A method and system for building power demand management based on big data

Abstract

This invention relates to the field of smart grid technology, specifically disclosing a building power demand management method and system based on big data. The method includes the following steps: acquiring multi-source electricity consumption data generated by smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces; performing time alignment and field matching according to a unified sampling time window to generate a building electricity consumption status sequence; generating a load profile based on area identifiers, equipment categories, operating status, and historical electricity consumption records; classifying load attributes; generating demand forecast values ​​and adjustable load scheduling queues; outputting hierarchical control commands; and adjusting the scheduling weight for the next control cycle based on power changes, switch status, and non-execution reason identifiers. This invention, through multi-source data alignment, load profiling, scheduling queues, and execution feedback linkage, establishes a closed-loop correspondence between demand forecasting basis, control commands, and the actual operating status of the building.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a building power demand management method and system based on big data. Background Technology

[0002] The field of smart grid technology typically revolves around data collection, transmission, storage, calculation, command issuance, and execution control for electricity production, transmission, distribution, metering, consumption, and dispatch. This involves data interaction between smart meters, building distribution cabinets, energy terminals, energy management servers, communication gateways, and databases. Traditional building power demand management methods and systems refer to technologies that manage the electricity consumption records and operational status of lighting, air conditioning, elevators, charging equipment, and office loads within a building. This typically involves smart meters, sub-metering devices, environmental sensors, and building controllers collecting power, electricity consumption, time of day, temperature, and equipment on / off status. This data is transmitted to a server via wired or wireless networks. The server then performs statistical analysis, comparison, and sorting based on historical electricity consumption curves, time-of-use pricing, load categories, and preset thresholds, and issues start / stop, load limiting, or time-of-day adjustment commands to distribution switches, air conditioning controllers, or energy terminals.

[0003] Traditional building power demand management relies on data uploaded by smart meters, sub-metering devices, environmental sensors, and building controllers. The server performs statistical comparisons based on historical electricity consumption curves, time-of-use pricing, load categories, and preset thresholds. However, in actual operation, there are differences in the sampling time, field granularity, and equipment status records of data from various sources. The correlation between power changes after the execution of control commands and the reasons for non-execution in determining the next cycle is insufficient, resulting in a lag in demand forecasting and limited matching between load scheduling order and the real-time operating status of the building and user-side constraints. Summary of the Invention

[0004] The purpose of this invention is to provide a building power demand management method and system based on big data. This method is used to form a closed-loop processing mode that can be used for demand forecasting, load profiling, scheduling queue generation, and scheduling weight correction by focusing on the correlation between the sampling time, field granularity, equipment status records, and control execution feedback of multi-source power consumption data in the building. This enables the building power demand management process to make control decisions based on consistent data sequences, distinguishable load attributes, and execution feedback records.

[0005] To achieve the above objectives, this invention provides a building power demand management method based on big data, comprising the following steps: acquiring multi-source power consumption data generated by smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces; performing time alignment and field matching on the multi-source power consumption data according to a unified sampling time window to generate a building power consumption status sequence; generating a load profile based on area identifiers, equipment categories, operating status, and historical power consumption records, and classifying the load profile into rigid loads, delayable loads, reduceable loads, and transferable loads; generating a demand forecast value for the next control cycle based on the building power consumption status sequence, ambient temperature changes, and executed control command status; generating an adjustable load scheduling queue based on time-of-use pricing, load categories, area priorities, equipment operating status, continuous operating time, and user-side constraints, and outputting hierarchical control commands; receiving power changes, switch status, and non-execution reason identifiers after the execution of the hierarchical control commands, and adjusting the scheduling weight for the next control cycle when the deviation between the actual power consumption value and the demand forecast value meets the correction trigger condition.

[0006] Furthermore, the steps of time alignment and field matching of the multi-source electricity consumption data according to the unified sampling time window include: determining the unified sampling time window by jointly using the electricity price time period boundary and the control cycle of the building controller; writing the total power record collected by the smart meter, the sub-item electricity record collected by the sub-metering device, the temperature record collected by the environmental sensor, the switch status record collected by the equipment controller, and the time-of-use electricity price record provided by the electricity price data interface into the corresponding time window respectively; when there are multiple records of the same device or the same area within the same time window, the record whose collection timestamp is close to the end of the time window is used as the field matching object; when the field matching object lacks an area identifier or device category, the area identifier or device category corresponding to the device number in the historical electricity consumption record is called to supplement it and generate the building electricity consumption status sequence.

