Method and device for predictive dynamic scheduling of power load of old building

Abstract

The invention provides an old building power load predictive dynamic scheduling method and equipment, and the method achieves the closed-loop feedback of a scheduling decision and the self-optimization of a model through the automatic obtaining of BIM equipment parameters, and the construction of a precise prediction model for key equipment in a stack room. The problems that the old book library power load prediction precision is insufficient, the scheduling decision is extensive, and a continuous optimization mechanism is lacked are effectively solved, the power supply safety and stability are remarkably improved, and internal precious articles are protected; bIM is used as a core data source for the first time, the key equipment attributes and topology of the stack library are automatically obtained, and a foundation is laid for fine management; the deep learning model is specially used for optimizing the characteristics of key equipment in a stack room, especially temperature and humidity sensitive equipment, and the prediction precision is obviously higher than that of a traditional method; the scheduling result and the actual load data are written back to the BIM in real time and are used for model retraining, so that the system is continuously self-promoted, and the method is particularly suitable for personalized requirements of old book libraries.

Description

Technical Field

[0001] This invention relates to a predictive dynamic scheduling method and equipment for power load in old buildings. Background Technology

[0002] The operation and maintenance management of old buildings, especially the power supply guarantee for important functional spaces (such as archives and library stacks), faces severe challenges. The power distribution facilities in these buildings are aging and have limited load-bearing capacity, while internal equipment (such as constant temperature and humidity air conditioners, precision dehumidifiers, and special lighting) has extremely high requirements for power stability. Instantaneous overload can easily lead to equipment shutdown or even damage to valuable items (such as ancient books and documents). Traditional power load forecasting methods mainly rely on historical meter data and simple environmental parameters (such as temperature), which have the following key shortcomings:

[0003] 1. Relying on manual meter reading or ordinary sensors makes it impossible to obtain detailed operating parameters of key equipment (such as specific equipment power, start / stop status, and set temperature and humidity thresholds) in real time and accurately.

[0004] 2. General prediction models are difficult to adapt to the complex operating logic of special equipment in old book warehouses (such as the nonlinear and intermittent high power consumption characteristics of high-precision temperature and humidity control equipment) and its extreme sensitivity to the environment.

[0005] 3. After scheduling decisions (such as shutting down equipment), there is a lack of a mechanism to feed back real-time load changes and their impacts to the building information model, making it impossible to form closed-loop optimization and knowledge accumulation.

[0006] 4. Failure to fully utilize the equipment's own attribute information (such as service area area, weight of its impact on the comfort of stored items, and topological location in the power distribution network) for refined scheduling decisions. Summary of the Invention

[0007] The purpose of this invention is to provide a predictive dynamic scheduling method and equipment for power load in old buildings.

[0008] To address the above problems, this invention provides a predictive dynamic scheduling method for power load in old buildings, comprising:

[0009] Build a detailed operation dataset at the equipment level for old buildings and store the detailed operation dataset in the BIM database;

[0010] Based on the refined operational dataset in the BIM database, long-term and short-term prediction models are trained.

[0011] Based on the short-term prediction model, the optimal or suboptimal shutdown plan is obtained, i.e., the set of equipment C that needs to be shut down.

[0012] Before the time point when the load is predicted to exceed the safe capacity of the power distribution system of the old building, execute the optimal or suboptimal shutdown scheme C obtained from the solution;

[0013] The actual load data collected in real time when executing the optimal or suboptimal shutdown plan C and the actual effect of executing the optimal or suboptimal shutdown plan C are used as new actual operating data and automatically written back to the BIM database of the corresponding key equipment in the BIM model; the short-term prediction model and the long-term prediction model are retrained using the new actual operating data.

[0014] Furthermore, in the above method, a refined operational dataset at the equipment level for old buildings is constructed, and this refined operational dataset is stored in a BIM database, including:

[0015] The basic attribute information of key equipment is automatically extracted from the building information model of old buildings, and the actual operation data and environmental parameters of the key equipment are monitored and collected in real time. The basic attribute information, real-time operation data and environmental parameters of the key equipment are fused and preprocessed to construct a refined operation dataset at the equipment level, and the refined operation dataset is stored in the BIM database.

