Multi-scene energy consumption dynamic prediction method for building photovoltaic curtain wall

Abstract

The application discloses a building photovoltaic curtain wall multi-scene energy consumption dynamic prediction method, relates to the technical field of building photovoltaic curtain wall energy consumption prediction, and comprises the following steps: performing unified time scale alignment and proportional normalization processing on photovoltaic power generation data and building load data of the building photovoltaic curtain wall, decomposing net energy consumption into a photovoltaic power generation contribution proportion and a building load contribution proportion, and constructing a net energy consumption contribution proportion evolution matrix; based on the net energy consumption contribution proportion evolution matrix, extracting a proportional displacement vector of the photovoltaic power generation contribution proportion and the building load contribution proportion in a time sequence, and calculating a proportional displacement consistency degree index. The application solves the problem that in the building photovoltaic curtain wall multi-scene energy consumption dynamic prediction, net energy consumption appearance stability cannot be identified, but internal contribution proportions are structurally replaced, realizes structural adaptive prediction effect based on dynamic adjustment of a replacement degree, multi-scene judgment boundary and synchronous update of prediction parameter weights.

Description

Technical Field

[0001] This invention relates to the field of building photovoltaic curtain wall energy consumption prediction technology, specifically to a method for dynamic prediction of energy consumption of building photovoltaic curtain walls in multiple scenarios. Background Technology

[0002] Multi-scenario dynamic energy consumption prediction for building-integrated photovoltaic (BIPV) curtain wall systems refers to an intelligent calculation method that comprehensively considers the coupling relationship between photovoltaic power generation behavior and the building's own energy consumption characteristics under various application scenarios, including different climate conditions, operating conditions, load modes, and energy management strategies. This method uses a data-driven model to predict the overall or net energy consumption of the building in real time or on a rolling basis. This prediction not only involves the building load variation pattern but also requires simultaneous analysis of multi-dimensional factors such as solar irradiance, ambient temperature, building orientation, thermal performance of the building envelope, and the operating status of energy-consuming equipment, thereby achieving a dynamic depiction of the coordinated changes in power generation and consumption. Existing technologies typically acquire meteorological data, photovoltaic module output power data, and building energy consumption data through data acquisition systems. Historical data is then cleaned and features extracted to establish predictive models based on regression analysis, time series models, or machine learning algorithms. In multi-scenario applications, different operating models are generally established through scenario segmentation or parameter pre-setting. The appropriate model is then selected for prediction calculation based on real-time monitoring data. Simultaneously, a rolling update mechanism is used to compare the actual operating data with the prediction results, and prediction accuracy is improved through parameter correction or model retraining. Overall, the implementation typically includes steps such as multi-source data acquisition, data preprocessing and feature engineering, scenario identification and classification, model building and training, dynamic prediction calculation, and result feedback correction, forming a closed-loop data-driven energy consumption prediction process.

[0003] The existing technology has the following shortcomings: In the dynamic prediction of energy consumption in multiple scenarios for building-integrated photovoltaic (BIPV) curtain walls, net energy consumption is jointly composed of photovoltaic (PV) power generation and building load. When PV power generation and building load remain stable within the same time scale, but their contribution ratios to net energy consumption undergo structural substitution, while the total net energy consumption still shows a continuous trend, the net energy consumption curve remains flat on the surface, but its internal energy composition has undergone a shift in dominant relationships. Because PV power generation and building load cancel each other out at the numerical level, the net energy consumption amplitude does not change significantly, making it difficult to reflect internal structural changes using only the net energy consumption value as a criterion. Existing dynamic prediction technologies for multi-scenario energy consumption of BIPV curtain walls cannot adjust the multi-scenario judgment boundaries based on the degree of substitution when the contribution ratios of PV power generation and building load to net energy consumption undergo structural changes. This makes it impossible to identify operating states where the net energy consumption appears stable but its internal composition changes, resulting in a lag in multi-scenario segmentation. This leads to directional deviations in the prediction results during the energy dominance transition phase, reducing the accuracy of multi-scenario judgment and the reliability of subsequent energy consumption prediction.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls, specifically including the following steps: S1. Perform unified time scale alignment and ratio normalization on the photovoltaic power generation data and building load data of building photovoltaic curtain walls, decompose the net energy consumption into the contribution ratio of photovoltaic power generation and the contribution ratio of building load, and construct the evolution matrix of net energy consumption contribution ratio. S2. Based on the net energy consumption contribution ratio evolution matrix, extract the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio in the time series, calculate the proportional displacement consistency index, and determine whether the contribution ratios of photovoltaic power generation and building load to net energy consumption have undergone structural changes based on the proportional displacement consistency index. S3. In the case of structural changes, perform dominant substitution projection calculation on the proportional displacement vector to generate the substitution intensity index and dominant conversion direction factor to determine the degree of substitution. S4. Based on the constituent substitution intensity index and the dominant conversion direction factor, establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, and adjust the judgment boundary of multiple scenarios according to the constituent substitution degree. S5. Based on the adjusted multi-scenario judgment boundary, the scenario weight distribution and prediction parameter weight are updated synchronously to complete the dynamic prediction of energy consumption of building photovoltaic curtain walls in multiple scenarios.

[0007] Preferably, S1 is as follows: The photovoltaic power generation data and building load data of the building photovoltaic curtain wall are resampled according to a preset unified time scale, and a time index mapping relationship is established. The photovoltaic power generation data and building load data are mapped to the same time index sequence. The data points corresponding to the missing time index are interpolated according to the numerical change ratio between adjacent time indices to generate a photovoltaic power generation data sequence and a building load data sequence after unified time scale alignment. Based on the photovoltaic power generation data sequence and building load data sequence after unified time scale alignment, the net energy consumption sequence is calculated by subtracting the photovoltaic power generation data from the building load data. The absolute value of the net energy consumption is the absolute value of the net energy consumption sequence under the corresponding time index. The photovoltaic power generation data and building load data are divided by the absolute value of the net energy consumption under the corresponding time index to generate the photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence. The photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence are arranged in matrix order according to time index. The net energy consumption contribution ratio evolution matrix is ​​constructed with time index as row coordinate and photovoltaic power generation contribution ratio and building load contribution ratio as column coordinate.

[0008] Preferably, S2 specifically includes the following steps: S201. Based on the net energy consumption contribution ratio evolution matrix, read the photovoltaic power generation contribution ratio column vector and the building load contribution ratio column vector respectively, calculate the ratio difference between adjacent time indices according to the time index order, arrange the ratio differences corresponding to consecutive time indices in sequence, and form the photovoltaic power generation contribution ratio ratio displacement vector and the building load contribution ratio displacement vector. S202. Based on the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio, the proportional displacement direction consistency and proportional displacement amplitude deviation are combined at the corresponding time index position to generate the proportional displacement consistency index, wherein the proportional displacement consistency index is composed of the direction consistency parameter and the amplitude deviation parameter. S203. Compare the proportional displacement consistency index with the preset structural change judgment interval. When the proportional displacement consistency index falls into the preset structural change judgment interval, it is determined that the proportion of photovoltaic power generation and building load to net energy consumption has undergone structural change; otherwise, it is determined that no structural change has occurred.

[0009] Preferably, S202 specifically refers to: At the corresponding time index positions, the number of times the photovoltaic power generation contribution ratio and the number of times the building load contribution ratio change in the same direction and in opposite directions are counted respectively. The difference between the number of times ... The displacement value difference between the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio is calculated at the corresponding time index position. The absolute value of the displacement value difference under continuous time index is accumulated. Then, the accumulated result is divided by the total number of time indexes for mean normalization to generate the proportional displacement amplitude deviation. The proportional displacement amplitude deviation is proportionally mapped to the preset amplitude benchmark to construct the amplitude deviation parameter. The directional consistency parameter and the amplitude deviation parameter are linearly combined according to preset weights to generate the proportional displacement consistency index, which is used to represent the degree of coordinated change between the proportional displacement vector of photovoltaic power generation contribution and the proportional displacement vector of building load contribution in the time series.