[0007] Furthermore, the building electricity consumption status sequence includes a total power field, a component electricity consumption field, an ambient temperature field, an equipment switch status field, a region identifier field, an equipment category field, and a time-of-use electricity price field, all arranged according to a unified sampling time window. When the direction of electricity consumption change corresponding to the total power field is consistent with the direction of electricity consumption change corresponding to the component electricity consumption field, the corresponding time window is marked as a time window that can participate in demand forecasting. When the direction of electricity consumption change corresponding to the total power field is inconsistent with the direction of electricity consumption change corresponding to the component electricity consumption field, the corresponding time window is marked as a time window to be verified, and the time window to be verified is excluded from the direct calculation input of the current demand forecast value.

[0008] Furthermore, the step of generating the load profile based on the area identifier, equipment category, operating status, and historical electricity consumption records includes: grouping the building electricity consumption status sequence according to the area identifier and equipment category to form an area equipment combination record; writing the continuous operating time, historical electricity consumption for the same period, number of switch status changes, and time-of-use pricing period percentage into the area equipment combination record; generating a load profile record based on the correspondence between the continuous operating time and the historical electricity consumption for the same period, and writing a load adjustable attribute field into the load profile record, the load adjustable attribute field being used to distinguish between rigid loads, delayable loads, reduceable loads, and transferable loads.

[0009] Furthermore, the process of writing the load adjustable attribute field includes: when the equipment category corresponding to the regional equipment combination record is a safe power supply equipment or a continuously operating equipment, the load adjustable attribute field is written as a rigid load; when the equipment category corresponding to the regional equipment combination record has a variable start time and the number of switch state changes meets the delay control condition, the load adjustable attribute field is written as a delayable load; when the continuous running time corresponding to the regional equipment combination record meets the reduction control condition and the electricity consumption of the same historical period meets the reduction judgment condition, the load adjustable attribute field is written as a reduceable load; when the time-of-use electricity price period ratio corresponding to the regional equipment combination record meets the peak-shifting judgment condition, the load adjustable attribute field is written as a transferable load.

[0010] Furthermore, the step of generating the demand forecast value for the next control cycle based on the building power consumption status sequence, ambient temperature changes, and executed control command status includes: extracting real-time power records and sub-item power consumption records adjacent to the next control cycle from the building power consumption status sequence; extracting historical power consumption records of the same period from historical power consumption records that belong to the same area, the same equipment category, and the same electricity price period as the next control cycle; correcting the historical power consumption records of the same period for air conditioning loads according to the temperature difference corresponding to the ambient temperature change; and generating the demand forecast value by deducting or delaying the status of the prediction input records for the corresponding equipment category according to the load limit, delayed start, or off-peak operation flags corresponding to the executed control command status.

[0011] Furthermore, after the next control cycle ends, the system receives the total power record collected by the smart meter and the sub-item power record collected by the sub-metering device, and converts the total power record and sub-item power record into actual power consumption. The system compares the deviation between the actual power consumption and the demand forecast. When the deviation meets the correction trigger condition, the system uses the area identifier, equipment category, and executed control command status corresponding to the deviation as the basis for a single update to adjust the scheduling weight of the next control cycle. When the deviation does not meet the correction trigger condition, the scheduling weight of the previous control cycle is used.

[0012] Further, the steps of generating the adjustable load scheduling queue based on time-of-use pricing, load category, regional priority, equipment operating status, continuous operating time, and user-side constraints, and outputting the hierarchical control instructions, include: establishing candidate scheduling records for delayable loads, loads that can be reduced, and loads that can be transferred, respectively; writing time-of-use pricing level, load category, regional priority, equipment operating status, continuous operating time, and user-side constraints into the candidate scheduling records; when a candidate scheduling record meets the user-side constraints and the equipment operating status allows adjustment, determining the queue position according to the regional priority and time-of-use pricing level; when a candidate scheduling record does not meet the user-side constraints or the equipment operating status does not allow adjustment, excluding the candidate scheduling record from the adjustable load scheduling queue for that period.

[0013] Furthermore, the process of outputting the hierarchical control instructions includes: sequentially reading candidate scheduling records according to their queue positions in the adjustable load scheduling queue; generating a load limiting instruction when a candidate scheduling record corresponds to a load that can be reduced; generating a delayed start instruction when a candidate scheduling record corresponds to a load that can be delayed; and generating a peak-shifting operation instruction when a candidate scheduling record corresponds to a load that can be shifted. The process also includes receiving the power changes, switch status, and non-execution reason identifiers after the execution of the load limiting instruction, delayed start instruction, or peak-shifting operation instruction, and writing the power changes, switch status, and non-execution reason identifiers into the execution feedback record corresponding to the candidate scheduling record.

