Load score determination method and device for smart building, equipment and medium

Abstract

The invention provides a load score determination method and device for an intelligent building, equipment and a medium, and the method comprises the steps: inputting a total load data curve of the intelligent building into a model, carrying out the decomposition of a load curve, and outputting the power utilization characteristics of various types of loads after decomposition; a plurality of load indexes at each moment are extracted from the electricity utilization characteristics of the various loads, the generalized additive model carries out weighted aggregation processing on the load indexes based on the dynamic weighting factors of the various loads at each moment, and the aggregated load indexes of the smart building at each moment are generated; and weighting the aggregated load indexes at each moment based on an analytic hierarchy process, and determining the load score of the smart building. By constructing a multi-dimensional flexible index and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, so that the flexibility capability of a building in different environments and states is accurately reflected, and a power grid is supported to realize differentiated scheduling and efficient response in actual operation.

Description

Technical Field

[0001] The present application relates to the technical field of energy Internet of Things, and in particular to a method, device, equipment and medium for determining a load score of a smart building. Background Art

[0002] Current analysis technologies for smart building loads mainly rely on two methods: intrusive load monitoring (ILM) and non-intrusive load monitoring (NILM). Intrusive methods usually obtain detailed load data by installing collection equipment in each load circuit, but their installation and maintenance costs are high, they are highly invasive to the existing system, and are difficult to deploy on a large scale. Non-intrusive methods extract load characteristics through overall meter data. Common methods include combinatorial optimization algorithms (such as particle swarm algorithms and chicken swarm algorithms) and machine learning methods (such as Markov models and shallow neural networks). These two methods have the following problems: insufficient decomposition capabilities make it difficult to adapt to the complex load structure of the building, and the evaluation dimension is single, making it difficult to support flexible scheduling and automated decision-making of the power grid. Therefore, how to accurately determine the load score of a smart building has become a technical issue that cannot be underestimated. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, device, equipment and medium for determining the load score of a smart building. By constructing a multi-dimensional flexibility index and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, thereby accurately reflecting the flexibility capabilities of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation.

[0004] The present application provides a method for determining a load score of a smart building. The method includes:

[0005] The total load data curve of the smart building is input into the KAN-Transformer model to decompose the load curve and output the power consumption characteristics of each type of load after decomposition;

[0006] Extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, and performing weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment using a generalized additive model to generate an aggregated load indicator for the smart building at each moment;

[0007] Based on the hierarchical analysis algorithm, the aggregated load indicators at each moment are weighted to determine the load score of the smart building at each moment.

[0008] In one possible implementation, the total load data curve of the smart building is input into the KAN-Transformer model for load curve decomposition, and the decomposed power consumption characteristics of each type of load are output, including:

[0009] Based on the nonlinear modeling network layer of the KAN-Transformer model, a nonlinear mapping process is performed on the total load data curve of the smart building to construct a nonlinear load feature;

[0010] The Transformer network layer based on the KAN-Transformer model performs self-attention mechanism time series modeling processing on the nonlinear load characteristics, and outputs the decomposed power consumption characteristics of various loads.

[0011] In a possible implementation, for each type of load, a dynamic weighting factor at each moment is determined by the following steps:

[0012] Determine a load state correction factor of the load at the moment based on the power regulation amount, maximum power capacity, and state correction factor of the load at the moment;

[0013] Determine the comfort correction factor of the load at the moment based on the actual environmental quantity at the moment, the comfort setting value of the load at the moment, and the adjustment coefficient;

[0014] Based on the sensitivity of the load to environmental fluctuations and the degree of influence of external environmental factors on the load at that moment, an external factor correction factor of the load at that moment is determined;

[0015] Determine a load complementarity correction factor for the load at the moment based on a power sequence correlation coefficient between the load and other loads and a sensitivity value of a complementarity relationship between the load and other loads at the moment;

[0016] The load state correction factor, the comfort correction factor, the external factor correction factor, and the load complementarity correction factor are multiplied together to determine a dynamic weighting factor at that moment.