[0010] Preferably, S3 specifically includes the following steps: S301. In the event of structural changes, obtain the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio respectively. Construct a proportional displacement combination vector at the same time index position. Use the symmetrical direction vector where the values ​​of the photovoltaic power generation contribution ratio displacement component and the building load contribution ratio displacement component are equal as the dominant substitution reference direction. Perform vector projection calculation on the proportional displacement combination vector to generate the dominant substitution projection value corresponding to each time index. S302. The absolute values ​​of the dominant substitution projection values ​​under the continuous time index are accumulated and normalized by dividing by the total number of time indices to generate the substitution intensity index. The number of times the dominant substitution projection value is positive and the number of times it is negative are counted. The difference between the positive and negative values ​​and the sum of the positive and negative values ​​are normalized to generate the dominant conversion direction factor. S303. The substitution intensity index and the dominant conversion direction factor are matched with the preset substitution classification intervals respectively. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is positive, the degree of substitution is determined to be power generation-dominated substitution. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is negative, the degree of substitution is determined to be load-dominated substitution. When the substitution intensity index does not fall into any intensity interval, the degree of substitution is determined to be a state where a stable substitution has not been formed.

[0011] Preferably, S301 is as follows: In the event of structural changes, the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio are read in time index order. At each time index position, the proportional displacement components of the photovoltaic power generation contribution ratio and the building load contribution ratio are extracted, and a two-dimensional proportional displacement combination vector is constructed using the two proportional displacement components. In the proportional displacement coordinate plane, a symmetrical direction vector with the same value and sign as the proportional displacement component contributed by photovoltaic power generation and the proportional displacement component contributed by building load is constructed as the dominant alternative reference direction. The projection length of the proportional displacement combination vector on the dominant alternative reference direction is calculated to generate the dominant alternative projection value at the corresponding time index position. The dominant substitution projection values ​​under continuous time index are arranged in time index order, and the dominant substitution projection values ​​are subjected to sign preservation processing to form a dominant substitution projection sequence containing directional information, which is used to subsequently generate the substitution intensity index and the dominant conversion direction factor.

[0012] Preferably, S4 is as follows: Based on the substitution intensity index and the dominant conversion direction factor, a proportional dimension feature vector is constructed, and the trend slope of the net energy consumption sequence is calculated in the corresponding time interval to generate a net energy consumption trend dimension vector. The proportional dimension feature vector and the net energy consumption trend dimension vector are mapped to the same coupled coordinate space to establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension. In the coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, the coordinate positions of the proportional dimension feature vector in the coupled coordinate space are extracted according to the degree of substitution. When the degree of substitution is power generation-led substitution, the projection offset of the proportional dimension feature vector along the negative axis of the net energy consumption trend is calculated; when the degree of substitution is load-led substitution, the projection offset of the proportional dimension feature vector along the positive axis of the net energy consumption trend is calculated; when the degree of substitution is that a stable substitution state has not been formed, it is determined that the proportional dimension feature vector does not produce a trend axis offset. The projection offset obtained from the corresponding degree of substitution is superimposed on the original multi-scene judgment boundary threshold, and the multi-scene judgment boundary is directionally shifted and adjusted so that the multi-scene judgment boundary forms an updated judgment interval in the coupling space of the proportional dimension and the net energy consumption trend dimension.

[0013] Preferably, S5 is as follows: Based on the adjusted multi-scenario judgment boundary, the proportional dimension feature vector and the net energy consumption trend dimension vector of the current time interval are matched in the coupling space of proportional dimension and net energy consumption trend dimension to determine the corresponding scenario category. Based on the interval span corresponding to the adjusted multi-scenario judgment boundary, the interval coverage ratio of each scenario is recalculated, the interval coverage ratio of each scenario is normalized, and the updated scenario weight distribution is generated. Based on the updated scenario weight distribution, the prediction parameter weights corresponding to each scenario are updated synchronously. The updated scenario weight distribution is used as a weight coefficient to reconstruct the prediction parameter weights, generating updated prediction parameter weights. The updated prediction parameter weights are then applied to the net energy consumption prediction calculation for subsequent time intervals.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention transforms the net energy consumption of building photovoltaic (PV) curtain walls from a single numerical expression into a dynamic proportional expression system of the contribution ratios of PV power generation and building load. It constructs a net energy consumption contribution ratio evolution matrix and extracts a proportional displacement vector in the time series dimension to form a proportional displacement consistency index. This enables precise identification of whether structural changes have occurred in the contribution ratios of PV power generation and building load to net energy consumption. While the total net energy consumption maintains a continuous changing trend, it can capture the transformation of the dominant relationship of internal energy composition, thus overcoming the structural change distortion problem caused by traditional methods that rely solely on net energy consumption values. By calculating the dominant substitution projection to generate a composition substitution intensity index and a dominant transformation direction factor, and mapping them to a coupled space of proportional and net energy consumption trend dimensions, the directional translation adjustment of multi-scenario judgment boundaries allows the scene division logic to evolve synchronously with changes in the internal proportional structure, thereby improving the sensitivity and accuracy of multi-scenario judgment.

[0015] 2. This invention, based on adjusting the multi-scenario judgment boundaries, synchronously updates the scenario weight distribution and prediction parameter weights, enabling changes in the proportional structure to be directly transmitted to the prediction level, avoiding directional deviations in prediction during the energy dominance relationship transition phase. By constructing a closed-loop dynamic update chain of proportional identification, substitution quantification, boundary adjustment, and weight reconstruction, it realizes the transformation from static judgment to structurally adaptive judgment in multi-scenario dynamic prediction of building photovoltaic curtain walls. This allows the prediction model to maintain judgment consistency and parameter matching even when the net energy consumption appearance is stable but the internal energy composition is substituted, improving the stability, reliability, and foresight of multi-scenario dynamic energy consumption prediction. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart illustrating the multi-scenario dynamic energy consumption prediction method for building photovoltaic curtain walls according to the present invention.

[0018] Figure 2This is a flowchart illustrating step S2 in the multi-scenario dynamic energy consumption prediction method for building photovoltaic curtain walls of the present invention.

[0019] Figure 3 This is a flowchart illustrating step S3 in the multi-scenario dynamic energy consumption prediction method for building photovoltaic curtain walls of the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figures 1 to 3 The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls, as shown, specifically includes the following steps: S1. Perform unified time scale alignment and ratio normalization on the photovoltaic power generation data and building load data of building photovoltaic curtain walls, decompose the net energy consumption into the contribution ratio of photovoltaic power generation and the contribution ratio of building load, and construct the evolution matrix of net energy consumption contribution ratio. In this embodiment, S1 specifically refers to: The photovoltaic power generation data and building load data of the building photovoltaic curtain wall are resampled according to a preset unified time scale, and a time index mapping relationship is established. The photovoltaic power generation data and building load data are mapped to the same time index sequence. The data points corresponding to the missing time index are interpolated according to the numerical change ratio between adjacent time indices to generate a photovoltaic power generation data sequence and a building load data sequence after unified time scale alignment. The resampling process for photovoltaic (PV) power generation data and building load data from building-integrated photovoltaic (BIPV) curtain walls, based on a pre-defined unified time scale, essentially transforms data with different original collection frequencies into a data sequence with the same time granularity. For example, PV power generation data might be recorded with shorter sampling intervals, while building load data might be recorded with longer sampling intervals. In this case, a unified time scale is first set, such as a fixed time interval as the target sampling interval, and then a continuous time index sequence is constructed based on this time interval. For the original PV power generation data and building load data, a time index mapping rule is used to map the original data to the target time index according to the correspondence between their respective timestamps and the target time index. When there is no original sampled value under a certain target time index, the numerical change ratio between adjacent time indices is calculated based on the data change trend of the two known time indices before and after it, and interpolation calculations are performed on the data points corresponding to the missing time index according to the proportional relationship. For example, if PV power generation data is missing at a certain time index, the power generation value of the current time index is estimated based on the power generation change amplitude of the two time points before and after it, according to the time interval ratio. This resampling and interpolation process ensures that the photovoltaic power generation data sequence and the building load data sequence are completely consistent in the time dimension, thus obtaining a photovoltaic power generation data sequence and a building load data sequence aligned with a unified time scale, providing basic data for subsequent net energy consumption calculation and proportional decomposition.