[0016] Furthermore, in the above method, the key equipment includes: a constant temperature and humidity air conditioning unit, a precision dehumidifier, a special lighting circuit, and a constant humidity device for the ancient book preservation cabinet;

[0017] The basic attribute information includes: equipment identifier, rated power, service area, equipment type (whether it is a temperature and humidity sensitive device) and its position in the power distribution network topology diagram;

[0018] The actual operating data includes: power consumption, operating status, set values, and actual temperature and humidity output;

[0019] The environmental parameters include the temperature and humidity inside and outside the old building.

[0020] Furthermore, in the above method, a long-term prediction model is trained based on a refined operational dataset from the BIM database, including:

[0021] Using a long short-term memory network or Transformer architecture and a refined dataset from several years of history, a long-term prediction model is trained to predict the overall and key equipment load trends in old buildings over the next few days to weeks. The long-term prediction model is used to capture the seasonal and periodic operating patterns of temperature and humidity sensitive equipment in old buildings.

[0022] Furthermore, in the above method, a short-term prediction model is trained based on a refined operational dataset from the BIM database, including:

[0023] A model combining an attention mechanism and a one-dimensional convolutional neural network is constructed as a short-term prediction model, and the short-term prediction model is trained using a refined running dataset from the last few hours. The attention mechanism enables the short-term prediction model to focus on key equipment and environmental factors in old buildings that have the greatest impact on load fluctuations. The one-dimensional convolutional neural network is responsible for extracting local fluctuation patterns in the load time series. The short-term prediction model is used to predict the precise load for the next few minutes to tens of minutes.

[0024] Furthermore, in the above method, based on the short-term prediction model, the optimal or suboptimal shutdown plan is obtained, including:

[0025] A trained short-term prediction model is used to predict the real-time load in old buildings. When the predicted real-time load exceeds the safe capacity of the building's power distribution system, an instantaneous feature vector X is calculated for each dispatchable critical device. i ;

[0026] Based on instantaneous feature vector X i This yields the optimal or suboptimal shutdown plan, i.e., the set of devices C that need to be shut down.

[0027] Furthermore, in the above method, the instantaneous feature vector X i for:

[0028] X i = <p i ,u i ,c i A i ,e i ,d i Among them,

[0029] p i : Indicates the rated power P of device i i The ratio to the required load reduction ΔP;

[0030] u i : Indicates the percentile of the historical usage duration of device i;

[0031] c i : Represents the comfort level of device i and the sensitivity coefficient for the preservation of collections;

[0032] A i : Represents the service area of ​​device i;

[0033] e i : Indicates the environmental dependency of device i;

[0034] d i : Indicates the electrical distance from device i to the power distribution core node.

[0035] Furthermore, in the above method, based on the instantaneous feature vector X i This yields the optimal or suboptimal shutdown plan, i.e., the set of devices C that need to be shut down, including:

[0036] Let C be the set of devices that need to be shut down;

[0037] Set constraints: ∑ i∈C P i ≥ΔP, meaning the total shutdown power must meet the requirements;

[0038] Where, ∑ i∈C P i : Represents the rated power P for each device i in set C. i The summation is performed, which is the sum of the rated power of all the shut-down equipment.

[0039] ΔP: Indicates the amount of load that the power distribution system of an old building needs to reduce;

[0040] Set optimization goals:

[0041] This means minimizing the impact of closure on the environmental comfort of old buildings and the preservation of collections; among which,

[0042] min: indicates finding the minimum value;

[0043] ∑ i∈C : indicates that the minimum value is summed over each device i in set C;

[0044] It represents the instantaneous rate of change of outdoor temperature, reflecting the intensity of external environmental disturbances;

[0045] min[∑ i∈C d i This means minimizing the power distribution network losses caused by shutdown operations.

[0046] Based on the constraints and optimization objectives, the optimal or suboptimal shutdown scheme, i.e. the set of devices C that need to be shut down, is solved using a multi-objective evolutionary algorithm or a weighted optimization algorithm.

[0047] According to another aspect of the invention, a computer-readable storage medium is provided that stores computer-executable instructions thereon, wherein when executed by a processor, the computer-executable instructions cause the processor to perform the method as described in any of the preceding claims.

[0048] According to another aspect of the present invention, a calculator device is provided, comprising:

[0049] Processor; and

[0050] A memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the method described in any of the preceding descriptions.