[0017] In one possible implementation, when the load index is a regulated power index, the generalized additive model performs weighted aggregation processing on the load index based on dynamic weighting factors of various types of loads at each moment to generate an aggregated load index of the smart building at each moment, including:

[0018] The generalized additive model performs weighted processing on the regulation power index of the corresponding load at each moment by using the dynamic weighting factor of each type of load at that moment, and determines the weighted regulation power index of each type of load at each moment;

[0019] The weighted regulation power indicators of various loads at each moment are aggregated to generate the aggregated regulation power indicator of the smart building at each moment.

[0020] In one possible implementation, the load indicator includes at least one of the following:

[0021] Adjust power index, response time index, climbing ability index and duration index.

[0022] In a possible implementation, when the load index is a regulating power index, multiple load indexes at each moment are extracted from the power consumption characteristics of each type of load using the following formula:

[0023] ΔP dr (t) = P fle (t)-P ref (t)

[0024] Among them, P fle (t) is the actual load power at time t, P ref (t) is the reference power at the current moment, ΔP dr (t) is the regulation power index at time t.

[0025] In one possible implementation, the weighted processing of the aggregated load index at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment includes:

[0026] Comparing the plurality of load indicators, determining a scale value of each load indicator, and constructing an indicator judgment matrix based on a transitivity calculation of the scale value of each load indicator;

[0027] Normalizing the aggregated load index at each moment to determine the normalized load index at each moment;

[0028] Calculating the indicator judgment matrix based on the hierarchical analysis algorithm to determine the weight value of each load indicator;

[0029] Based on the weight values ​​of the load indicators and the normalized load indicators at each moment, the load score of the smart building at each moment is determined.

[0030] The present application also provides a device for determining a load score of a smart building. The device includes:

[0031] The load decomposition module is used to input the total load data curve of the smart building into the KAN-Transformer model for load curve decomposition processing, and output the power consumption characteristics of each type of load after decomposition;

[0032] An index extraction module is used to extract multiple load indicators at each moment from the power consumption characteristics of each type of load, and a generalized additive model performs weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment to generate an aggregated load indicator for the smart building at each moment;

[0033] The indicator weighting module is used to perform weighted processing on the aggregated load indicators at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment.

[0034] An embodiment of the present application also provides an electronic device, comprising: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the steps of the load score determination method for a smart building as described above are performed.

[0035] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the computer program executes the steps of the method for determining the load score of a smart building as described above.

[0036] The embodiments of the present application provide a method, device, equipment and medium for determining the load score of a smart building. The load score determination method includes: inputting the total load data curve of the smart building into the KAN-Transformer model to perform load curve decomposition processing, and outputting the power consumption characteristics of each type of load after decomposition; extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, and using a generalized additive model to perform weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment to generate the aggregated load indicators of the smart building at each moment; and performing weighted processing on the aggregated load indicators at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment. By constructing multi-dimensional flexibility indicators and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, thereby accurately reflecting the flexibility capabilities of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation.

[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 This is one of the flow charts of a method for determining a load score of a smart building provided in an embodiment of the present application;

[0040] Figure 2 This is a second flow chart of a method for determining a load score of a smart building provided in an embodiment of the present application;

[0041] Figure 3 A schematic diagram of the structure of a load score determination device for a smart building provided in an embodiment of the present application;

[0042] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work falls within the scope of protection of the present application.

[0044] First, the application scenarios to which this application is applicable are introduced. This application can be applied in the field of energy Internet of Things technology.

[0045] Research has found that the current analysis technology for smart building loads mainly relies on two methods: intrusive load monitoring (ILM) and non-intrusive load monitoring (NILM). Intrusive methods usually install collection equipment in each load circuit to obtain detailed load data, but their installation and maintenance costs are high, they are highly invasive to the existing system, and are difficult to deploy on a large scale. Non-intrusive methods extract load characteristics through overall meter data. Common methods include combinatorial optimization algorithms (such as particle swarm algorithms and chicken swarm algorithms) and machine learning methods (such as Markov models and shallow neural networks). These two methods have the following problems: insufficient decomposition capabilities make it difficult to adapt to the complex load structure of the building, and the evaluation dimension is single, making it difficult to support flexible scheduling and automated decision-making of the power grid. Therefore, how to accurately determine the load score of a smart building has become a technical issue that cannot be underestimated.

[0046] Based on this, the embodiments of the present application provide a method, device, equipment and medium for determining the load score of a smart building. By constructing multi-dimensional flexibility indicators and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, thereby accurately reflecting the flexibility capabilities of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation.