[0022] Among them, photovoltaic power generation data of building photovoltaic curtain walls refers to the power generation or electricity generation data output in real time by building photovoltaic curtain wall components, and building load data refers to the power or electricity consumption data of buildings within the same time range; preset unified time scale refers to the fixed time interval standard determined before data processing, used to unify data with different sampling frequencies; resampling processing refers to reconstructing the original data into a new time series according to the unified time scale; time index mapping relationship refers to the correspondence rules between the original timestamp and the target time index; same time index sequence refers to the time axis shared by photovoltaic power generation data and building load data; data point corresponding to missing time index refers to the data position where the original sampled value could not be directly matched at the target time index; the numerical change ratio between adjacent time indices refers to the correspondence between the data change amplitude and time interval between two consecutive valid time points; interpolation calculation refers to the process of estimating the value at the missing time index based on the change ratio; the photovoltaic power generation data sequence and building load data sequence after unified time scale alignment refer to two sets of data sequences that are completely consistent in time dimension after resampling and interpolation processing and can be directly used for subsequent net energy consumption calculation.

[0023] Based on the photovoltaic power generation data sequence and building load data sequence after unified time scale alignment, the net energy consumption sequence is calculated by subtracting the photovoltaic power generation data from the building load data. The absolute value of the net energy consumption is the absolute value of the net energy consumption sequence under the corresponding time index. The photovoltaic power generation data and building load data are divided by the absolute value of the net energy consumption under the corresponding time index to generate the photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence. Based on the photovoltaic (PV) power generation data sequence and building load data sequence aligned to a unified time scale, the net energy consumption sequence is first obtained by subtracting the PV power generation data from the building load data at each time index. This reflects the actual energy surplus or deficit of the building at the corresponding time index. If the building load data is greater than the PV power generation data at a certain time index, the net energy consumption is positive, indicating that the building has external power demand; if the PV power generation data is greater than the building load data, the net energy consumption is negative, indicating that there is surplus power generation. The absolute value of the net energy consumption is the absolute magnitude of the net energy consumption value at the corresponding time index, used to eliminate the influence of energy direction on the proportional calculation. Subsequently, at each time index, the PV power generation data and building load data are divided by the absolute value of the net energy consumption at the corresponding time index, respectively, to obtain the PV power generation contribution ratio sequence and the building load contribution ratio sequence. For example, if the building load is at a high level and the PV power generation is at a low level at a certain time index, the absolute value of the net energy consumption is larger, the PV power generation contribution ratio is smaller, and the building load contribution ratio is larger; when the two values ​​are close, the ratios will be similar. This process transforms the original energy values ​​into a contribution ratio relative to net energy consumption, providing a comparable basis for subsequent identification of changes in internal dominant relationships.

[0024] The net energy consumption sequence refers to the time series formed by subtracting photovoltaic (PV) power generation data from building load data at each time index, used to represent the energy balance state of the building at each time point. The absolute value of net energy consumption refers to the absolute magnitude of the net energy consumption sequence at the corresponding time index, used as a benchmark for proportional normalization. The PV power generation contribution ratio sequence refers to the proportional expression of PV power generation data relative to the absolute value of net energy consumption, used to reflect the relative contribution of PV power generation in the composition of net energy consumption. The building load contribution ratio sequence refers to the proportional expression of building load data relative to the absolute value of net energy consumption, used to reflect the relative proportion of building load in the composition of net energy consumption. By constructing the PV power generation contribution ratio sequence and the building load contribution ratio sequence, net energy consumption can be transformed from a single numerical expression into a proportional composition expression, allowing subsequent analysis to focus on changes in internal composition rather than solely relying on the magnitude of net energy consumption values.

[0025] The photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence are arranged in matrix order according to time index. The net energy consumption contribution ratio evolution matrix is ​​constructed with time index as row coordinate and photovoltaic power generation contribution ratio and building load contribution ratio as column coordinate.

[0026] Arranging the photovoltaic (PV) power generation contribution ratio sequence and the building load contribution ratio sequence in matrix order according to time index essentially transforms the two ratio sequences, originally arranged in a one-dimensional time sequence, into a two-dimensional data representation. Each time index corresponds to a row of data records, containing both the PV power generation contribution ratio and the building load contribution ratio. In practice, a continuous list of time indices is first generated according to a unified time scale. Then, the PV power generation contribution ratio sequence and the building load contribution ratio sequence are traversed in time index order. At each time index position, the corresponding two ratio values ​​are extracted and written into a two-dimensional array or data table structure in a fixed column order. For example, if the PV power generation contribution ratio is one value and the building load contribution ratio is another value at a certain time index, the corresponding row in the matrix is ​​recorded as a vector containing these two ratios. As the time index increases, the matrix is ​​filled row by row, thus forming a complete ratio evolution data structure. This arrangement transforms the ratio change process into a two-dimensional data form suitable for vector operations and trajectory analysis, providing direct input for subsequent ratio displacement vector calculations and evolution trend analysis.

[0027] Matrix arrangement refers to organizing multiple time series data into a two-dimensional data set according to fixed row and column rules. The time index, represented by row coordinates, indicates that each row of the matrix corresponds to a specific time point, reflecting the order of time evolution. The contribution ratios of photovoltaic power generation and building load, represented by column coordinates, indicate that each column of the matrix corresponds to a proportional component, reflecting the internal composition of net energy consumption. The net energy consumption contribution ratio evolution matrix is ​​a two-dimensional data set formed by sequentially arranging the proportional vectors under all time indices, used to express the overall trajectory of the changes in the contribution ratios of photovoltaic power generation and building load over time. By constructing the net energy consumption contribution ratio evolution matrix, the proportional changes can be expanded from single-point comparisons to continuous evolution analysis, providing a holistic basis for expressing changes in internal energy composition.

[0028] S2. Based on the net energy consumption contribution ratio evolution matrix, extract the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio in the time series, calculate the proportional displacement consistency index, and determine whether the contribution ratios of photovoltaic power generation and building load to net energy consumption have undergone structural changes based on the proportional displacement consistency index. In this embodiment, S2 specifically includes the following steps: S201. Based on the net energy consumption contribution ratio evolution matrix, read the photovoltaic power generation contribution ratio column vector and the building load contribution ratio column vector respectively, calculate the ratio difference between adjacent time indices according to the time index order, arrange the ratio differences corresponding to consecutive time indices in sequence, and form the photovoltaic power generation contribution ratio ratio displacement vector and the building load contribution ratio displacement vector. Extracting the proportional displacement vector based on the net energy consumption contribution ratio evolution matrix essentially transforms the change of the ratio over time from a "static proportional value" into a "dynamic trajectory." In practice, firstly, the column vectors of photovoltaic power generation contribution ratio and building load contribution ratio are read from the net energy consumption contribution ratio evolution matrix, ensuring each column vector is arranged in time index order. Then, the sequence of time indexes is traversed sequentially, calculating the difference between the proportional value of the later time index and the proportional value of the previous time index. This difference is used as the change in the ratio within that time interval. Next, the proportional differences between all adjacent time indices are arranged sequentially in ascending order of time indices, forming the proportional displacement vectors of both the photovoltaic power generation contribution ratio and the building load contribution ratio. For example, if the photovoltaic power generation contribution ratio changes from a lower value to a higher value at a certain time index, the corresponding proportional difference for that time period is positive; if it changes from a higher value to a lower value, the proportional difference is negative. Through this continuous difference calculation, the proportional evolution matrix can be transformed into a proportional change trajectory expression, thus providing basic data for subsequent directional consistency and amplitude deviation analysis.

[0029] The column vectors for the contribution ratios of photovoltaic (PV) power generation and building load are two sets of proportional data extracted from the net energy consumption contribution ratio evolution matrix. They represent the numerical arrangement of the PV power generation contribution ratio and building load contribution ratio at each time index. The ratio difference between adjacent time indices refers to the change in the ratio value at two consecutive time index positions, reflecting the increasing or decreasing trend of the ratio within a unit time interval. The proportional displacement vectors of the PV power generation contribution ratio and the building load contribution ratio are vector data sets formed by arranging the proportional differences of all adjacent time indices in chronological order, representing the displacement trajectory of the ratio throughout the entire time series. By constructing proportional displacement vectors, the changes in the proportion composition can be extended from single-point comparisons to continuous change analysis, giving the internal energy composition evolution quantifiable temporal dynamic characteristics.