[0051] Compared with existing technologies, this invention aims to provide a predictive dynamic scheduling method for power load in old buildings (especially focusing on library scenarios) based on BIM and artificial intelligence. By automatically acquiring BIM equipment parameters, constructing accurate prediction models for key library equipment, realizing closed-loop feedback of scheduling decisions and model self-optimization, it effectively solves the problems of insufficient power load prediction accuracy, coarse scheduling decisions, and lack of continuous optimization mechanisms in old libraries, significantly improving power supply safety and stability, and protecting valuable items inside.

[0052] This invention is the first to use BIM as the core data source, automatically acquiring the attributes and topology of key equipment in the library, laying the foundation for refined management. It enables refined equipment-level prediction: the deep learning model is specifically optimized for the characteristics of key library equipment (especially temperature and humidity sensitive equipment), achieving significantly higher prediction accuracy than traditional methods.

[0053] This invention writes the scheduling results and actual load data back to the BIM in real time and uses it for model retraining, forming a closed loop of "prediction-decision-execution-feedback-optimization", enabling the system to continuously improve itself, and is especially suitable for the personalized needs of old book warehouses.

[0054] This invention defines and quantifies the unique "comfort and collection preservation sensitivity coefficient" (c) of the library. i ")" and "environmental dependence (e)" i This will make scheduling decisions more scientific and protect precious collections to the greatest extent possible.

[0055] This invention significantly reduces the risk of equipment failure and temperature and humidity out-of-control conditions caused by power overload in old book warehouses through accurate prediction and intelligent optimization scheduling, ensuring the safety of collections and the normal operation of functional spaces. Attached Figure Description

[0056] Figure 1 This is a flowchart of a predictive dynamic scheduling method for power load in old buildings according to an embodiment of the present invention. Detailed Implementation

[0057] The present invention will now be described in further detail with reference to the accompanying drawings.

[0058] In a typical configuration of this application, the terminal, the device of the service network, and the trusted party all include one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0059] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0060] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0061] like Figure 1 As shown, this invention provides a predictive dynamic scheduling method for power load in old buildings, the method comprising:

[0062] Step S1, BIM-based equipment-level power load data integration and processing:

[0063] The basic attribute information of key equipment is automatically extracted from the building information model (BIM) of old buildings such as libraries. The actual operating data and environmental parameters of the key equipment are monitored and collected in real time. The basic attribute information, real-time operating data and environmental parameters of the key equipment are fused and preprocessed, including data alignment, missing value imputation, outlier detection and correction, in order to construct a refined operating dataset at the equipment level and store the refined operating dataset in the BIM database.

[0064] Preferably, the key equipment includes: a constant temperature and humidity air conditioning unit, a precision dehumidifier, a special lighting circuit, and a constant humidity device for the ancient book preservation cabinet.

[0065] The basic attribute information includes: device identifier, rated power P i Service area A i The type of equipment, whether it is a temperature and humidity sensitive device, and its location in the power distribution network topology diagram, etc.

[0066] The actual operating data includes: power consumption, operating status, set values, and actual temperature and humidity output;

[0067] The environmental parameters include the temperature and humidity inside and outside the library.

[0068] Step S2: Construct and train deep learning prediction models for key equipment in old buildings such as libraries.

[0069] Model input enhancement: The refined operational datasets from several years of historical data obtained in step 1 from the BIM database are used as model input features to train long-term and short-term prediction models; wherein, the long-term and short-term prediction models pay particular attention to the correlation between the operating modes of temperature and humidity sensitive equipment in the library and the environment.

[0070] Preferably, the long-term prediction model adopts a long short-term memory network (LSTM) or Transformer architecture and uses a refined dataset from several years of history to train a long-term prediction model to predict the overall load trend of the library and key equipment in the next few days to several weeks. The long-term prediction model is used to capture the seasonal and periodic operating patterns of equipment in old buildings that are sensitive to temperature and humidity.

[0071] Here, the long-term prediction model can capture the seasonal and periodic patterns of the periodic operation of special equipment in old buildings, such as temperature and humidity sensitive equipment in libraries, like the humidity control device for ancient book preservation cabinets.

