A method for calculating energy saving of a lighting system after energy-saving reconstruction of building lighting

Abstract

The application discloses a kind of based on building lighting energy-saving reconstruction after lighting system energy-saving quantity calculation method, it is related to intelligent lighting technical field, including, using multimodal data fusion mechanism integration preprocessed environmental monitoring data and lighting quality image parameter, construct initial energy consumption benchmark value model to carry out regression analysis to comprehensive feature data, update initial energy consumption benchmark value model, generate final energy consumption benchmark value model;Remote energy consumption detection and diagnosis center compares and analyzes the energy consumption reference value of initial energy consumption benchmark value model with the energy consumption reference value of final energy consumption benchmark value model, combines dynamic energy-saving quantity correction mechanism, outputs energy-saving quantity calculation result.The application generates comprehensive feature data by using multimodal data fusion mechanism integration preprocessed environmental monitoring data and lighting quality image parameter, realizes the unified modeling to the multiple factors of influence energy consumption.

Description

Technical Field

[0001] This invention relates to the field of intelligent lighting technology, and in particular to a method for calculating the energy savings of a lighting system after energy-saving renovation of building lighting. Background Technology

[0002] In the field of lighting energy conservation assessment, energy saving calculation, as an important technical means to measure the effectiveness of energy-saving renovations, is widely used in building energy management, green lighting projects, and energy service contracts. Traditional methods typically compare and analyze historical energy consumption data with measured data after renovations, using linear regression or benchmark models to establish energy consumption prediction mechanisms. These are combined with environmental monitoring parameters such as light intensity, temperature, and humidity to construct energy consumption reference values ​​for evaluating energy-saving effects. Furthermore, some methods introduce multi-source data fusion strategies to improve the completeness of feature representation and modeling accuracy.

[0003] However, existing methods still have limited utilization of image parameters related to lighting quality, leading to biases in the energy consumption model's portrayal of actual operating conditions. Current methods typically use statistical modeling of multi-source environmental parameters and historical energy consumption data to indirectly reflect the operating status of lighting equipment, thus compensating for model bias caused by insufficient fusion of lighting quality image parameters. Especially under complex environmental conditions, relying solely on environmental monitoring data is insufficient to comprehensively reflect the energy consumption characteristics of the lighting system, affecting the accuracy and stability of energy saving calculations. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for calculating the energy savings of a lighting system after energy-saving renovation of building lighting to solve the problem of deviation in the characterization of actual operating conditions by energy consumption models.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, this invention provides a method for calculating energy savings of a lighting system after energy-saving renovation of building lighting. The method includes: collecting environmental monitoring data and lighting quality image parameters, and preprocessing them; integrating the preprocessed environmental monitoring data and lighting quality image parameters using a multimodal data fusion mechanism to generate comprehensive feature data; constructing an initial energy consumption benchmark model to perform regression analysis on the comprehensive feature data and outputting an energy consumption reference value; based on the energy consumption reference value, carrying out energy-saving renovations on the lighting renovation area, collecting environmental monitoring data and lighting quality image parameters of the area after the renovation, generating a post-renovation lighting quality assessment, and updating the initial energy consumption benchmark model to generate a final energy consumption benchmark model; and a remote energy consumption detection and diagnosis center comparing and analyzing the energy consumption reference values ​​of the initial and final energy consumption benchmark models, and, combined with a dynamic energy-saving correction mechanism, outputting the energy-saving calculation results.

[0007] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation as described in this invention, the environmental monitoring data includes light intensity, air temperature, and humidity; The lighting quality image parameters include brightness distribution, illuminance uniformity, color temperature consistency, and glare index; The preprocessing includes data cleaning, normalization, feature extraction, and data fusion.

[0008] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation described in this invention, the step of integrating preprocessed environmental monitoring data and lighting quality image parameters using a multimodal data fusion mechanism to generate comprehensive feature data includes the following specific steps: Time alignment and matching are performed on the preprocessed environmental monitoring data and lighting quality image parameters to generate a time-matched sample set. Feature extraction and encoding are performed on the time-matched sample set to form a structured matrix; A multimodal data fusion mechanism is used to fuse features of a structured matrix to generate comprehensive feature data.

[0009] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation as described in this invention, the specific steps for constructing an initial energy consumption benchmark model, performing regression analysis on comprehensive characteristic data, and outputting energy consumption reference values ​​are as follows: Extract the structured fusion feature matrix from the comprehensive feature data as the input variable, and extract the energy consumption values ​​corresponding to the structured fusion feature matrix from the historical real energy consumption data to generate the output variable; The input and output variables are divided into training sets, and the training sets are modeled and trained using linear regression to generate an initial energy consumption baseline model. The comprehensive feature data is input into the initial energy consumption baseline model for analysis to obtain energy consumption reference values.