[0047] See also Figure 1 , Figure 1 This is one of the flow charts of a method for determining a load score of a smart building provided in an embodiment of the present application. Figure 1 As shown in , the load score determination method provided in the embodiment of the present application includes:

[0048] S101: Input the total load data curve of the smart building into the KAN-Transformer model to perform load curve decomposition processing, and output the power consumption characteristics of each type of load after decomposition.

[0049] In this step, the KAN-Transformer model is used to process the total load data curve of the smart building, and the original total power curve is decomposed into the power consumption characteristics of multiple types of typical loads.

[0050] Among them, the total load data curve is the original time series data of the total load of the smart building.

[0051] In one possible implementation, the total load data curve of the smart building is input into the KAN-Transformer model for load curve decomposition, and the decomposed power consumption characteristics of each type of load are output, including:

[0052] A: Based on the nonlinear modeling network layer of the KAN-Transformer model, the total load data curve of the smart building is nonlinearly mapped to construct a nonlinear load feature.

[0053] Here, the input of the nonlinear modeling network layer is the original time series data of the total load of the smart building, and the complex nonlinear load characteristics are constructed through B-spline nonlinear mapping.

[0054] B: Based on the Transformer network layer of the KAN-Transformer model, the nonlinear load characteristics are subjected to time series modeling processing using a self-attention mechanism, and the decomposed power consumption characteristics of each type of load are output.

[0055] Here, the input of the Transformer layer is the nonlinear load characteristics processed by the nonlinear modeling network layer. Time series modeling is performed through the self-attention mechanism to output the power consumption characteristics of various decomposed loads.

[0056] The load decomposition process of the KAN-Transformer model uses the KAN layer (nonlinear modeling network layer) to perform nonlinear feature extraction on the input data. This, combined with the Transformer model's self-attention mechanism, accurately identifies the characteristics of flexible and basic loads. In the KAN-Transformer model, the input load data first undergoes nonlinear feature extraction and transformation through the KAN layer. This process, through a series of B-spline functions and activation functions, converts the input data into a feature representation that better reflects the load characteristics and variation patterns. Next, after positional encoding, these features are input to the Transformer module, where a multi-head attention mechanism is used to model temporal dependencies and decompose them using an encoder-decoder structure. This architecture leverages the Transformer's strengths in temporal modeling while leveraging the KAN layer's enhanced nonlinear modeling capabilities, resulting in highly accurate load decomposition results.

[0057] S102: Extract multiple load indicators at each moment from the power consumption characteristics of each type of load, and use a generalized additive model to perform weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment to generate the aggregated load indicators of the smart building at each moment.

[0058] In this step, multiple load indicators at each moment are extracted from the power consumption characteristics of various loads. The generalized additive model performs weighted aggregation processing on the load indicators according to the dynamic weighting factors of various loads at each moment to generate the aggregated load indicators of the smart building at each moment.

[0059] In a possible implementation manner, the load index includes at least one of the following: a regulation power index, a response time index, a climbing ability index, and a duration index.

[0060] Here, according to the operating status, comfort constraints, environmental factors and complementary relationships of various loads, indicators including adjustable power, response time, climbing ability and duration are extracted.

[0061] In a possible implementation, when the load index is a regulating power index, multiple load indexes at each moment are extracted from the power consumption characteristics of each type of load using the following formula:

[0062] ΔP dr (t) = P fle (t)-P ref (t)

[0063] Among them, P fle (t) is the actual load power at time t, P ref (t) is the reference power at the current moment, ΔP dr (t) is the regulation power index at time t.

[0064] In one possible implementation, the response time indicator is determined in the following manner:

[0065]

[0066] Where t(ΔP(t)≥90%ΔP tar (t)) is the moment when the building power regulation reaches 90% of the target regulation for the first time, t sig T is the moment when the building receives the adjustment instruction from the power grid or energy system. Res It is the response time indicator.

[0067] In one possible implementation, the gradeability index is determined by:

[0068]

[0069] Among them, P fle (t) is the actual load power (kW) of the building participating in the flexibility response at time t; P tar (t) is the target power of flexibility response at time t (kW), θ RR It is an indicator of climbing ability.