[0072] Preferably, the short-term prediction model is constructed by combining an attention mechanism and a one-dimensional convolutional neural network (1D-CNN) as the short-term prediction model, and trained using a refined running dataset (second-level / minute-level) from the most recent few hours. The attention mechanism enables the short-term prediction model to focus on key equipment (such as the response of air conditioners to sudden changes in temperature and humidity) and environmental factors (such as sudden changes in outdoor temperature) in older buildings that have the greatest impact on load fluctuations. The CNN is responsible for extracting local fluctuation patterns in the load time series. The short-term prediction model is used to predict the accurate load for the next few minutes to tens of minutes.

[0073] Step S3, Intelligent shutdown decision based on multi-objective optimization:

[0074] Using the short-term prediction model trained in step S2, real-time load prediction is performed in the old building. When the real-time predicted load exceeds the safe capacity of the library's power distribution system:

[0075] Assess the impact of equipment shutdown: Calculate the instantaneous feature vector X for each schedulable critical device. i = <p i ,ui ,c i A i ,e i ,d i >:

[0076] p i Rated power P of device i i (From BIM) The ratio of the load reduction ΔP to the required load reduction (Pi / ΔP);

[0077] u i : Percentile of historical usage time for device i (calculated by sorting);

[0078] c i : The comfort level and preservation sensitivity of equipment i (defined for older buildings such as book depositories). For example, the comfort level and preservation sensitivity of air conditioning, which directly affects the temperature and humidity of the core ancient book storage area, are much higher than the comfort level and preservation sensitivity of lighting in ordinary reading areas. ci is quantified through measurement or preset rules;

[0079] A i : Service area of ​​device i (directly derived from BIM attributes);

[0080] e i : Environmental dependence of device i. The ei value is 0 for non-temperature and humidity sensitive devices (such as general lighting); the ei value of temperature and humidity sensitive devices (air conditioners, dehumidifiers) is the absolute value of the Pearson correlation coefficient between air conditioner power and outdoor temperature (reflecting the degree to which they are affected by the external environment);

[0081] d i Electrical distance from device i to the core power distribution node (sum of shortest path impedances, calculated using topology information extracted from BIM);

[0082] Multi-objective optimization decision: Let C be the set of devices that need to be shut down;

[0083] Set constraints: ∑ i∈C P i ≥ΔP, meaning the total shutdown power must meet the requirements;

[0084] Where, ∑ i∈C P i : Represents the rated power P for each device i in set C. i The summation is performed, which is the sum of the rated power of all the shut-down equipment.

[0085] ΔP: Indicates the amount of load that needs to be reduced in the library's power distribution system.

[0086] Optimization goal:

[0087] This means minimizing the impact of closure on the comfort of the stacks environment and the preservation of the collections; among which,

[0088] min: indicates finding the minimum value;

[0089] ∑ i∈C : indicates that the minimum value is summed over each device i in set C;

[0090] It represents the instantaneous rate of change of outdoor temperature, reflecting the intensity of external environmental disturbances;

[0091] min[∑ i∈C d i This refers to minimizing distribution network losses caused by shutdown operations (line losses due to the location of the shut-down equipment). Here, when equipment needs to be shut down to reduce power load, minimizing the total electrical distance from the shut-down equipment to the core distribution node helps reduce distribution network losses.

[0092] Based on the constraints and optimization objectives, the optimal or suboptimal shutdown scheme, i.e. the set of devices C that need to be shut down, is solved using a multi-objective evolutionary algorithm (such as NSGA-II) or a weighted optimization algorithm.

[0093] Here, there are two optimization objectives: first, to minimize the impact of shutdown on the comfort of the library environment and the preservation of the collection; and second, to minimize the power distribution network losses caused by the shutdown operation. Finally, a multi-objective evolutionary algorithm (such as NSGA-II) or a weighted optimization algorithm is used to solve for the optimal or suboptimal shutdown scheme, i.e., the set of devices that need to be shut down is C.

[0094] Step S4, edge execution and BIM data backhaul closed-loop optimization:

[0095] Prediction and decision-making modules can be deployed on edge computing nodes in old buildings such as library sites to optimize edge execution and BIM data backhaul in a closed loop.

[0096] Anti-delay compensation: Before the predicted time point when the load will exceed the safe capacity of the power distribution system of an old building such as a library, execute the solved optimal or suboptimal shutdown scheme C; among which, a dynamic overload early warning buffer threshold can be set; and a rolling optimization strategy (such as every 5-10 seconds) is used to update the predicted time point when the load will exceed the safe capacity of the power distribution system of the library and the optimal or suboptimal shutdown scheme C.