[0014] This invention also provides a building power demand management system based on big data. The system supports the aforementioned big data-based building power demand management technology solution and includes a multi-source data access unit, a state sequence construction unit, a load profile generation unit, a demand forecasting unit, a scheduling queue generation unit, a control command generation unit, and an execution feedback acquisition unit. The multi-source data access unit communicates with smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces, and sends multi-source power consumption data to the state sequence construction unit. The state sequence construction unit performs time alignment and field matching on the multi-source power consumption data and outputs a building power consumption state sequence. The load profile generation unit generates a load profile based on the building power consumption state sequence. The demand forecasting unit generates demand forecast values, the scheduling queue generation unit generates an adjustable load scheduling queue, the control command generation unit outputs hierarchical control commands, and the execution feedback acquisition unit transmits execution feedback to the demand forecasting unit and the scheduling queue generation unit.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by aligning and matching the multi-source electricity consumption data generated by smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces according to a unified sampling time window, a consistent data foundation can be formed for the building's electricity consumption status sequence. Loads with different adjustable attributes are divided into load profiles, and an adjustable load scheduling queue is generated by combining time-of-use pricing, regional priority, equipment operating status, continuous operating time, and user-side constraints. Control commands correspond to load attributes. By receiving power changes, switch status, and non-execution reason identifiers, and adjusting the scheduling weight of the next control cycle when the deviation meets the correction trigger condition, subsequent demand forecasting and scheduling order follow the execution feedback changes. Attached Figure Description

[0016] 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.

[0017] Figure 1 This is the main flowchart for building power demand management in this invention; Figure 2 This is a schematic diagram illustrating the time window alignment of multi-source power consumption data in this invention; Figure 3 This is a schematic diagram illustrating the load profile and load attribute division of the present invention; Figure 4 This is a schematic diagram illustrating the demand forecasting and feedback correction status of the present invention; Figure 5 This is a diagram illustrating the collaborative relationships within the building power demand management system of this invention. Detailed Implementation

[0018] The technical solution of the present invention will be further described below with reference to embodiments. Subsequent embodiments are described within the boundaries of the claims, the invention content, and the disclosed technical content, without including undisclosed specific numerical values, parameter ranges, experimental data, formula derivations, or comparison results in the specification. In subsequent embodiments, smart meters, sub-metering devices, environmental sensors, equipment controllers, electricity price data interfaces, building controllers, distribution switches, air conditioning controllers, and energy-consuming terminals serve as data sources, control transmission objects, or execution objects in building power demand management, respectively; the server serves as the processing entity for receiving, processing, judging, outputting, and providing feedback corrections.

[0019] In subsequent embodiments, the multi-source electricity consumption data refers to the data set generated by smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces and entered into the server processing flow; the building electricity consumption status sequence refers to the serialized record formed after the data is aligned to time and matched to fields according to a unified sampling time window; the load profile refers to the load characteristic record formed based on area identification, equipment category, operating status, and historical electricity consumption records; the demand forecast value refers to the electricity demand result used for scheduling judgment in the next control cycle; the adjustable load scheduling queue refers to the sequence of control objects formed after sorting the adjustable loads; and the hierarchical control instructions refer to the load limiting, delayed start, peak-shifting operation, or maintenance operation control instructions output to the power distribution switch, air conditioning controller, or energy terminal.

[0020] The preset conditions, judgment conditions, correction trigger conditions, delay control conditions, reduction control conditions, reduction judgment conditions, and peak-shaving judgment conditions in the embodiments can be determined by building management strategies, equipment rated power, equipment operation protection requirements, user-side constraints, and historical operation data.

[0021] For ease of implementation, conditions can be represented by preset thresholds, preset ratios, preset durations, preset times, or combinations thereof. Different thresholds can be configured for different buildings or different equipment categories.

[0022] The thresholds, coefficients, and calculation methods given in the subsequent embodiments are used to illustrate possible implementations of the present invention and do not constitute the sole limitation on the scope of protection.

[0023] Please see Figures 1 to 4 This embodiment provides a building power demand management method based on big data. This method is applicable to the demand management process of power-consuming objects such as lighting, air conditioning, elevators, charging equipment, and office loads within a building. The server receives power consumption, environmental, status, and electricity price records generated during building operation through data connections with smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces. After forming a unified data sequence, it performs load profiling, demand forecasting, scheduling queue generation, instruction output, and feedback correction.