[0070] In one possible implementation, the duration indicator is determined in the following manner:

[0071]

[0072] Where ΔP(t) is the load power adjustment amount at time t during the building's flexibility response period, that is, the absolute value of the difference between the actual power after adjustment and the reference power (kW); ΔPtar (t) is the target value of the power regulation at time t, that is, the absolute value of the difference between the reference power and the target power (kW); t0 is the start time of the flexible response or regulation; t end is the end time of flexibility response or adjustment; dt is the time step, T eff It is an indicator of climbing ability.

[0073] In a possible implementation, for each type of load, a dynamic weighting factor at each moment is determined by the following steps:

[0074] a: Based on the power regulation amount, maximum power capacity and state correction factor of the load at that moment, determine the load state correction factor of the load at that moment.

[0075] Here, the load state correction factor is determined by the following formula:

[0076]

[0077] Among them, w state,i (t) is the load state correction factor at the current time t, ΔP i (t) is the power regulation amount that load i can provide at time t, P max,i is the maximum power capacity of load i, f load,i (t) is the state correction factor of load i at time t, which is used to reflect the impact of factors such as the current load rate and start-stop status of the load on its regulation ability. The state correction factor dynamically adjusts the weight according to the state of the load.

[0078] b: Based on the actual environmental quantity at that moment, the comfort setting value of the load at that moment, and the adjustment coefficient, determine the comfort correction factor of the load at that moment.

[0079] Here, the comfort correction factor is determined by the following formula:

[0080] w comfort,i (t) = 1-α|X real (t)-X set,i |

[0081] Among them, X real (t) is the actual environmental quantity (such as temperature, brightness, etc.), X set,i is the comfort setting value of load i, α is the adjustment coefficient. When the actual environment is farther away from the setting value, the adjustable margin is more constrained, so the weight is smaller, w comfort,i (t) is the comfort correction factor.

[0082] c: Based on the load's sensitivity to environmental fluctuations and the degree of influence of external environmental factors on the load at that moment, determine the external factor correction factor of the load at that moment.

[0083] Here, the external factor correction factor is determined by the following formula:

[0084] w env,i (t) = 1 + γ i ·f env (t)

[0085] Among them, γ i is the sensitivity of load i to environmental fluctuations, f env (t) is the influence of external environmental factors (such as ambient temperature, electricity price fluctuation, etc.) on the load regulation willingness at time t, w env,i (t) is the external factor correction factor.

[0086] d: Based on the power sequence correlation coefficient between the load and other loads at the moment and the sensitivity value of the complementarity relationship between the load and other loads, determine the load complementarity correction factor of the load at the moment.

[0087] Here, the load complementarity correction factor is determined by the following formula:

[0088]

[0089] Among them, β i is the sensitivity of the complementary relationship between load i and other loads, corr(P i (t),P j (t)) is the correlation coefficient of the power series of load i and load j during the regulation period starting from time t, which is used to quantify the relationship between the power fluctuation trends of the two loads. If the correlation coefficient is positive (>0), it means that the power change trends of the two loads are similar, that is, both are high or low at the same time; if the correlation coefficient is negative (<0), it means that the two trends are complementary, and when one increases, the other decreases. In this case, the two can form a good complementary relationship, which is more conducive to overall balance and regulation. comp,i (t) is the load complementarity correction.

[0090] e: Multiply the load state correction factor, the comfort correction factor, the external factor correction factor, and the load complementarity correction factor to determine a dynamic weighting factor at that moment.

[0091] Here, the load state correction factor, the comfort correction factor, the external factor correction factor, and the load complementarity correction factor are multiplied together to determine the dynamic weighting factor at that moment.

[0092] In one possible implementation, when the load index is a regulated power index, the generalized additive model performs weighted aggregation processing on the load index based on dynamic weighting factors of various types of loads at each moment to generate an aggregated load index of the smart building at each moment, including:

[0093] (1) The generalized additive model performs weighted processing on the regulation power index of the corresponding load at each moment by using the dynamic weighting factor of each type of load at that moment, and determines the weighted regulation power index of each type of load at each moment.

[0094] Here, the generalized additive model performs weighted processing on the regulation power index of the corresponding load at each moment through the dynamic weighting factor of each type of load at that moment, and determines the weighted regulation power index of each type of load at each moment.