[0010] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation as described in this invention, the steps include: conducting energy-saving renovations on the lighting renovation area based on energy consumption reference values, collecting environmental monitoring data and lighting quality image parameters of the renovated area, and generating a post-renovation lighting quality assessment. Analyze energy consumption reference values, identify areas with abnormal energy consumption, output lighting energy-saving renovation plans, and carry out energy-saving renovations on the lighting renovation areas according to the plans, generating the renovated lighting areas. Collect environmental monitoring data and lighting quality image parameters of the area after the lighting renovation, generate post-renovation characteristic data, analyze them, and obtain an assessment of the lighting quality after the renovation.

[0011] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation as described in this invention, the step of updating the initial energy consumption benchmark model to generate the final energy consumption benchmark model refers to using data fusion and feature mapping methods to overlay the post-renovation lighting quality assessment onto the training set as an extended training set, and updating the initial energy consumption benchmark model according to the data distribution characteristics of the extended training set to output the final energy consumption benchmark model.

[0012] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving renovation as described in this invention, the remote energy consumption detection and diagnosis center compares and analyzes the energy consumption reference values ​​of the initial energy consumption benchmark model and the final energy consumption benchmark model. The specific steps are as follows: The modified feature data are input into the initial energy consumption benchmark model and the final energy consumption benchmark model respectively for inference and prediction, and a comparative energy consumption reference value is generated. By using the difference calculation method, the energy consumption reference values ​​are compared and analyzed over time periods to obtain preliminary energy-saving results.

[0013] As a preferred embodiment of the energy-saving calculation method for lighting systems after building lighting energy-saving retrofitting described in this invention, the specific steps for combining a dynamic energy-saving correction mechanism to output the energy-saving calculation result are as follows: Based on the actual operating conditions of lighting and the operating status of lighting equipment, a dynamic adjustment factor is constructed, and the preliminary energy-saving results are corrected by a weighted correction method to generate the corrected energy-saving results; The preliminary energy-saving results and the revised energy-saving results are compared and evaluated, and the energy-saving calculation results are output in conjunction with the lighting energy-saving renovation plan.

[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the energy-saving calculation method for a lighting system after energy-saving renovation of building lighting as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the energy-saving calculation method for a lighting system after energy-saving renovation of building lighting as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By integrating preprocessed environmental monitoring data and lighting quality image parameters through a multimodal data fusion mechanism, comprehensive feature data is generated, realizing unified modeling of multiple factors affecting energy consumption. This improves the expressive integrity of the structured fusion feature matrix and the consistency of model input, enabling the energy consumption benchmark model to more comprehensively and accurately reflect the operating status of the lighting system. By updating the initial energy consumption benchmark model and generating the final energy consumption benchmark model, expanding the training set based on the modified lighting quality assessment data, and dynamically adjusting the model parameters, adaptive updating of the model with changes in state is achieved, enhancing the timeliness and applicability of energy consumption prediction. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method for calculating energy savings in lighting systems after energy-saving retrofitting of building lighting.

[0019] Figure 2 This is a flowchart of a multimodal data fusion mechanism.

[0020] Figure 3 Flowchart for building and updating energy consumption baseline models.

[0021] Figure 4 A flowchart for energy saving calculation and comparative analysis. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for calculating the energy savings of a lighting system after energy-saving renovation of building lighting, including the following steps: S1. Collect environmental monitoring data and lighting quality image parameters, and perform preprocessing.

[0026] Environmental monitoring data includes light intensity, air temperature, and humidity; It should be noted that multiple sensors are deployed in the lighting area to collect environmental monitoring data at regular intervals, for example, every 5 minutes, to ensure continuous and stable environmental status information. Illuminance is recorded in lux, reflecting the combined illuminance level of natural and artificial light sources within the lighting area; air temperature is recorded in degrees Celsius to analyze the impact of environmental thermal conditions on the operation of lighting equipment; humidity is recorded as a relative percentage, reflecting the moisture content in the air, to help determine the microclimate environment in which the lighting system operates. Illuminance, air temperature, and humidity are standardized using a unified timestamp format.

[0027] Lighting quality image parameters include luminance distribution, illuminance uniformity, color temperature consistency, and glare index; Specifically, lighting quality image parameters are extracted from image data captured under standard lighting conditions. Brightness distribution is recorded in candela per square meter, reflecting the spatial variation of luminous intensity at different locations within the illuminated area. Illuminance uniformity is expressed as a percentage, calculated from the average illuminance difference in key areas of the image, describing the evenness of light distribution within the illuminated area. Color temperature consistency is recorded in Kelvin, reflecting the spatial stability of the light source color through statistical analysis of color temperature values ​​at multiple sampling points in the image. Glare index is calculated according to standard recommended methods, quantified based on the contrast between bright areas and background brightness in the image, with a value range of, for example, 1 to 9, used to assess the impact of lighting on visual comfort. Standard recommended methods refer to standardized technical processes adopted in a specific field to ensure consistency and comparability in data collection, analysis, or evaluation. Brightness distribution, illuminance uniformity, color temperature consistency, and glare index are standardized using a unified timestamp format.