[0024] S1. Acquire multi-source electricity consumption data. The server obtains total power records from smart meters, individual electricity consumption records from sub-metering devices, temperature records from environmental sensors, switch status records from device controllers, and time-of-use electricity price records from the electricity price data interface. Total power records reflect the overall electricity consumption status of the building; individual electricity consumption records reflect the electricity consumption status of different circuits, areas, or equipment categories; temperature records are used to determine the demand of air conditioning loads; switch status records reflect the operating status of energy-consuming terminals; and time-of-use electricity price records are used for subsequent scheduling queue sorting. After these records enter the server, they are identified using the collection timestamp, device number, area identifier, device category, and source interface to form a data input set to be processed.

[0025] S2 performs unified sampling time window processing on multi-source electricity consumption data. The server determines the unified sampling time window based on the electricity price time period boundary and the control cycle of the building controller. The total power record collected by the smart meter, the individual electricity consumption record collected by the sub-metering device, the temperature record collected by the environmental sensor, the switch status record collected by the equipment controller, and the time-of-use electricity price record provided by the electricity price data interface are written into the corresponding time window respectively. The corresponding time window is used to place data from different sources, different sampling times, and different field granularities under the same processing benchmark, avoiding the direct mixing of data under different time calibers in subsequent prediction and scheduling. For example, if the building controller's control cycle is 15 minutes and the time-of-use pricing period boundary is 10:00, when a certain basic time window is from 9:50 to 10:05, this basic time window crosses the time-of-use pricing period boundary. The server then divides it into two unified sampling time windows: 9:50 to 10:00 and 10:00 to 10:05. When a certain basic time window does not cross the time-of-use pricing period boundary, the server directly uses that basic time window as the unified sampling time window. Through this processing, the total power record, individual power consumption record, temperature record, switch status record, and time-of-use pricing record within the same unified sampling time window all correspond to the same time-of-use pricing caliber and control cycle caliber.

[0026] S3 performs field matching on records written to the unified sampling time window. The server uses records with the same device number, area identifier, or device category within the same time window as candidate matching objects. When multiple records of the same device or area exist within the same time window, the server uses the record whose collection timestamp is closest to the end of the time window as the field matching object. This process allows the device status at the end of the control cycle to serve as the input status for subsequent demand prediction. If the field matching object lacks an area identifier or device category, the server uses the area identifier or device category corresponding to the device number in historical electricity consumption records to supplement it; if the supplementation basis cannot form a definite correspondence, the relevant record is kept in a pending verification state and is not used as direct input for the current demand prediction.

[0027] S4 generates a building electricity consumption status sequence. After time alignment and field matching, the server generates the following fields arranged according to a unified sampling time window: total power, individual electricity consumption, ambient temperature, equipment on / off status, area identifier, equipment category, and time-of-use electricity price. These fields constitute the recording units of building electricity consumption status within the same time window, and the recording units are arranged in chronological order to form the building electricity consumption status sequence. When forming the sequence, the server also performs a consistency check on the direction of electricity consumption change between the total power field and the individual electricity consumption field. When the direction of electricity consumption change corresponding to the total power field is consistent with the direction of electricity consumption change corresponding to the individual electricity consumption field, the corresponding time window is marked as a time window that can participate in demand forecasting; when the direction of electricity consumption change corresponding to the total power field is inconsistent with the direction of electricity consumption change corresponding to the individual electricity consumption field, the corresponding time window is marked as a time window to be verified, and the direct calculation input for the current demand forecast value is excluded. The time window to be verified is retained in the sequence for subsequent data source tracing and execution feedback, but it does not directly participate in the generation of the current forecast input.

[0028] S5, Generate Load Profile. The server groups the building's electricity consumption status sequence according to region identifier and device category, forming a region device combination record. The region device combination record contains continuous runtime, historical electricity consumption for similar time periods, number of switch status changes, and the percentage of time-of-use pricing periods. Continuous runtime is derived from the continuity of device switch status in adjacent time windows; historical electricity consumption for similar time periods is derived from records matching the region, device category, and electricity price period in historical electricity consumption records; the number of switch status changes is derived from status change records uploaded by the device controller; and the percentage of time-of-use pricing periods is derived from the distribution of the time-of-use pricing field in the corresponding time window. The server generates a load profile record based on the correspondence between continuous runtime and historical electricity consumption for similar time periods, and writes the load adjustability attribute field into the load profile record.