[0030] S202. Based on the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio, the proportional displacement direction consistency and proportional displacement amplitude deviation are combined at the corresponding time index position to generate the proportional displacement consistency index, wherein the proportional displacement consistency index is composed of the direction consistency parameter and the amplitude deviation parameter. S203. Compare the proportional displacement consistency index with the preset structural change judgment interval. When the proportional displacement consistency index falls into the preset structural change judgment interval, it is determined that the proportion of photovoltaic power generation and building load to net energy consumption has undergone structural change; otherwise, it is determined that no structural change has occurred.

[0031] Comparing the proportional displacement consistency index with a preset structural change judgment interval essentially transforms the quantified comprehensive change index into a clear structural change judgment result. In practice, firstly, the distribution range of the proportional displacement consistency index during stable operation and dominant transition phases is statistically analyzed based on historical operating data. Then, a preset structural change judgment interval is set based on statistical quantile intervals or fixed interval boundaries. This interval can be defined as the range where the index is above a certain upper boundary or below a certain lower boundary. Subsequently, during actual operation, the real-time calculated proportional displacement consistency index is compared with the preset structural change judgment interval. When the proportional displacement consistency index falls within the preset structural change judgment interval, a structural change is determined to have occurred in the proportion of photovoltaic power generation and building load's contribution to net energy consumption; otherwise, no structural change is determined. For example, in historical samples, when the internal dominant relationship changes, the proportional displacement consistency index is concentrated in a negative interval. This interval is then set as the structural change judgment interval, and a structural change is determined when the real-time index enters this interval. By using interval judgment rather than single-point threshold comparison, the ability to stably identify structural changes can be improved.

[0032] The preset structural change judgment interval refers to the range of index values ​​predetermined based on the historical distribution pattern of the proportional displacement consistency index, used to distinguish between stable proportional evolution states and structural substitution states. A structural change in the contribution ratio of photovoltaic power generation and building load to net energy consumption refers to a state where the contribution ratios of photovoltaic power generation and building load show a continuous directional consistency or continuous reverse deviation in the time series, thus causing a shift in the internal dominant relationship. The proportional displacement consistency index, as a comprehensive indicator, is used to characterize the degree of coordination of the proportional displacement vector in terms of direction and amplitude. Interval comparison refers to matching the proportional displacement consistency index with the preset structural change judgment interval in terms of numerical range, thereby outputting the structural change judgment result. By introducing the preset structural change judgment interval, continuous change indicators can be transformed into clear structural change judgment criteria, providing triggering conditions for subsequent multi-scenario judgment boundary adjustments.

[0033] In this embodiment, S202 specifically refers to: At the corresponding time index positions, the number of times the photovoltaic power generation contribution ratio and the number of times the building load contribution ratio change in the same direction and in opposite directions are counted respectively. The difference between the number of times ... The process of counting the number of same-direction and opposite-direction changes in the proportional displacement vectors of photovoltaic power generation contribution and building load contribution at corresponding time index positions essentially involves comparing the direction of change of these two proportional displacement vectors point-by-point throughout the time series. Specifically, the displacement values ​​of both vectors at the same time index position are read sequentially according to the time index. The direction of change is determined by the sign of the displacement values: a same-direction change is counted when the two values ​​have the same sign, and an opposite-direction change is counted when the signs are opposite. After traversing the entire time series, the number of same-direction and opposite-direction changes are obtained. Then, the difference between the number of same-direction and opposite-direction changes is divided by the sum of the number of same-direction and opposite-direction changes for normalization, limiting the result to between -1 and +1, thus generating a direction consistency parameter. For example, when most time index positions show same-direction changes, the direction consistency parameter is close to +1; when most time index positions show opposite-direction changes, the direction consistency parameter is close to -1; and when the number of same-direction and opposite-direction changes is close to zero, the direction consistency parameter is close to zero. This normalization calculation transforms the original directional statistical results into standardized indicators, preserving directional consistency information while eliminating the influence of time series length on the calculation results, thus providing a stable basis for directional determination in the subsequent construction of the proportional displacement consistency index.

[0034] The displacement value difference between the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio is calculated at the corresponding time index position. The absolute value of the displacement value difference under continuous time index is accumulated. Then, the accumulated result is divided by the total number of time indexes for mean normalization to generate the proportional displacement amplitude deviation. The proportional displacement amplitude deviation is proportionally mapped to the preset amplitude benchmark to construct the amplitude deviation parameter. Calculating the displacement difference between the proportional displacement vector of photovoltaic power generation contribution and the proportional displacement vector of building load contribution at the corresponding time index position essentially quantifies the difference in the variation amplitude of the two proportional displacement vectors at the same time index. In practice, firstly, the displacement values ​​of the two proportional displacement vectors at the same time index position are read sequentially according to the time index, and the difference between them is taken as the displacement value difference at that time index. Then, the absolute values ​​of all displacement value differences obtained under consecutive time indices are accumulated to eliminate the influence of directional factors on amplitude statistics, ensuring that the amplitude deviation at each time index participates in the accumulation as a positive value. After completing the absolute value accumulation, the accumulated result is divided by the total number of time indices and averaged to obtain the proportional displacement amplitude deviation. For example, when the variation amplitude of photovoltaic power generation contribution is significantly greater than that of building load contribution within a certain time interval, the displacement value difference at the corresponding time index is large. After absolute value accumulation and averaged, a high proportional displacement amplitude deviation will be formed. Subsequently, the proportional displacement amplitude deviation is proportionally mapped to a preset amplitude benchmark, which can be determined based on the historical displacement differential distribution. The current proportional displacement amplitude deviation is then mapped to a unified scale range to construct the amplitude deviation parameter.

[0035] The displacement numerical difference refers to the numerical difference between the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio at the same time index position, used to reflect the difference in amplitude variation; the absolute value accumulation processing refers to taking the absolute value of the displacement numerical difference at all time indices and then continuously accumulating it, used to eliminate the influence of direction and strengthen the overall statistics of amplitude difference; the mean normalization processing refers to dividing the result of the absolute value accumulation by the total number of time indices, so that the proportional displacement amplitude deviation is not affected by the time length; the proportional displacement amplitude deviation is a centralized expression of the amplitude difference within the entire time series; the preset amplitude benchmark is a reference value used to calibrate the scale of amplitude difference; the proportional mapping refers to the scaling transformation of the proportional displacement amplitude deviation according to the preset amplitude benchmark; the amplitude deviation parameter is a standardized amplitude index after mapping, used to participate in the construction of the proportional displacement consistency index.

[0036] The directional consistency parameter and the amplitude deviation parameter are linearly combined according to preset weights to generate the proportional displacement consistency index, which is used to represent the degree of coordinated change between the proportional displacement vector of photovoltaic power generation contribution and the proportional displacement vector of building load contribution in the time series.

[0037] The linear combination of the directional consistency parameter and the amplitude deviation parameter with preset weights is a process of comprehensively expressing directional consistency information and amplitude deviation information on the same scale. In practice, firstly, preset weights are set according to the importance of the directional consistency parameter and the amplitude deviation parameter in the overall judgment logic. These preset weights can be determined based on historical sample distribution or empirical proportions, and the sum of the weights must be a fixed constant. Then, within the same time interval, the directional consistency parameter and the amplitude deviation parameter are multiplied by their respective preset weights and then weighted and summed to obtain the proportional displacement consistency index. For example, when the directional consistency parameter is close to positive one and the amplitude deviation parameter is small, with a high proportion of directional weight, the weighted result will tend towards a higher positive value, indicating a strong coordinated change in proportional displacement. When the directional consistency parameter is close to negative one and the amplitude deviation parameter is large, with a high proportion of amplitude weight, the weighted result will tend towards a negative range, indicating a significant reverse deviation in proportional displacement. Through this linear combination operation, while maintaining the independent calculation of directional and amplitude features, the two can be integrated into a single index, avoiding the bias of a single parameter on the judgment result.

[0038] The preset weights are coefficients used to adjust the proportion of the directional consistency parameter and the amplitude deviation parameter in the comprehensive calculation; their values ​​can be set according to the historical data distribution. Linear combination operation refers to the weighted summation of multiple parameters, allowing information from multiple dimensions to be expressed on a unified scale. The proportional displacement consistency index is a comprehensive index generated by fusing the directional consistency parameter and the amplitude deviation parameter; it is used to characterize the degree of coordinated change between the proportional displacement vector of photovoltaic power generation contribution and the proportional displacement vector of building load contribution in the time series. By constructing the proportional displacement consistency index, the direction and amplitude differences of proportional changes can be unified into a single judgment index, providing a stable quantitative basis for subsequent structural change judgments.

