 is characterized in that: The adaptive weight function satisfies W t The floor value of the window function, that is, the window size, W t Adjust parameters for adaptive window, Where: W base is the minimum window size, k1 is used to control the window increment, and λ controls the sensitivity of the window to the rate of change. That is, the rate of change of the signal.

6. The adaptive energy-saving control method based on device topology and digital twin according to claim 4 is characterized in that: In step S3, key parameter features are extracted based on the time series analysis method; Based on the equipment operation rules and historical data trends, Lasso regression is used for feature selection to screen out key influencing factors and dynamically update the parameter pool.

7. The adaptive energy-saving control method based on device topology and digital twin according to claim 4 is characterized in that: In step S4, the Pearson correlation coefficient correlation analysis method is used to evaluate the degree of influence of each parameter on the energy-saving model; the Sobol index is used to assign sensitivity weights to parameters with higher influence, the model input variables are adjusted based on the sensitivity weights, and the parameter combination is adjusted using Bayesian optimization to improve the adaptability and prediction accuracy of the model under different load conditions.

8. The adaptive energy-saving control method based on device topology and digital twin according to claim 1 is characterized in that: In step S5, a state-action-reward mechanism is defined to build a reinforcement learning environment. The Deep Q Network (DQN) deep reinforcement learning algorithm is used to intelligently optimize the energy-saving strategy and learn the optimal energy-saving strategy under different device configurations: Where γ is the discount factor and α is the learning rate. Online learning is performed under different working conditions to adjust the decision boundary of the reinforcement learning model to enhance the dynamic adaptability of the model. The Q value is updated using the following formula:

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 8.

Images