[0029] S6, Write the load adjustable attribute field. The server writes the load adjustable attribute field based on the equipment category, number of switch state changes, continuous running time, historical electricity consumption during similar periods, and the proportion of time-of-use pricing periods corresponding to the regional equipment combination record. If the equipment category corresponding to the regional equipment combination record is a safe power supply device or a continuously operating device, the server writes the load adjustable attribute field as a rigid load. If the equipment category corresponding to the regional equipment combination record has a variable start time and the number of switch state changes meets the delay control condition, the server writes the load adjustable attribute field as a delayable load. If the continuous running time corresponding to the regional equipment combination record meets the reduction control condition and the historical electricity consumption during similar periods meets the reduction judgment condition, the server writes the load adjustable attribute field as a reduceable load. If the proportion of time-of-use pricing periods corresponding to the regional equipment combination record meets the peak-shaving judgment condition, the server writes the load adjustable attribute field as a transferable load. The above writing process ensures that the load object forms a distinguishable and callable state record before scheduling, preventing load objects under different constraints from being directly mixed into the same control object set.

[0030] S7 generates the demand forecast for the next control cycle. The server extracts real-time power records and sub-item power records adjacent to the next control cycle from the building's power consumption status sequence, and extracts historical power consumption records for the same period in the same area, equipment category, and electricity price period as the next control cycle from historical power consumption records. For air conditioning loads, the server corrects the historical power consumption records for the same period based on the temperature difference corresponding to changes in ambient temperature. For loads involved in executed control commands, the server deducts or postpones the status of the corresponding equipment category's forecast input records based on the load limiting, delayed start, or off-peak operation flags corresponding to the executed control command status. Status deduction is used to ensure that the power consumption status of loads that have been limited reflects the power consumption status after the load limiting in the forecast input of the next control cycle, while status postponement is used to ensure that loads with delayed start or off-peak operation reflect the demand status after the change in the start-up period in the forecast input of the next control cycle. Specifically, the state reduction amount corresponding to the load limiting instruction is determined based on the target power reduction in the load limiting instruction, the average power of the corresponding equipment category within the most recent several unified sampling time windows, and the actual response power of historical similar control instructions; the state delay processing corresponding to the delayed start instruction includes moving the predicted input record of the corresponding equipment category from the original start control cycle to the target start control cycle; the state delay processing corresponding to the off-peak operation instruction includes moving the predicted input record of the corresponding equipment category from the control cycle corresponding to the high electricity price period to the control cycle corresponding to the flat or valley period; the server generates the demand prediction value based on the above input records and state processing results, and sends the demand prediction value to the subsequent scheduling queue generation process.

[0031] S8 generates an adjustable load scheduling queue. The server creates candidate scheduling records for deferred loads, slashable loads, and transferable loads, and writes time-of-use pricing level, load category, area priority, equipment operating status, continuous runtime, and user-side constraints into the candidate scheduling records. The time-of-use pricing level represents the price status of the corresponding time window, the load category comes from the load adjustable attribute field, the area priority comes from the area order in the building management strategy, the equipment operating status comes from the switch status records uploaded by the equipment controller, the continuous runtime comes from the state continuity relationship of adjacent time windows, and the user-side constraints come from the equipment operating constraints already set in the building power management. The server filters the candidate scheduling records. When a candidate scheduling record meets the user-side constraints and the equipment operating status allows adjustment, the server determines the queue position of the candidate scheduling record in the adjustable load scheduling queue according to the area priority and time-of-use pricing level. When a candidate scheduling record does not meet the user-side constraints or the equipment operating status does not allow adjustment, the server excludes the candidate scheduling record from the adjustable load scheduling queue for that cycle. The above filtering and sorting process establishes a correspondence between the scheduling object and the real-time operating status of the building, user-side constraints, and price status.

[0032] S9 outputs hierarchical control instructions. The server reads candidate scheduling records sequentially according to their queue positions in the adjustable load scheduling queue and generates corresponding control instructions based on the load adjustable attribute fields of the candidate scheduling records. When a candidate scheduling record corresponds to a load that can be reduced, the server generates a load limiting instruction; when a candidate scheduling record corresponds to a load that can be delayed, the server generates a delayed start instruction; when a candidate scheduling record corresponds to a load that can be transferred, the server generates a peak-shifting operation instruction; when a candidate scheduling record corresponds to a rigid load or does not meet the adjustment conditions, the candidate scheduling record is not used as the scheduling control object for this time. The server transmits the load limiting instruction, delayed start instruction, or peak-shifting operation instruction to the building controller, which then transmits the instructions to the power distribution switch, air conditioning controller, or energy consumption terminal. The power distribution switch, air conditioning controller, or energy consumption terminal changes the operating status of the corresponding equipment according to the received instructions, forming execution feedback that can be collected by the equipment controller or metering device.