[0039] S3. In the case of structural changes, perform dominant substitution projection calculation on the proportional displacement vector to generate the substitution intensity index and dominant conversion direction factor to determine the degree of substitution. In this embodiment, S3 specifically includes the following steps: S301. In the event of structural changes, obtain the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio respectively. Construct a proportional displacement combination vector at the same time index position. Use the symmetrical direction vector where the values ​​of the photovoltaic power generation contribution ratio displacement component and the building load contribution ratio displacement component are equal as the dominant substitution reference direction. Perform vector projection calculation on the proportional displacement combination vector to generate the dominant substitution projection value corresponding to each time index. S302. The absolute values ​​of the dominant substitution projection values ​​under the continuous time index are accumulated and normalized by dividing by the total number of time indices to generate the substitution intensity index. The number of times the dominant substitution projection value is positive and the number of times it is negative are counted. The difference between the positive and negative values ​​and the sum of the positive and negative values ​​are normalized to generate the dominant conversion direction factor. The absolute values ​​of the dominant substitution projection values ​​under continuous time indices are accumulated and normalized by dividing by the total number of time indices. This is to quantify the intensity level of the proportional dominant substitution from an overall time perspective. Specifically, the dominant substitution projection sequence is first traversed according to the time index order. The absolute value of each dominant substitution projection value is taken and accumulated item by item to eliminate the influence of directional factors on the intensity statistics, ensuring that the projection amplitude at each time index position is positive in the calculation. After accumulating the absolute values ​​of all time indices, the accumulated result is divided by the total number of time indices for normalization, thus generating the substitution intensity index. For example, in a certain structural change phase, if the amplitudes of the dominant substitution projection values ​​at multiple time index positions are large, the accumulated absolute value result will be high. Even after normalization to the total number of time indices, it will still remain at a high level, indicating a high substitution intensity. Simultaneously, the number of times the dominant substitution projection value is positive and the number of times it is negative are counted. The difference between positive and negative counts and the sum of positive and negative counts are normalized to generate the dominant conversion direction factor. When the number of positive values ​​significantly exceeds the number of negative values, the dominant conversion direction factor tends towards the positive range; when the number of negative values ​​predominates, the dominant conversion direction factor tends towards the negative range. By calculating intensity statistics and direction statistics separately, quantitative results in both the magnitude and direction of substitution can be obtained simultaneously.

[0040] Normalization involves dividing the cumulative result by the total number of time indices to ensure that the substitution intensity index is unaffected by time length. The substitution intensity index is a centralized expression of the overall level of the dominant substitution projection amplitude under continuous time indices, used to characterize the substitution intensity during structural change phases. Normalization calculation involves proportionalizing the difference between positive and negative frequencies to the total frequencies, limiting the dominant conversion direction factor to a uniform numerical range. The dominant conversion direction factor is a quantitative indicator of the directional distribution in the dominant substitution projection sequence, used to represent the dominant direction of the substitution trend. The synergistic generation of the substitution intensity index and the dominant conversion direction factor provides a two-dimensional basis for subsequent classification of the degree of substitution.

[0041] S303. The substitution intensity index and the dominant conversion direction factor are matched with the preset substitution classification intervals respectively. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is positive, the degree of substitution is determined to be power generation-dominated substitution. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is negative, the degree of substitution is determined to be load-dominated substitution. When the substitution intensity index does not fall into any intensity interval, the degree of substitution is determined to be a state where a stable substitution has not been formed.

[0042] Matching the substitution intensity index and the dominant switching direction factor with preset substitution grading intervals is a process of transforming continuous numerical indicators into discrete substitution degree determination results. In practice, firstly, based on the statistical distribution of the substitution intensity index in historical structural change samples, multiple intensity intervals are pre-divided, and clear numerical boundaries are set for each interval. Then, during operation, the real-time calculated substitution intensity index is matched with each intensity interval one by one. When the substitution intensity index falls into a corresponding intensity interval, a joint determination is made based on the sign of the dominant switching direction factor. If the substitution intensity index falls into the corresponding intensity interval and the dominant switching direction factor is positive, the substitution degree is determined to be power generation-dominated substitution; if the substitution intensity index falls into the corresponding intensity interval and the dominant switching direction factor is negative, the substitution degree is determined to be load-dominated substitution; if the substitution intensity index does not fall into any intensity interval, the substitution degree is determined to be a state where stable substitution has not yet been formed. For example, when the substitution intensity index reaches a high intensity range and the dominant conversion direction factor is positive, it indicates that the contribution of photovoltaic power generation plays a dominant role in the structural substitution process; when the index is low or does not enter any intensity range, it indicates that the substitution has not yet formed a stable structure. By combining interval matching and direction determination, continuous indicators can be transformed into clear substitution classification results.

[0043] The preset substitution grading intervals are multiple intervals pre-divided based on the numerical range of the substitution intensity index, used to distinguish different intensity levels. A substitution intensity index falling into a corresponding intensity interval means the index value is between the upper and lower boundaries of a preset interval. Power generation-dominated substitution indicates that the contribution ratio of photovoltaic power generation is dominant during the substitution process in the structural change phase; load-dominated substitution indicates that the contribution ratio of building load is dominant during the substitution process; a substitution intensity index not falling into any intensity interval means the index value has not reached any preset intensity boundary requirement; and the absence of a stable substitution state indicates that the proportional substitution has not yet formed a sustained dominant relationship. By constructing preset substitution grading intervals and combining them with a dominant conversion direction factor for two-dimensional determination, the degree of substitution can have clear grading rules and directional distinction standards, providing a deterministic classification basis for subsequent multi-scenario determination boundary adjustments.

[0044] In this embodiment, S301 specifically refers to: In the event of structural changes, the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio are read in time index order. At each time index position, the proportional displacement components of the photovoltaic power generation contribution ratio and the building load contribution ratio are extracted, and a two-dimensional proportional displacement combination vector is constructed using the two proportional displacement components. In the event of structural changes, reading the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio in time index order essentially involves synchronously pairing two independent one-dimensional displacement sequences within the same time coordinate system. Specifically, the proportional displacement vectors are first traversed sequentially according to the time index sequence. At each time index position, the corresponding photovoltaic power generation contribution ratio displacement component and building load contribution ratio displacement component are extracted. These two components are then used as two coordinate components at the same time point to construct a two-dimensional proportional displacement combination vector, where the horizontal component corresponds to the photovoltaic power generation contribution ratio displacement component, and the vertical component corresponds to the building load contribution ratio displacement component. For example, if the photovoltaic power generation contribution ratio displacement component is positive and the building load contribution ratio displacement component is negative at a certain time index, the two-dimensional proportional displacement combination vector is located in a specific quadrant in the plane coordinate system; if both are positive, it is located in another quadrant. By constructing the two-dimensional proportional displacement combination vector, the two originally separate proportional displacement change trajectories can be integrated into a vector expression with directional and amplitude characteristics. This transforms the proportional change relationship from a separate difference into a planar vector motion expression, providing a unified data foundation for subsequent dominant alternative reference direction construction and vector projection calculation.

[0045] The photovoltaic (PV) power generation contribution ratio displacement component refers to the change in the PV power generation contribution ratio at a certain time index relative to the previous time index. The building load contribution ratio displacement component refers to the change in the building load contribution ratio at the same time index relative to the previous time index. The two-dimensional proportional displacement combination vector is a vector data structure formed by combining these two proportional displacement components in a fixed coordinate order. It is used to express the joint change state of the two components in the proportional displacement coordinate plane. By constructing the two-dimensional proportional displacement combination vector, the directional and magnitude relationships of proportional changes can be simultaneously incorporated into a unified calculation framework, thus establishing a clear geometric expression basis for dominant alternative projection calculations.