[0095] (2): Aggregate the weighted regulation power indicators of various loads at each moment to generate the aggregated regulation power indicator of the smart building at each moment.

[0096] Here, the weighted regulation power index at each moment is aggregated to generate the aggregated regulation power index of the smart building at each moment.

[0097] In smart buildings, different types of loads often exhibit heterogeneity and temporal dynamics. To better manage demand response and dispatchability for these loads, a method is needed that accurately characterizes the scalability of various load types over different time periods. A temporal weighted aggregation method based on a generalized additive model leverages the load's adjustable power and corresponding dynamic weighting factors. By considering factors such as load operating status, comfort constraints, hysteresis effects, inter-load complementarity, and external environmental influences, it achieves accurate aggregation and calculation of heterogeneous loads.

[0098] S103: Perform weighted processing on the aggregated load index at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment.

[0099] In this step, the hierarchical analysis algorithm is used to perform weighted processing on the aggregated load indicators at each moment to determine the load score of the smart building at each moment.

[0100] See also Figure 2 , Figure 1 This is one of the flow charts of a method for determining a load score of a smart building provided in an embodiment of the present application. Figure 2 As shown in , the load score determination method provided in the embodiment of the present application includes:

[0101] S201: Compare the multiple load indices, determine the scale value of each load indices, and construct an indicator judgment matrix based on the transitivity calculation of the scale value of each load indices.

[0102] In this step, various evaluation indicators are compared according to expert opinions or user requirements, and sorted in a non-decreasing order of importance. i with x i+1 Compare and record the corresponding scale value as t i , and then calculate the other element values ​​in the judgment matrix according to the transitivity of the indicator importance, so as to obtain the indicator judgment matrix.

[0103] A scale value of 1 indicates equal importance, a scale value of 1.2 indicates slightly important, and a scale value of 1.4 indicates strongly important.

[0104] S202: performing normalization processing on the aggregated load index at each moment to determine the normalized load index at each moment.

[0105] Here, the adjustable power index is normalized by the following formula:

[0106]

[0107] Among them, P i is the adjustable power of load i; P max is the maximum adjustable power, Ω p (t i ) is the normalized adjustable power index at time t.

[0108] Here, the response time indicator is normalized by the following formula:

[0109]

[0110] Among them, T Res,max is the maximum acceptable response time in this scenario, T Res,i is the actual response time of load i, Ω T (t i ) is the normalized response time indicator at time t.

[0111] Here, the climbing ability index is normalized by the following formula:

[0112]

[0113] Among them, θ i is the actual climbing ability of load i, θ max is the maximum climbing ability, Ω R (ri ) is the normalized climbing ability index.

[0114] Here, the duration indicator is normalized by the following formula:

[0115]

[0116] Among them, T eff,i is the duration of load i, T eff,max is the maximum duration, Ω D (d i ) is the duration indicator after normalization.

[0117] S203: Calculating the indicator judgment matrix based on the hierarchical analysis algorithm to determine the weight value of each load indicator.

[0118] In this step, the index judgment matrix is ​​calculated using a hierarchical analysis algorithm to determine the weight value of each load index.

[0119] In this way, the weight value of each load index is determined by the following formula:

[0120]

[0121] Among them, w i is the weight value of the i-th load index, It represents the product of all elements in the i-th row of the judgment matrix R, and n is the number of load indicators.

[0122] S204: Based on the weight values ​​of the load indicators and the normalized load indicators at each moment, determine the load score of the smart building at each moment.

[0123] In this step, the load score of the smart building at each moment is determined based on the weight value of each load indicator and the normalized load indicator at each moment.

[0124] Here, the score S of the smart building is determined through the following steps:

[0125] S=w p Ω p +w r Ω r +w c Ω c +w t Ω t

[0126] Among them, Ω P is the normalized adjustment power index, Ω ris the normalized response time index, Ω c is the normalized climbing ability index, Ω t is the duration index after normalization, w P To adjust the weight value of the power index, w r is the weight value of the response time indicator, w c is the weight value of the climbing ability index, w t is the weight value of the duration indicator.