[0033] S10: Receive execution feedback and correct scheduling weights. The server receives power changes, switch statuses, and non-execution reason identifiers after the execution of load limiting instructions, delayed start instructions, or peak-shifting operation instructions, and writes these information into the execution feedback record corresponding to the candidate scheduling record. After the next control cycle ends, the server receives the total power record collected by the smart meter and the sub-item power record collected by the sub-metering device, and converts the total power record and sub-item power record into actual electricity consumption. The server compares the deviation between the actual electricity consumption and the demand forecast. When the deviation meets the correction trigger condition, it adjusts the scheduling weights for the next control cycle based on the area identifier, equipment category, and executed control instruction status corresponding to the deviation. When the deviation does not meet the correction trigger condition, the scheduling weights of the previous control cycle are used. The scheduling weights are then incorporated into the subsequent adjustable load scheduling queue generation process, ensuring that the queue position for the next control cycle can accommodate the execution feedback status of the previous cycle.

[0034] S11, Feedback Loop and Record Recall. The server stores the building power consumption status sequence, load profile, demand forecast, adjustable load scheduling queue, hierarchical control instructions, and execution feedback records into corresponding storage areas. These records are recalled in subsequent control cycles: the building power consumption status sequence is used to form new real-time inputs, the load profile is used to maintain load categories and adjustable attributes, the demand forecast is used to determine deviations from actual power consumption, the adjustable load scheduling queue is used to determine the control sequence, and the hierarchical control instructions and execution feedback records are used to determine the executed status and reasons for non-execution. Through the above record recall relationship, the server forms a closed-loop processing process with data sequence, forecast, scheduling, execution, and feedback as the main lines after each control cycle.

[0035] In this embodiment, records that are validly included in prediction and scheduling should meet the following conditions: complete fields, corresponding time windows, clear source interfaces, and verifiable relationships between the total power field and the sub-item power fields. Records with missing fields but which can be definitively correlated with historical power consumption records through device numbers can be used for area identification or device category supplementation; records that cannot be definitively correlated enter a pending verification state. The pending verification state is not equivalent to deleting the record, but rather prevents the record from being directly input into the current demand prediction value and retains it for subsequent data traceability. The non-execution reason identifier in the execution feedback is used to indicate the non-execution of instructions due to device offline, user-side constraints, operation protection, or communication status; the non-execution reason identifier enters the scheduling weight correction process of the next control cycle as part of the execution feedback record, but does not replace the device operating status field itself.

[0036] In this embodiment, the load limiting command, delayed start command, and off-peak operation command are all specific output states of hierarchical control commands. The load limiting command corresponds to loads that can be reduced, and its execution target can be a power distribution switch, air conditioning controller, or energy-consuming terminal that supports power reduction. The delayed start command corresponds to loads that can be delayed, and its execution target is an energy-consuming terminal whose start-up time can be adjusted. The off-peak operation command corresponds to loads that can be transferred, and its execution target is an energy-consuming terminal whose operating period can be adjusted to other electricity price periods or electricity consumption periods. The maintain operation state is used for objects that do not meet the adjustment conditions, are rigid loads, or are restricted by user-side constraints. The above command states correspond to the load adjustable attribute field, so that the command generation process does not use a single power threshold as the sole control basis, but combines load attributes, operating states, and constraints to form the control output.

[0037] Please see Figure 5 This embodiment provides a building power demand management system based on big data, which supports the aforementioned building power demand management method based on big data. The system includes a multi-source data access unit, a state sequence construction unit, a load profile generation unit, a demand forecasting unit, a scheduling queue generation unit, a control command generation unit, and an execution feedback acquisition unit. The multi-source data access unit communicates with smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces, receiving total power records, sub-item power records, temperature records, switch status records, and time-of-use price records, and sending the multi-source power consumption data to the state sequence construction unit. The state sequence construction unit performs time alignment and field matching on the multi-source power consumption data, outputting a building power consumption state sequence. The load profile generation unit generates a load profile based on the building power consumption state sequence and writes the load adjustable attribute fields. The demand forecasting unit generates demand forecast values ​​based on the building power consumption state sequence, ambient temperature changes, and the status of executed control commands. The scheduling queue generation unit generates an adjustable load scheduling queue based on time-of-use price, load category, area priority, equipment operating status, continuous operating time, and user-side constraints. The control command generation unit converts the adjustable load scheduling queue into hierarchical control commands and transmits them to the power distribution switch, air conditioning controller, or energy terminal via the building controller. The execution feedback acquisition unit receives power changes, switch status, and non-execution reason identifiers, and transmits the execution feedback to the demand forecasting unit and the scheduling queue generation unit.