[0046] In the proportional displacement coordinate plane, a symmetrical direction vector with the same value and sign as the proportional displacement component contributed by photovoltaic power generation and the proportional displacement component contributed by building load is constructed as the dominant alternative reference direction. The projection length of the proportional displacement combination vector on the dominant alternative reference direction is calculated to generate the dominant alternative projection value at the corresponding time index position. Constructing a symmetrical direction vector in the proportional displacement coordinate plane, where the proportional displacement components contributed by photovoltaic power generation and those contributed by building load are equal in value and have the same sign, is to establish a standard reference direction for measuring the cooperative substitution relationship between the two types of proportional displacements. Specifically, a direction vector is defined in the two-dimensional proportional displacement coordinate plane, where the horizontal and vertical components have the same value and maintain the same sign, thus forming a symmetrical direction vector extending diagonally. When both proportional displacement components increase or decrease simultaneously, the two-dimensional proportional displacement combined vector will tend towards this symmetrical direction. Subsequently, at each time index position, the projection length of the proportional displacement combined vector on the dominant substitution reference direction is calculated. Specifically, the projection length is determined by calculating the degree of directional consistency between the proportional displacement combined vector and the dominant substitution reference direction, combined with the magnitude of the proportional displacement combined vector. The obtained projection length serves as the dominant substitution projection value for the corresponding time index position. For example, when the proportional displacement component of photovoltaic power generation and the proportional displacement component of building load are close in value and have the same sign at the same time index, the projection length of the proportional displacement combination vector in the symmetrical direction is larger, indicating that the two exhibit a strong synergistic substitution trend. When the two components are in opposite directions, the projection length in the symmetrical direction is smaller or even negative, indicating that the substitution direction has deviated. The symmetrical direction vector with equal value and the same sign is used to define the dominant substitution reference direction, which serves as the reference axis for projection calculation. The projection length in the dominant substitution reference direction is used to quantify the magnitude of the proportional displacement combination vector component on the reference axis. The dominant substitution projection value at the corresponding time index position is an expression of the intensity and direction of the substitution trend at a single time point. Through this calculation process, the two-dimensional proportional displacement relationship can be transformed into a quantitative result in a single direction, providing a continuous projection data basis for the subsequent generation of the substitution intensity index and the dominant conversion direction factor.

[0047] The dominant substitution projection values ​​under continuous time index are arranged in time index order, and the dominant substitution projection values ​​are subjected to sign preservation processing to form a dominant substitution projection sequence containing directional information, which is used to subsequently generate the substitution intensity index and the dominant conversion direction factor.

[0048] Arranging the dominant substitution projection values ​​under continuous time indices in time index order is to maintain the continuous expression of the dominant substitution trend in the time dimension. Specifically, firstly, based on a unified time index sequence, the dominant substitution projection values ​​calculated at each time index position are sequentially stored as a one-dimensional numerical sequence, ensuring that each element in the sequence corresponds to a unique time index. Then, sign-preserving processing is performed on each dominant substitution projection value; that is, while retaining the absolute magnitude of the value, its positive or negative sign information is preserved without eliminating or transforming the sign, thus forming a dominant substitution projection sequence containing directional information. For example, if the dominant substitution projection value is continuously positive within a certain time period, it indicates that the proportional displacement combination vector extends in the same direction as the dominant substitution reference direction; if it is continuously negative in another time period, it indicates that the dominant substitution direction has reversed. Sign-preserving processing means that the absolute value or uniformization of positive and negative signs is not performed during data processing, so that the projection values ​​contain both magnitude and directional information; the dominant substitution projection sequence is a set of one-dimensional projection values ​​arranged in time order and retaining sign characteristics. By constructing a dominant substitution projection sequence containing directional information, the magnitude of the projection values ​​can be statistically analyzed and the symbol distribution can be statistically analyzed in subsequent processing, thus providing a continuous and complete data foundation for the generation of the substitution intensity index and the dominant conversion direction factor.

[0049] S4. Based on the constituent substitution intensity index and the dominant conversion direction factor, establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, and adjust the judgment boundary of multiple scenarios according to the constituent substitution degree. In this embodiment, S4 specifically refers to: Based on the substitution intensity index and the dominant conversion direction factor, a proportional dimension feature vector is constructed, and the trend slope of the net energy consumption sequence is calculated in the corresponding time interval to generate a net energy consumption trend dimension vector. The proportional dimension feature vector and the net energy consumption trend dimension vector are mapped to the same coupled coordinate space to establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension. Constructing a proportional dimension feature vector based on the substitution intensity index and the dominant conversion direction factor is a process of unifying the expression of substitution intensity and substitution direction information during structural change stages. Specifically, the substitution intensity index can be used as the amplitude component, and the dominant conversion direction factor as the direction indicator component, combined to form a two-dimensional proportional dimension feature vector. The amplitude represents the substitution intensity level, and the direction component represents whether power generation or load is dominant. Simultaneously, the trend slope of the net energy consumption sequence is calculated within the same time interval. This can be achieved by linear fitting within a time window or continuous difference accumulation calculation of the net energy consumption sequence, yielding the overall trend of net energy consumption over time, thus generating a net energy consumption trend dimension vector. For example, if the substitution intensity index is high and the dominant conversion direction factor is positive within a certain time interval, while the net energy consumption sequence shows a downward trend, then the proportional dimension feature vector and the net energy consumption trend dimension vector form a specific angular relationship in coordinate space. Subsequently, the proportional dimension feature vector and the net energy consumption trend dimension vector are mapped to the same coupled coordinate space, so that the two types of vectors can be expressed in position under a unified coordinate system. This establishes a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, and allows the changes in internal composition and the external energy consumption trend to be associated within the same analytical framework.

[0050] The proportional dimension feature vector is a vector expression formed by combining the substitution intensity index and the dominant conversion direction factor, used to characterize the amplitude and direction of proportional structure changes; the net energy consumption sequence is a set of building net energy change data arranged according to time index; trend slope calculation refers to the quantification of the overall change direction and rate of change of the net energy consumption sequence within a selected time interval; the net energy consumption trend dimension vector is a vector expression representing the change direction and intensity of net energy consumption based on the trend slope result; the same coupled coordinate space refers to the space in which the proportional dimension feature vector and the net energy consumption trend dimension vector are uniformly mapped to the same two-dimensional or multi-dimensional coordinate system for coordinate positioning; the coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension refers to the mapping structure in this coordinate space that reflects the degree of correlation between internal proportional changes and external net energy consumption trends through vector position relationships. By establishing this coupled mapping relationship, quantitative coordinate basis can be provided for subsequent directional adjustments of multi-scenario judgment boundaries.

[0051] In the coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, the coordinate positions of the proportional dimension feature vector in the coupled coordinate space are extracted according to the degree of substitution. When the degree of substitution is power generation-led substitution, the projection offset of the proportional dimension feature vector along the negative axis of the net energy consumption trend is calculated; when the degree of substitution is load-led substitution, the projection offset of the proportional dimension feature vector along the positive axis of the net energy consumption trend is calculated; when the degree of substitution is that a stable substitution state has not been formed, it is determined that the proportional dimension feature vector does not produce a trend axis offset. In the coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, extracting the coordinate positions of the proportional dimension feature vectors in the coupled coordinate space according to the degree of substitution is a process of directionally associating the structural substitution results with the direction of the net energy consumption trend. Specifically, firstly, a trend axis is established in the coupled coordinate space along the direction of the net energy consumption trend dimension vector, distinguishing between the positive and negative trend axes. Then, the dominant substitution direction is determined based on the degree of substitution. When the degree of substitution is power generation-dominated, the proportional dimension feature vector is projected onto the negative trend axis of the net energy consumption trend, and the projection length of the proportional dimension feature vector on this negative trend axis is calculated, obtaining the projection offset of the proportional dimension feature vector along the negative trend axis of the net energy consumption trend. When the degree of substitution is load-dominated, the proportional dimension feature vector is projected onto the positive trend axis of the net energy consumption trend, and the projection offset of the proportional dimension feature vector along the positive trend axis of the net energy consumption trend is calculated. When the degree of substitution is that a stable substitution state has not yet been formed, no projection calculation is performed on the proportional dimension feature vector along the trend axis, keeping the original coordinate positions unchanged. For example, if, within a certain time interval, the substitution is dominated by power generation and the net energy consumption trend is downward, then the projection offset of the proportional dimension feature vector on the negative trend axis will be used as the basis for boundary adjustment; if the substitution is dominated by load, then the projection offset will be calculated along the positive trend axis. By distinguishing between different substitution types and trend directions, direction-sensitive boundary adjustment control can be achieved.