[0127] In this application, the traditional hierarchical analysis algorithm may fail the consistency check when constructing the judgment matrix, affecting the reliability of the evaluation results. To this end, this application introduces a scaling construction method to improve the hierarchical analysis algorithm. By directly constructing the judgment matrix using a predetermined scale, the inherent consistency of the matrix is ​​ensured, avoiding the complexity of consistency checks in traditional methods and improving the scientific nature and accuracy of the evaluation process.

[0128] In the specific implementation method, first, the total load electricity consumption data and related auxiliary information of the smart building are collected; secondly, the KAN-Transformer model is used to process the original load curve to achieve non-invasive decomposition of multiple types of typical loads; then, based on factors such as the operating status, comfort, electricity price sensitivity and load complementarity of various loads, a generalized additive model is used to construct a time series weighting mechanism to extract multi-dimensional flexibility indicators including adjustable power, response time, climbing ability and duration; then, the above indicators are normalized and weighted through the improved hierarchical analysis method to calculate the comprehensive flexibility score for each time period; finally, the flexibility time series score result is output for reference by the power supply network dispatching system to achieve quantitative evaluation and dispatching optimization of the load regulation capability of the smart building.

[0129] In this application, the application is able to decompose the operating status of various typical loads in real time based on the total load data of the smart building, and by constructing a multi-dimensional flexibility index and a dynamic weighting mechanism, quantitatively evaluate the load regulation potential of each time period, thereby accurately reflecting the flexibility of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation. Furthermore, the KAN-Transformer model introduced in this application has nonlinear modeling and time-series correlation learning capabilities. It can complete the overall load deconstruction and evaluation without the need to deploy hardware collection devices for each type of load separately, thereby improving system deployment efficiency, reducing implementation costs, and enhancing the adaptability and intelligence level of the load flexibility quantification process.

[0130] An embodiment of the present application provides a method for determining a load score of a smart building, which includes: inputting the total load data curve of the smart building into a KAN-Transformer model to perform load curve decomposition processing, and outputting the power consumption characteristics of each type of load after decomposition; extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, and using a generalized additive model to weightedly aggregate the load indicators based on the dynamic weighting factors of each type of load at each moment to generate the aggregated load indicators of the smart building at each moment; and weighting the aggregated load indicators at each moment based on a hierarchical analysis algorithm to determine the load score of the smart building at each moment.

[0131] The embodiment of the present application provides a method for determining the load score of a smart building, which includes: inputting the total load data curve of the smart building into the KAN-Transformer model to perform load curve decomposition processing, and outputting the power consumption characteristics of each type of load after decomposition; extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, and using a generalized additive model to perform weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment to generate the aggregated load indicators of the smart building at each moment; and performing weighted processing on the aggregated load indicators at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment. By constructing multi-dimensional flexibility indicators and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, thereby accurately reflecting the flexibility capabilities of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation.

[0132] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a load score determination device for a smart building provided in an embodiment of the present application. Figure 3 As shown in , the load score determination device 300 of the smart building includes:

[0133] The load decomposition module 310 is used to input the total load data curve of the smart building into the KAN-Transformer model to perform load curve decomposition processing and output the power consumption characteristics of each type of load after decomposition;

[0134] An index extraction module 320 is configured to extract multiple load indices at each moment from the power consumption characteristics of each type of load, and perform weighted aggregation processing on the load indices based on the dynamic weighting factors of each type of load at each moment using a generalized additive model to generate an aggregated load index for the smart building at each moment;

[0135] The index weighting module 330 is used to perform weighted processing on the aggregated load index at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment.

[0136] Furthermore, when the load decomposition module 310 is used to input the total load data curve of the smart building into the KAN-Transformer model to perform load curve decomposition processing and output the power consumption characteristics of various loads after decomposition, the load decomposition module 310 is specifically used to:

[0137] Based on the nonlinear modeling network layer of the KAN-Transformer model, a nonlinear mapping process is performed on the total load data curve of the smart building to construct a nonlinear load feature;

[0138] The Transformer network layer based on the KAN-Transformer model performs self-attention mechanism time series modeling processing on the nonlinear load characteristics, and outputs the decomposed power consumption characteristics of various loads.