[0038] During system operation, the multi-source data access unit serves as the data entry point, the state sequence construction unit as the unified data processing node, the load profile generation unit as the load attribute generation node, the demand forecasting unit as the demand result generation node for the next control cycle, the scheduling queue generation unit as the control object sorting node, the control command generation unit as the control output node, and the execution feedback acquisition unit as the feedback return node. The data flow between these units is consistent with the aforementioned method steps: multi-source electricity consumption data enters the state sequence construction unit to form a building electricity consumption state sequence; the building electricity consumption state sequence enters the load profile generation unit and the demand forecasting unit; the load profile and demand forecast values ​​enter the scheduling queue generation unit; the adjustable load scheduling queue enters the control command generation unit; hierarchical control commands are output to the execution objects; and execution feedback is returned to the demand forecasting unit and the scheduling queue generation unit. Thus, the system carries the input, processing, output, and feedback correction processes of the aforementioned method through the data transmission relationships between the units.

[0039] In the above specific embodiments, the method steps, system units, data fields, status records, judgment conditions, instruction types, and feedback relationships are only used to explain the possible implementations of the present invention and should not limit the present invention to the specific embodiments listed. Without departing from the scope of the claims and the original disclosure, any equivalent substitutions, equivalent modifications, equivalent combinations, order adjustments, module-to-module replacements, equivalent transformations of field names, equivalent acceptance of the execution subject, or equivalent changes in the carrier form that can be conceived by those skilled in the art should fall within the scope of protection of this patent; however, the aforementioned equivalent changes should not extend to unclaimed topics, and should not alter the substantive correspondence between multi-source power consumption data, building power consumption status sequences, load profiles, demand forecasts, adjustable load scheduling queues, hierarchical control instructions, and execution feedback records through name changes.

Claims

1. A building power demand management method based on big data, characterized in that, Acquire multi-source electricity consumption data generated by smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces; The multi-source electricity consumption data is time-aligned and field-matched according to a unified sampling time window to generate a building electricity consumption status sequence. A load profile is generated based on the area identifier, equipment category, operating status, and historical electricity consumption records. Based on the load profile, the load is classified into rigid loads, delayable loads, reduceable loads, and transferable loads. Based on the building power consumption status sequence, ambient temperature changes, and the status of executed control commands, a demand forecast value for the next control cycle is generated. An adjustable load scheduling queue is generated based on time-of-use pricing, load category, regional priority, equipment operating status, continuous operating time, and user-side constraints, and hierarchical control instructions are output. The system receives power changes, switch status, and non-execution reason identifiers after the execution of the hierarchical control command. When the deviation between the actual power consumption value and the predicted demand value meets the correction trigger condition, the system adjusts the scheduling weight for the next control cycle.

2. The building power demand management method based on big data according to claim 1, characterized in that, The steps for time alignment and field matching of the multi-source electricity consumption data according to a unified sampling time window include: A unified sampling time window is determined by the electricity price time period boundary and the control cycle of the building controller. The total power record collected by the smart meter, the sub-item power record collected by the sub-metering device, the temperature record collected by the environmental sensor, the switch status record collected by the equipment controller, and the time-of-use electricity price record provided by the electricity price data interface are written into the corresponding time window respectively. When there are multiple records of the same device or the same area in the same time window, the record whose collection timestamp is close to the end of the time window is used as the field matching object. When the field matching object is missing the area identifier or equipment category, the area identifier or equipment category corresponding to the equipment number in the historical electricity consumption record is called to supplement it and generate the building electricity consumption status sequence.

3. The building power demand management method based on big data according to claim 2, characterized in that, The building electricity consumption status sequence includes a total power field, a component electricity consumption field, an ambient temperature field, an equipment switch status field, a region identifier field, an equipment category field, and a time-of-use electricity price field, all arranged according to a unified sampling time window. When the electricity consumption change direction corresponding to the total power field is consistent with the electricity consumption change direction corresponding to the component electricity consumption field, the corresponding time window is marked as a time window that can participate in demand forecasting. When the electricity consumption change direction corresponding to the total power field is inconsistent with the electricity consumption change direction corresponding to the component electricity consumption field, the corresponding time window is marked as a time window to be verified, and the time window to be verified is excluded from the direct calculation input of the current demand forecast value.

4. The building power demand management method based on big data according to claim 1, characterized in that, The steps for generating the load profile based on area identification, equipment category, operating status, and historical electricity consumption records include: The building's electricity consumption status sequence is grouped according to area identifier and equipment category to form an area equipment combination record. The area equipment combination record is written with continuous running time, historical electricity consumption in the same period, number of switch status changes, and the percentage of time-of-use electricity price periods. The load profile record is generated based on the correspondence between continuous running time and historical electricity consumption in the same period, and the load adjustable attribute field is written into the load profile record. The load adjustable attribute field is used to distinguish rigid loads, delayable loads, loads that can be reduced, and loads that can be transferred.