[0052] The projection offset of the proportional dimension feature vector along the negative axis of the net energy consumption trend refers to the projection length of the proportional dimension feature vector in the coupled coordinate space along the opposite direction of the net energy consumption trend vector, used to represent the degree of offset in the negative trend direction under the condition of power generation-dominated substitution. The projection offset of the proportional dimension feature vector along the positive axis of the net energy consumption trend refers to the projection length of the proportional dimension feature vector in the same direction as the net energy consumption trend vector, used to represent the degree of offset in the positive trend direction under the condition of load-dominated substitution. The proportional dimension feature vector does not generate a trend axis offset, meaning that when a stable substitution state has not been formed, no trend axis direction projection calculation is performed, and the original position in the coupled coordinate space is maintained. This processing, which distinguishes the projection direction according to the substitution type, provides a clear directional basis for adjusting the boundary judgment in multiple scenarios.

[0053] The projection offset obtained from the corresponding degree of substitution is superimposed on the original multi-scene judgment boundary threshold, and the multi-scene judgment boundary is directionally shifted and adjusted so that the multi-scene judgment boundary forms an updated judgment interval in the coupling space of the proportional dimension and the net energy consumption trend dimension.

[0054] Superimposing the projection offset obtained from the degree of substitution onto the original multi-scenario decision boundary threshold is a process of directly transforming structural substitution information into changes in the position of the decision boundary. Specifically, the coordinate values ​​of the original multi-scenario decision boundary threshold in the coupled space of the proportional dimension and the net energy consumption trend dimension are first read. This threshold can be represented as the boundary line position dividing different operating scenarios in the coupled coordinate space. Then, based on the projection offset determined by the degree of substitution, this projection offset is superimposed onto the coordinates corresponding to the original multi-scenario decision boundary threshold according to its direction attribute. When the projection offset is positive, the boundary moves along the positive axis of the net energy consumption trend; when the projection offset is negative, the boundary moves along the negative axis of the net energy consumption trend. For example, if a negative projection offset is obtained under the condition of power generation-dominated substitution, this offset is superimposed onto the original multi-scenario decision boundary threshold, causing the decision boundary to shift in the negative direction of the net energy consumption trend in the coupled coordinate space, thereby changing the subsequent scenario division position. By directly applying the projection offset to the boundary threshold, the multi-scenario decision logic and the proportional structural changes can be linked, rather than being fixed to the original threshold.

[0055] The original multi-scenario decision boundary threshold refers to the numerical boundary coordinates used to distinguish different operating scenarios in the coupled space of the proportional dimension and the net energy consumption trend dimension. Directional translation adjustment refers to the numerical movement of the decision boundary threshold along a specific coordinate axis based on the sign and magnitude of the projection offset, without changing the boundary morphology. The updated decision interval refers to the scene boundary range redefined in the coupled space after directional translation adjustment. Through this directional translation adjustment based on the degree of substitution, the multi-scenario decision boundaries can move synchronously when the proportional structure is substituted, thereby maintaining the dynamic consistency between scene division and changes in internal energy structure.

[0056] S5. Based on the adjusted multi-scenario judgment boundary, the scenario weight distribution and prediction parameter weight are updated synchronously to complete the dynamic prediction of energy consumption of building photovoltaic curtain walls in multiple scenarios.

[0057] In this embodiment, S5 specifically refers to: Based on the adjusted multi-scenario judgment boundary, the proportional dimension feature vector and the net energy consumption trend dimension vector of the current time interval are matched in the coupling space of proportional dimension and net energy consumption trend dimension to determine the corresponding scenario category. Based on the interval span corresponding to the adjusted multi-scenario judgment boundary, the interval coverage ratio of each scenario is recalculated, the interval coverage ratio of each scenario is normalized, and the updated scenario weight distribution is generated. Based on the adjusted multi-scenario judgment boundaries, interval matching is performed on the proportional dimension feature vector and net energy consumption trend dimension vector of the current time interval in the coupled space of proportional dimension and net energy consumption trend dimension. This process remaps the real-time running state to the updated scene classification system. Specifically, firstly, the range of each scene interval formed by the adjusted multi-scenario judgment boundaries is read in the coupled space. Then, the proportional dimension feature vector and net energy consumption trend dimension vector constructed within the current time interval are mapped to coordinate positions in the coupled space. The corresponding scene category is determined by determining which scene interval the coordinate point falls into. Subsequently, based on the span of each scene interval defined by the adjusted multi-scenario judgment boundaries, the coverage ratio of each scene interval in the overall coupled space is calculated. Combined with the distribution density of the proportional dimension feature vector and net energy consumption trend dimension vector within the current time interval, the interval coverage ratio of each scene is recalculated. Finally, the interval coverage ratios of each scene are uniformly scaled to keep the sum of all scene weights fixed, thereby generating an updated scene weight distribution. For example, when the feature vectors of the proportional dimension fall into the power generation-dominated replacement scenario range within a certain time interval, the coverage ratio of that scenario range increases, and the corresponding scenario weight increases after normalization.

[0058] The coupled space of proportional dimension and net energy consumption trend dimension refers to the two-dimensional coordinate space after uniformly mapping the proportional dimension feature vector and the net energy consumption trend dimension vector. The proportional dimension feature vector and the net energy consumption trend dimension vector of the current time interval represent the coordinate expressions of the current operating state in the two dimensions of proportional structure change and net energy consumption trend change. The corresponding scenario category refers to the type of operating scenario determined according to the interval division of the coupled space. The interval coverage ratio of each scenario refers to the proportion of the coordinate range occupied by each scenario interval in the coupled space. The updated scenario weight distribution refers to the set of scenario weights after being reallocated according to the interval coverage ratio. Through this weight reconstruction based on the interval coverage ratio of the coupled space, the scenario weight distribution can be made to change synchronously with the adjustment of the judgment boundary.

[0059] Based on the updated scenario weight distribution, the prediction parameter weights corresponding to each scenario are updated synchronously. The updated scenario weight distribution is used as a weight coefficient to reconstruct the prediction parameter weights, generating updated prediction parameter weights. The updated prediction parameter weights are then applied to the net energy consumption prediction calculation for subsequent time intervals.

[0060] Based on the updated scene weight distribution, the prediction parameter weights corresponding to each scene are updated synchronously. This process directly transmits the scene-level changes to the prediction model parameter level. Specifically, firstly, an independent set of prediction parameter weights is pre-configured for each scene. This set can include the weight proportions of different input features in the prediction calculation. After obtaining the updated scene weight distribution, it is used as a weight coefficient matrix to reconstruct the prediction parameter weights corresponding to each scene. That is, according to the weight proportions of each scene in the scene weight distribution, the prediction parameter weights for the corresponding scenes are proportionally amplified or reduced, and then linearly combined when multiple scene weights are superimposed to generate the updated prediction parameter weights. For example, when the weight of the power generation-dominated substitution scene increases in the updated scene weight distribution, a higher weighting ratio is given to the prediction parameter weights corresponding to that scene, causing the prediction model to focus more on the impact of power generation-related feature variables in subsequent time intervals. Subsequently, the updated prediction parameter weights are applied to the net energy consumption prediction calculation in subsequent time intervals, ensuring that the prediction results reflect the weight shifts brought about by changes in scene structure.

[0061] The prediction parameter weights for each scenario refer to a pre-defined set of parameter weights for different operating scenarios, used to control the influence of each input variable in the prediction calculation. The weight coefficients are proportional coefficients derived from the updated scenario weight distribution and are used in the reconstruction calculation of the prediction parameter weights. Weighted reconstruction processing refers to proportionally adjusting the original prediction parameter weights according to the weight coefficients to form a new combination of parameter weights. The updated prediction parameter weights refer to the set of parameters formed after the scenario weights have been redistributed, used for subsequent net energy consumption prediction calculations. By synchronously updating the scenario weight distribution and the prediction parameter weights, dynamic linkage between multi-scenario decision boundary adjustments and the internal parameters of the prediction model can be achieved.