[0139] Furthermore, the indicator extraction module 320 determines the dynamic weighting factor at each moment through the following steps:

[0140] Determine a load state correction factor of the load at the moment based on the power regulation amount, maximum power capacity, and state correction factor of the load at the moment;

[0141] Determine the comfort correction factor of the load at the moment based on the actual environmental quantity at the moment, the comfort setting value of the load at the moment, and the adjustment coefficient;

[0142] Based on the sensitivity of the load to environmental fluctuations and the degree of influence of external environmental factors on the load at that moment, an external factor correction factor of the load at that moment is determined;

[0143] Determine a load complementarity correction factor for the load at the moment based on a power sequence correlation coefficient between the load and other loads and a sensitivity value of a complementarity relationship between the load and other loads at the moment;

[0144] The load state correction factor, the comfort correction factor, the external factor correction factor, and the load complementarity correction factor are multiplied together to determine a dynamic weighting factor at that moment.

[0145] Furthermore, when the load index is a regulated power index, the generalized additive model performs weighted aggregation processing on the load index based on the dynamic weighting factors of various load types at each moment to generate the aggregated load index of the smart building at each moment. The index extraction module 320 is specifically configured to:

[0146] The generalized additive model performs weighted processing on the regulation power index of the corresponding load at each moment by using the dynamic weighting factor of each type of load at that moment, and determines the weighted regulation power index of each type of load at each moment;

[0147] The weighted regulation power indicators of various loads at each moment are aggregated to generate the aggregated regulation power indicator of the smart building at each moment.

[0148] Furthermore, the index extraction module 320 extracts multiple load indexes at each moment from the power consumption characteristics of each type of load using the following formula:

[0149] ΔP dr (t) = P fle (t)-P ref (t)

[0150] Among them, P fle (t) is the actual load power at time t, P ref (t) is the reference power at the current moment, ΔP dr (t) is the regulation power index at time t.

[0151] Furthermore, when the indicator weighting module 330 is used to perform weighted processing on the aggregated load indicators at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment, the indicator weighting module 330 is specifically used to:

[0152] Comparing the plurality of load indicators, determining a scale value of each load indicator, and constructing an indicator judgment matrix based on a transitivity calculation of the scale value of each load indicator;

[0153] Normalizing the aggregated load index at each moment to determine the normalized load index at each moment;

[0154] Calculating the indicator judgment matrix based on the hierarchical analysis algorithm to determine the weight value of each load indicator;

[0155] Based on the weight values ​​of the load indicators and the normalized load indicators at each moment, the load score of the smart building at each moment is determined.

[0156] The embodiment of the present application provides a load score determination device for a smart building, the load score determination device comprising: a load decomposition module for inputting the total load data curve of the smart building into a KAN-Transformer model for load curve decomposition processing, and outputting the power consumption characteristics of each type of load after decomposition; an index extraction module for extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, a generalized additive model for weighted aggregation processing of the load indicators based on the dynamic weighting factors of each type of load at each moment, and generating the aggregated load indicators of the smart building at each moment; an index weighting module for weighting the aggregated load indicators at each moment based on a hierarchical analysis algorithm to determine the load score of the smart building at each moment. By constructing a multi-dimensional flexibility index and a dynamic weighting mechanism, the load regulation potential of each time period is quantitatively evaluated, thereby accurately reflecting the flexibility capabilities of the building in different environments and states, and supporting the power grid to achieve differentiated scheduling and efficient response in actual operation.

[0157] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown in FIG, the electronic device 400 includes a processor 410 , a memory 420 and a bus 430 .

[0158] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, the above-mentioned Figure 1 as well as Figure 2 The steps of the method for determining the load score of a smart building in the method embodiment shown are specifically implemented in the method embodiment and will not be repeated here.

[0159] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 as well as Figure 2 The steps of the method for determining the load score of a smart building in the method embodiment shown are specifically implemented in the method embodiment and will not be repeated here.

[0160] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0161] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. There may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interface, indirect coupling or communication connection of devices or units, which may be electrical, mechanical or other forms.

[0162] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0163] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0164] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0165] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The scope of protection of the present application is not limited thereto. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-mentioned embodiments within the technical scope disclosed in the present application, or perform equivalent replacements for some of the technical features thereof. These modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for determining a load score of a smart building, characterized in that: The load score determination method includes: The total load data curve of the smart building is input into the KAN-Transformer model for load curve decomposition, and the power consumption characteristics of each type of load after decomposition are output; Extracting multiple load indicators at each moment from the power consumption characteristics of each type of load, and performing weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment using a generalized additive model to generate an aggregated load indicator for the smart building at each moment; Based on the hierarchical analysis algorithm, the aggregated load indicators at each moment are weighted to determine the load score of the smart building at each moment.