[0097] BIM Data Update and Model Self-Learning: The actual load data collected in real time when executing the optimal or suboptimal shutdown plan C, and the actual effects of executing the optimal or suboptimal shutdown plan C (such as which equipment was shut down, the actual load reduction, and the temperature and humidity changes in key areas of the library after shutdown) are used as new actual operating data and automatically written back to the BIM database of the corresponding key equipment in the BIM model; using the new actual operating data, the short-term prediction model and long-term prediction model (especially the short-term prediction model) of step S2 are retrained periodically or incrementally online to improve their prediction accuracy and adaptability for the specific library environment.

[0098] Example 1: Automatic Acquisition of BIM Data

[0099] A BIM model of the library conforming to IFC standards was used. An interface program was developed to automatically parse the BIM file and extract the basic attribute information of key equipment (such as the variable frequency constant temperature and humidity air conditioning unit with ID AHU-101): Equipment Identifier Name = "Main Air Conditioner for the Ancient Books Library", Rated Power = 15kW, Service Area ServeSpace = "Rare Books Library (Area = 200m)". 2 Device Type = "HVAC", Location in the distribution network topology diagram: ElectricalPanel = "Panel-B-2". Simultaneously, obtain the total electrical path impedance (c) from the distribution system BIM sub-model to the main transformer. i ).

[0100] Example 2: Library Sensitivity Coefficient (c) i )set up

[0101] An assessment of a library's stacks was conducted, and equipment c was configured. i value:

[0102] Air conditioning units serving the "Special Collections of Ancient Books" (core area): c i =0.9;

[0103] Air conditioning units serving the "General Document Reading Area": ​​c i =0.4;

[0104] Humidity cabinet for ancient book storage: c i =0.85;

[0105] Main lighting in the reading area: c i =0.2;

[0106] Public corridor lighting: c i =0.1.

[0107] Example 3: Prediction Model Update and Results

[0108] Initial Deployment: This system was deployed in the stacks of a history library in a certain city. In the initial phase (within one week), the mean absolute percentage error (MAPE) of the short-term prediction model was 8.5%.

[0109] Closed-loop operation: The system runs continuously, updating the BIM database every 5 seconds with actual load data and scheduling results (such as the IDs of shut-down equipment, actual power changes, and temperature and humidity fluctuations). Incremental training of the short-term model is performed every Sunday morning using new data from the past week.

[0110] Performance Improvement: After three months of operation, the short-term prediction model's MAPE decreased to 5.2%. During a period of unusually high summer temperatures, the system predicted overload risk five minutes in advance (a warning the original method failed to provide), and intelligently shut down some public lighting and non-core reading area air conditioning (c i The temperature and humidity fluctuations in the core area were controlled within ±0.5℃ / ±2%RH (compliant with GB / T 30227-2013 "Basic Requirements for Ancient Book Repositories in Libraries"), and the impact on users in non-core areas was minimal.

[0111] Safety Enhancement: Within six months of system operation, the system successfully predicted and mitigated 12 overload risks. No equipment protection tripping or temperature and humidity exceeding the standard due to overload occurred in the library. The failure rate of risk response estimated by traditional manual monitoring methods during the same period was approximately 15-20%.

[0112] Example 4: Optimization Decision Calculation

[0113] At a certain moment, the predicted load exceeds the limit ΔP = 20kW. The system calculates the Xi vector of all schedulable devices (30 units in total). After solving the problem using the NSGA-II optimization algorithm, the optimal solution set is obtained. The administrator selects a balanced solution: shut down device C = {public corridor lighting group (5kW, c...}). i =0.1,d i Small), one air conditioner (10kW, c) in the general reading area i =0.4,d i (Chinese), spare dehumidifier for document restoration room (5kW, c) i =0.3, non-core area)}, total shutdown 20kW. This plan targets the core area of ​​ancient books (c i High-end equipment is unaffected, and network loss increases only slightly.

[0114] According to another aspect of the invention, a computer-readable storage medium is provided that stores computer-executable instructions thereon, wherein when executed by a processor, the computer-executable instructions cause the processor to perform the method as described in any of the preceding claims.