5. The building power demand management method based on big data according to claim 4, characterized in that, The writing process of the load adjustable attribute field includes: When the equipment category corresponding to the regional equipment combination record is a safe power supply equipment or a continuously operating equipment, the load adjustable attribute field is written as a rigid load; when the equipment category corresponding to the regional equipment combination record has a variable start time and the number of switch state changes meets the delay control condition, the load adjustable attribute field is written as a delayable load; when the continuous running time corresponding to the regional equipment combination record meets the reduction control condition and the electricity consumption of the same historical period meets the reduction judgment condition, the load adjustable attribute field is written as a reduceable load; when the time-of-use electricity price period ratio corresponding to the regional equipment combination record meets the peak-shifting judgment condition, the load adjustable attribute field is written as a transferable load.

6. The building power demand management method based on big data according to claim 1, characterized in that, The steps for generating the demand forecast value for the next control cycle based on the building power consumption status sequence, ambient temperature changes, and the status of executed control commands include: Extract real-time power records and sub-item power records adjacent to the next control cycle from the building power consumption status sequence; extract historical power consumption records of the same period in the same area, equipment category, and electricity price period as the next control cycle from historical power consumption records; correct the historical power consumption records of the same period for air conditioning loads based on the temperature difference corresponding to the ambient temperature change; and generate demand forecast values ​​by deducting or delaying the status of the prediction input records of the corresponding equipment category based on the load limit, delayed start, or staggered operation flags corresponding to the executed control command status.

7. The building power demand management method based on big data according to claim 6, characterized in that, After the next control cycle ends, the system receives the total power record collected by the smart meter and the sub-item power record collected by the sub-item metering device, and converts the total power record and sub-item power record into the actual power consumption value. Compare the deviation between the actual electricity consumption value and the demand forecast value. When the deviation meets the correction trigger condition, use the area identifier, equipment category and executed control command status corresponding to the deviation as the basis for a single update to adjust the scheduling weight of the next control cycle. When the deviation does not meet the correction triggering condition, the scheduling weight of the previous control cycle is used.

8. The building power demand management method based on big data according to claim 1, characterized in that, The steps of generating the adjustable load scheduling queue based on time-of-use pricing, load category, area priority, equipment operating status, continuous operating duration, and user-side constraints, and outputting the hierarchical control instructions, include: Candidate scheduling records are established for deferred loads, loads that can be reduced, and loads that can be transferred. The time-of-use tariff level, load category, regional priority, equipment operating status, continuous operating time, and user-side constraints are written into the candidate scheduling record. When the candidate scheduling record meets the user-side constraints and the equipment operating status allows adjustment, the queue position is determined according to the regional priority and time-of-use tariff level. When the candidate scheduling record does not meet the user-side constraints or the equipment operating status does not allow adjustment, the candidate scheduling record is excluded from the current adjustable load scheduling queue.

9. The building power demand management method based on big data according to claim 8, characterized in that, The process of outputting the hierarchical control commands includes: Candidate scheduling records are read sequentially according to their queue positions in the adjustable load scheduling queue. When a candidate scheduling record corresponds to a load that can be reduced, a load limiting instruction is generated. When a candidate scheduling record corresponds to a load that can be delayed, a delayed start instruction is generated. When a candidate scheduling record corresponds to a load that can be transferred, a peak-shifting operation instruction is generated. The power changes, switch statuses, and non-execution reason identifiers after the execution of the load limiting instruction, delayed start instruction, or peak-shifting operation instruction are received, and the power changes, switch statuses, and non-execution reason identifiers are written into the execution feedback record corresponding to the candidate scheduling record.

10. A building power demand management system based on big data, used to support the building power demand management method based on big data as described in any one of claims 1 to 9, characterized in that, It includes a multi-source data access unit, a state sequence construction unit, a load profile generation unit, a demand prediction unit, a scheduling queue generation unit, a control instruction generation unit, and an execution feedback acquisition unit; The multi-source data access unit communicates with smart meters, sub-metering devices, environmental sensors, equipment controllers, and electricity price data interfaces, and sends multi-source electricity consumption data to the state sequence construction unit; The state sequence construction unit performs time alignment and field matching on the multi-source electricity consumption data and outputs a building electricity consumption state sequence. The load profile generation unit generates a load profile based on the building's power consumption status sequence; The demand forecasting unit generates demand forecast values, the scheduling queue generation unit generates adjustable load scheduling queues, the control instruction generation unit outputs hierarchical control instructions, and the execution feedback acquisition unit transmits execution feedback to the demand forecasting unit and the scheduling queue generation unit.

Images