2. The load score determination method according to claim 1, characterized in that: The total load data curve of the smart building is input into the KAN-Transformer model for load curve decomposition, and the power consumption characteristics of each type of load after decomposition are output, including: Based on the nonlinear modeling network layer of the KAN-Transformer model, a nonlinear mapping process is performed on the total load data curve of the smart building to construct a nonlinear load feature; The Transformer network layer based on the KAN-Transformer model performs self-attention mechanism time series modeling processing on the nonlinear load characteristics, and outputs the decomposed power consumption characteristics of various loads.

3. The load score determination method according to claim 1, wherein: For each type of load, the dynamic weighting factor at each moment is determined by the following steps: Determine a load state correction factor of the load at the moment based on the power regulation amount, maximum power capacity, and state correction factor of the load at the moment; Determine the comfort correction factor of the load at the moment based on the actual environmental quantity at the moment, the comfort setting value of the load at the moment, and the adjustment coefficient; Based on the sensitivity of the load to environmental fluctuations and the degree of influence of external environmental factors on the load at that moment, an external factor correction factor of the load at that moment is determined; Determine a load complementarity correction factor for the load at the moment based on a power sequence correlation coefficient between the load and other loads and a sensitivity value of a complementarity relationship between the load and other loads at the moment; The load state correction factor, the comfort correction factor, the external factor correction factor, and the load complementarity correction factor are multiplied together to determine a dynamic weighting factor at that moment.

4. The load score determination method according to claim 1, wherein: When the load index is a regulated power index, the generalized additive model performs weighted aggregation processing on the load index based on the dynamic weighting factors of various load types at each moment to generate the aggregated load index of the smart building at each moment, including: The generalized additive model performs weighted processing on the regulation power index of the corresponding load at each moment by using the dynamic weighting factor of each type of load at that moment, and determines the weighted regulation power index of each type of load at each moment; The weighted regulation power indicators of various loads at each moment are aggregated to generate the aggregated regulation power indicator of the smart building at each moment.

5. The load score determination method according to claim 1, wherein: The load indicator includes at least one of the following: Adjust power index, response time index, climbing ability index and duration index.

6. The load score determination method according to claim 1, wherein: In view of the fact that the load index is a regulating power index, multiple load indexes at each moment are extracted from the power consumption characteristics of each type of load using the following formula: ΔP dr (t)=P fle (t)-P ref (t) Among them, P fle (t) is the actual load power at time t, P ref (t) is the reference power at the current moment, ΔP dr (t) is the regulation power index at time t.

7. The load score determination method according to claim 1, characterized in that: The weighted processing of the aggregated load index at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment includes: Comparing the plurality of load indicators, determining a scale value of each load indicator, and constructing an indicator judgment matrix based on a transitivity calculation of the scale value of each load indicator; Normalizing the aggregated load index at each moment to determine the normalized load index at each moment; Calculating the indicator judgment matrix based on the hierarchical analysis algorithm to determine the weight value of each load indicator; Based on the weight values ​​of the load indicators and the normalized load indicators at each moment, the load score of the smart building at each moment is determined.

8. A load score determination device for a smart building, characterized in that: The load score determining device comprises: The load decomposition module is used to input the total load data curve of the smart building into the KAN-Transformer model for load curve decomposition processing, and output the power consumption characteristics of each type of load after decomposition; An index extraction module is used to extract multiple load indicators at each moment from the power consumption characteristics of each type of load, and a generalized additive model performs weighted aggregation processing on the load indicators based on the dynamic weighting factors of each type of load at each moment to generate an aggregated load indicator for the smart building at each moment; The indicator weighting module is used to perform weighted processing on the aggregated load indicators at each moment based on the hierarchical analysis algorithm to determine the load score of the smart building at each moment.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate through the bus. When the processor is running, the machine-readable instructions execute the steps of the method for determining the load score of a smart building as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for determining a load score of a smart building according to any one of claims 1 to 7 are executed.

Images