[0115] According to another aspect of the present invention, a calculator device is provided, comprising:

[0116] Processor; and

[0117] A memory configured to store computer-executable instructions, which, when executed, cause the processor to perform the method described in any of the preceding descriptions.

[0118] In summary, this invention aims to provide a predictive dynamic scheduling method for power load in old buildings (especially focusing on library scenarios) based on BIM and artificial intelligence. By automatically acquiring BIM equipment parameters, constructing accurate prediction models for key library equipment, and realizing closed-loop feedback of scheduling decisions and model self-optimization, it effectively solves the problems of insufficient power load prediction accuracy, coarse scheduling decisions, and lack of continuous optimization mechanisms in old libraries, significantly improving power supply safety and stability, and protecting valuable items inside.

[0119] This invention is the first to use BIM as the core data source, automatically acquiring the attributes and topology of key equipment in the library, laying the foundation for refined management. It enables refined equipment-level prediction: the deep learning model is specifically optimized for the characteristics of key library equipment (especially temperature and humidity sensitive equipment), achieving significantly higher prediction accuracy than traditional methods.

[0120] This invention writes the scheduling results and actual load data back to the BIM in real time and uses it for model retraining, forming a closed loop of "prediction-decision-execution-feedback-optimization", enabling the system to continuously improve itself, and is especially suitable for the personalized needs of old book warehouses.

[0121] This invention defines and quantifies the unique "comfort and collection preservation sensitivity coefficient" (c) of the library. i ")" and "environmental dependence (e)" i This will make scheduling decisions more scientific and protect precious collections to the greatest extent possible.

[0122] This invention significantly reduces the risk of equipment failure and temperature and humidity out-of-control conditions caused by power overload in old book warehouses through accurate prediction and intelligent optimization scheduling, ensuring the safety of collections and the normal operation of functional spaces.

[0123] This invention can reduce manual inspection and intervention, optimize equipment operation, and potentially reduce energy consumption and equipment wear and tear.

[0124] For detailed descriptions of the various device embodiments of the present invention, please refer to the corresponding sections of the various method embodiments; they will not be repeated here.

[0125] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

[0126] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present invention (including associated data structures) can be stored in a computer-readable recording medium, such as RAM memory, a magnetic or optical drive, a floppy disk, or similar devices. Furthermore, some steps or functions of the present invention can be implemented in hardware, for example, as circuitry that works with a processor to perform the various steps or functions.

[0127] Furthermore, a portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. The program instructions invoking the methods of the invention may be stored in a fixed or removable recording medium, and / or transmitted via a data stream in a broadcast or other signal-carrying medium, and / or stored in the working memory of a computer device operating according to the program instructions. Here, an embodiment of the invention includes an apparatus comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, the apparatus is triggered to operate the methods and / or technical solutions based on the foregoing embodiments of the invention.

[0128] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in the apparatus claims may also be implemented by a single unit or device in software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any particular order.

Claims

1. A method for predictive dynamic scheduling of electrical loads in an old building, characterized in that, include: Build a detailed operation dataset at the equipment level for old buildings and store the detailed operation dataset in the BIM database; Based on the refined operational dataset in the BIM database, long-term and short-term prediction models are trained. Based on the short-term prediction model, the optimal or suboptimal shutdown plan is obtained, i.e., the set of equipment C that needs to be shut down. Before the time point when the load is predicted to exceed the safe capacity of the power distribution system of the old building, execute the optimal or suboptimal shutdown scheme C obtained from the solution; The actual load data collected in real time when executing the optimal or suboptimal shutdown plan C and the actual effect of executing the optimal or suboptimal shutdown plan C are used as new actual operating data and automatically written back to the BIM database of the corresponding key equipment in the BIM model; the short-term prediction model and the long-term prediction model are retrained using the new actual operating data.

2. The method of claim 1, wherein the old building power load predictive dynamic scheduling method is characterized by, Construct a detailed operational dataset for equipment in old buildings, and store this dataset in a BIM database, including: The basic attribute information of key equipment is automatically extracted from the building information model of old buildings, and the actual operation data and environmental parameters of the key equipment are monitored and collected in real time. The basic attribute information, real-time operation data and environmental parameters of the key equipment are fused and preprocessed to construct a refined operation dataset at the equipment level, and the refined operation dataset is stored in the BIM database.