[0062] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0063] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0064] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0066] The units described as separate components may or may not be physically separate. 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0067] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0068] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamic prediction of energy consumption in multiple scenarios for building-integrated photovoltaic (BIPV) curtain walls, characterized in that, Specifically, the following steps are included: S1. Perform unified time scale alignment and ratio normalization on the photovoltaic power generation data and building load data of building photovoltaic curtain walls, decompose the net energy consumption into the contribution ratio of photovoltaic power generation and the contribution ratio of building load, and construct the evolution matrix of net energy consumption contribution ratio. S2. Based on the net energy consumption contribution ratio evolution matrix, extract the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio in the time series, calculate the proportional displacement consistency index, and determine whether the contribution ratios of photovoltaic power generation and building load to net energy consumption have undergone structural changes based on the proportional displacement consistency index. S3. In the case of structural changes, perform dominant substitution projection calculation on the proportional displacement vector to generate the substitution intensity index and dominant conversion direction factor to determine the degree of substitution. S4. Based on the constituent substitution intensity index and the dominant conversion direction factor, establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, and adjust the judgment boundary of multiple scenarios according to the constituent substitution degree. S5. Based on the adjusted multi-scenario judgment boundary, the scenario weight distribution and prediction parameter weight are updated synchronously to complete the dynamic prediction of energy consumption of building photovoltaic curtain walls in multiple scenarios.

2. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 1, characterized in that, S1 specifically refers to: The photovoltaic power generation data and building load data of the building photovoltaic curtain wall are resampled according to a preset unified time scale, and a time index mapping relationship is established. The photovoltaic power generation data and building load data are mapped to the same time index sequence. The data points corresponding to the missing time index are interpolated according to the numerical change ratio between adjacent time indices to generate a photovoltaic power generation data sequence and a building load data sequence after unified time scale alignment. Based on the photovoltaic power generation data sequence and building load data sequence after unified time scale alignment, the net energy consumption sequence is calculated by subtracting the photovoltaic power generation data from the building load data. The absolute value of the net energy consumption is the absolute value of the net energy consumption sequence under the corresponding time index. The photovoltaic power generation data and building load data are divided by the absolute value of the net energy consumption under the corresponding time index to generate the photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence. The photovoltaic power generation contribution ratio sequence and the building load contribution ratio sequence are arranged in matrix order according to time index. The net energy consumption contribution ratio evolution matrix is ​​constructed with time index as row coordinate and photovoltaic power generation contribution ratio and building load contribution ratio as column coordinate.

3. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 1, characterized in that, S2 specifically includes the following steps: S201. Based on the net energy consumption contribution ratio evolution matrix, read the photovoltaic power generation contribution ratio column vector and the building load contribution ratio column vector respectively, calculate the ratio difference between adjacent time indices according to the time index order, arrange the ratio differences corresponding to consecutive time indices in sequence, and form the photovoltaic power generation contribution ratio ratio displacement vector and the building load contribution ratio displacement vector. S202. Based on the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio, the proportional displacement direction consistency and proportional displacement amplitude deviation are combined at the corresponding time index position to generate the proportional displacement consistency index, wherein the proportional displacement consistency index is composed of the direction consistency parameter and the amplitude deviation parameter. S203. Compare the proportional displacement consistency index with the preset structural change judgment interval. When the proportional displacement consistency index falls into the preset structural change judgment interval, it is determined that the proportion of photovoltaic power generation and building load to net energy consumption has undergone structural change; otherwise, it is determined that no structural change has occurred.

4. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 3, characterized in that, S202 specifically refers to: At the corresponding time index positions, the number of times the photovoltaic power generation contribution ratio and the number of times the building load contribution ratio change in the same direction and in opposite directions are counted respectively. The difference between the number of times ... The displacement value difference between the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio is calculated at the corresponding time index position. The absolute value of the displacement value difference under continuous time index is accumulated. Then, the accumulated result is divided by the total number of time indexes for mean normalization to generate the proportional displacement amplitude deviation. The proportional displacement amplitude deviation is proportionally mapped to the preset amplitude benchmark to construct the amplitude deviation parameter. The directional consistency parameter and the amplitude deviation parameter are linearly combined according to preset weights to generate the proportional displacement consistency index, which is used to represent the degree of coordinated change between the proportional displacement vector of photovoltaic power generation contribution and the proportional displacement vector of building load contribution in the time series.

5. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 1, characterized in that, S3 specifically includes the following steps: S301. In the event of structural changes, obtain the proportional displacement vector of the photovoltaic power generation contribution ratio and the proportional displacement vector of the building load contribution ratio respectively. Construct a proportional displacement combination vector at the same time index position. Use the symmetrical direction vector where the values ​​of the photovoltaic power generation contribution ratio displacement component and the building load contribution ratio displacement component are equal as the dominant substitution reference direction. Perform vector projection calculation on the proportional displacement combination vector to generate the dominant substitution projection value corresponding to each time index. S302. The absolute values ​​of the dominant substitution projection values ​​under the continuous time index are accumulated and normalized by dividing by the total number of time indices to generate the substitution intensity index. The number of times the dominant substitution projection value is positive and the number of times it is negative are counted. The difference between the positive and negative values ​​and the sum of the positive and negative values ​​are normalized to generate the dominant conversion direction factor. S303. The substitution intensity index and the dominant conversion direction factor are matched with the preset substitution classification intervals respectively. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is positive, the degree of substitution is determined to be power generation-dominated substitution. When the substitution intensity index falls into the corresponding intensity interval and the dominant conversion direction factor is negative, the degree of substitution is determined to be load-dominated substitution. When the substitution intensity index does not fall into any intensity interval, the degree of substitution is determined to be a state where a stable substitution has not been formed.

6. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 5, characterized in that, S301 specifically refers to: In the event of structural changes, the proportional displacement vectors of the photovoltaic power generation contribution ratio and the building load contribution ratio are read in time index order. At each time index position, the proportional displacement components of the photovoltaic power generation contribution ratio and the building load contribution ratio are extracted, and a two-dimensional proportional displacement combination vector is constructed using the two proportional displacement components. In the proportional displacement coordinate plane, a symmetrical direction vector with the same value and sign as the proportional displacement component contributed by photovoltaic power generation and the proportional displacement component contributed by building load is constructed as the dominant alternative reference direction. The projection length of the proportional displacement combination vector on the dominant alternative reference direction is calculated to generate the dominant alternative projection value at the corresponding time index position. The dominant substitution projection values ​​under continuous time index are arranged in time index order, and the dominant substitution projection values ​​are subjected to sign preservation processing to form a dominant substitution projection sequence containing directional information, which is used to subsequently generate the substitution intensity index and the dominant conversion direction factor.

7. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 1, characterized in that, S4 specifically refers to: Based on the substitution intensity index and the dominant conversion direction factor, a proportional dimension feature vector is constructed, and the trend slope of the net energy consumption sequence is calculated in the corresponding time interval to generate a net energy consumption trend dimension vector. The proportional dimension feature vector and the net energy consumption trend dimension vector are mapped to the same coupled coordinate space to establish a coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension. In the coupled mapping relationship between the proportional dimension and the net energy consumption trend dimension, the coordinate positions of the proportional dimension feature vector in the coupled coordinate space are extracted according to the degree of substitution. When the degree of substitution is power generation-led substitution, the projection offset of the proportional dimension feature vector along the negative axis of the net energy consumption trend is calculated; when the degree of substitution is load-led substitution, the projection offset of the proportional dimension feature vector along the positive axis of the net energy consumption trend is calculated; when the degree of substitution is that a stable substitution state has not been formed, it is determined that the proportional dimension feature vector does not produce a trend axis offset. The projection offset obtained from the corresponding degree of substitution is superimposed on the original multi-scene judgment boundary threshold, and the multi-scene judgment boundary is directionally shifted and adjusted so that the multi-scene judgment boundary forms an updated judgment interval in the coupling space of the proportional dimension and the net energy consumption trend dimension.

8. The method for dynamic prediction of energy consumption in multiple scenarios for building photovoltaic curtain walls according to claim 1, characterized in that, S5 specifically refers to: Based on the adjusted multi-scenario judgment boundary, the proportional dimension feature vector and the net energy consumption trend dimension vector of the current time interval are matched in the coupling space of proportional dimension and net energy consumption trend dimension to determine the corresponding scenario category. Based on the interval span corresponding to the adjusted multi-scenario judgment boundary, the interval coverage ratio of each scenario is recalculated, the interval coverage ratio of each scenario is normalized, and the updated scenario weight distribution is generated. Based on the updated scenario weight distribution, the prediction parameter weights corresponding to each scenario are updated synchronously. The updated scenario weight distribution is used as a weight coefficient to reconstruct the prediction parameter weights, generating updated prediction parameter weights. The updated prediction parameter weights are then applied to the net energy consumption prediction calculation for subsequent time intervals.

Images