3. The method of claim 2, wherein the method further comprises: The key equipment includes: a constant temperature and humidity air conditioning unit, a precision dehumidifier, a special lighting circuit, and a constant humidity device for the ancient book preservation cabinet; The basic attribute information includes: equipment identifier, rated power, service area, equipment type (whether it is a temperature and humidity sensitive device) and its position in the power distribution network topology diagram; The actual operating data includes: power consumption, operating status, set values, and actual temperature and humidity output; The environmental parameters include the temperature and humidity inside and outside the old building.

4. The method of claim 1, wherein the old building power load predictive dynamic scheduling method is characterized by, A long-term prediction model was trained based on a refined operational dataset from the BIM database, including: Using a long short-term memory network or Transformer architecture and a refined dataset from several years of history, a long-term prediction model is trained to predict the overall and key equipment load trends in old buildings over the next few days to weeks. The long-term prediction model is used to capture the seasonal and periodic operating patterns of temperature and humidity sensitive equipment in old buildings.

5. The method of claim 1, wherein the old building power load predictive dynamic scheduling method is characterized by, A short-term prediction model was trained based on a refined operational dataset from the BIM database, including: A model combining an attention mechanism and a one-dimensional convolutional neural network is constructed as a short-term prediction model, and the short-term prediction model is trained using a refined running dataset from the last few hours. The attention mechanism enables the short-term prediction model to focus on key equipment and environmental factors in older buildings that have the greatest impact on load fluctuations. The one-dimensional convolutional neural network is responsible for extracting local fluctuation patterns in the load time series. The short-term prediction model is used to predict the precise load for the next few minutes to tens of minutes.

6. The method of claim 1, wherein the old building power load predictive dynamic scheduling method is characterized by, Based on the aforementioned short-term prediction model, the optimal or suboptimal shutdown schemes are obtained, including: Using the trained short-term prediction model for real-time load prediction within the old building, when the real-time predicted load will exceed the safe capacity of the old building's power distribution system, calculate the instantaneous feature vector X for each dispatchable critical device i ; Based on the instantaneous eigenvector X i The optimal or suboptimal shutdown scheme, i.e. the set of devices C that need to be closed, is obtained.

7. The method of claim 6, wherein the method further comprises: The instantaneous feature vector X i is: X i = <p i , u i , c i , A i , e i , d i >, wherein, p i : denotes the rated power P of the device i i : the ratio of the actual power P to the required reduced load amount ΔP; u i : denotes the historical usage duration percentile of device i; c i : represents the comfort and collection preservation sensitivity coefficient of the device i; A i : denotes the service area of device i; e i : represents the environment dependence degree of device i; d i : represents the electrical distance of device i to the power distribution core node.

8. The method of claim 7, wherein the old building power load predictive dynamic scheduling method is characterized by, Based on the instantaneous eigenvector X i The optimal or suboptimal shutdown scheme, i.e. the set of devices C that need to be closed, is obtained by Let C be the set of devices that need to be shut down; Set constraints:∑ i∈C P i ≥ΔP, i.e. total shutdown power needs to meet requirements; where∑ i∈C P i : denotes the summing of the rated power P i of each device i in the set C, that is, the sum of the rated powers of all devices that are shut down. ΔP: Indicates the amount of load that the power distribution system of an old building needs to reduce; Set optimization goals: i.e. minimizing the impact of the shutdown on the environmental comfort and preservation of the collections of the old building; wherein, min: means to find the minimum value; ∑ i∈C : denotes the sum over each device i in the set C for which the minimum is sought. represents the outdoor temperature instantaneous change rate, reflecting the external environment disturbance intensity; min[∑ i∈C d i ], i.e. minimizing the distribution network loss caused by the switching off operation; Based on the constraint conditions and optimization objectives, and using a multi-objective evolutionary algorithm or a weighted optimization algorithm to solve the optimal or suboptimal shutdown scheme, that is, the set of devices C that need to be closed.

9. A computer-readable storage medium having stored thereon computer- executable instructions, wherein, The computer executable instructions, when executed by the processor, cause the processor to perform the method of any one of claims 1 to 8.

10. A computing device, wherein, Comprising: a processor; and a memory arranged to store computer executable instructions which, when executed, cause the processor to perform the method of any one of claims 1 to 8.

Images