[0028] Preprocessing includes data cleaning, normalization, feature extraction, and data fusion; Furthermore, data cleaning removes outliers and missing values ​​from environmental monitoring data and lighting quality image parameters according to the set time range and numerical reasonableness rules. For example, data points with light intensity exceeding the physical reasonable range are removed. Normalization transforms the parameter differences between environmental monitoring data and lighting quality images into a unified numerical range. For example, air temperature and humidity are scaled to the range of 0 to 1 to eliminate the impact of parameter differences on subsequent analysis. The set time range and numerical reasonableness rules refer to the standards used in the data cleaning process to judge the validity of environmental monitoring data and lighting quality image parameters based on the actual collection time and physical meaning. These standards are used to identify and remove data records that exceed the reasonable range.

[0029] Feature extraction extracts representative information from environmental monitoring data and lighting quality image parameters, such as brightness distribution and illuminance uniformity as key features from lighting quality image parameters. Data fusion aligns and merges environmental monitoring data and lighting quality image parameters according to a unified timestamp, generating a structured dataset for modeling. Through these processes, preprocessed environmental monitoring data and lighting quality image parameters are finally generated.

[0030] S2. Use a multimodal data fusion mechanism to integrate preprocessed environmental monitoring data and lighting quality image parameters to generate comprehensive feature data. Construct an initial energy consumption benchmark model to perform regression analysis on the comprehensive feature data and output energy consumption reference values.

[0031] The time stamp matching method is used to perform time alignment and matching on the preprocessed environmental monitoring data and lighting quality image parameters to generate a time matching sample set. Specifically, a fixed time window matching strategy is adopted, such as setting a time tolerance range of ±30 seconds. Environmental monitoring data records and lighting quality image parameter records existing within the same time range are matched one-to-one to ensure consistency in the time dimension between the environmental monitoring data and the lighting quality image parameters. Data records for which no corresponding time match can be found, such as unilateral data records exceeding the set time window, are retained separately but not included in the time matching sample set. After completing the time alignment and matching operations, a set consisting of environmental monitoring data records and lighting quality image parameter records is output as the time matching sample set.

[0032] Feature extraction and encoding of time-matched sample sets are performed using image feature extraction methods to form a structured matrix; It should be noted that for the four parameters included in the time-matched sample set—brightness distribution, illuminance uniformity, color temperature consistency, and glare index—quantifiable numerical features were extracted using image analysis methods. The mean and variance of the brightness distribution were calculated using pixel-based statistical methods, such as averaging the pixel intensity per candela per square meter. Illuminance uniformity was calculated using image region segmentation and local illuminance comparison, with the result expressed as a percentage. The color temperature consistency index was obtained through color space conversion and recorded as a value in Kelvin units. The glare index was obtained based on the relationship between the brightness of bright areas and the background in the image, with the quantified result being an integer between 1 and 9. The extracted numerical features were numerically encoded according to a unified format and concatenated with the corresponding environmental monitoring data from the same time-matched sample set according to a unified timestamp, forming a multi-dimensional feature vector containing both environmental and image features. This vector was then organized into a structured matrix.

[0033] A multimodal data fusion mechanism is used to fuse features of the structured matrix to generate comprehensive feature data; Furthermore, image feature vectors and environmental feature vectors are extracted from the structured matrix and concatenated using a time-stamp-based row-by-row matching method. For example, at a certain time point, the image feature vector contains brightness distribution of 0.72, illuminance uniformity of 0.85, color temperature consistency of 5600 Kelvin, and glare index of 4. The corresponding environmental feature vector contains light intensity of 0.68, air temperature of 23 degrees Celsius, and humidity of 58%. The image feature vector and environmental feature vector are horizontally merged in chronological order to form a joint feature vector. The joint feature vectors are standardized to ensure that they are within the same numerical range. For example, all joint feature vectors are scaled to the range of 0 to 1. The final output is a structured dataset composed of image feature vectors and environmental feature vectors, which serves as comprehensive feature data.

[0034] The structured fusion feature matrix in the comprehensive feature data is extracted using the feature data extraction method and used as the input variable. The energy consumption value corresponding to the structured fusion feature matrix in the historical real energy consumption data is extracted using the target variable extraction method to generate the output variable. It should be noted that a field-selection-based feature data extraction method is adopted to extract multimodal input feature sets from comprehensive feature data as structured fusion feature matrices. Each set of structured fusion feature matrices represents a complete multimodal feature vector. The multimodal input feature sets include brightness distribution, illuminance uniformity, color temperature consistency, glare index, illuminance intensity, air temperature, and humidity. Based on timestamp matching, historical energy consumption values ​​corresponding to the structured fusion feature matrix are extracted from historical energy consumption data to form output variables. For example, if the actual electricity consumption corresponding to an input variable in a certain hour is 0.45 kWh, it will be selected as the output variable for that hour. This ensures that each input variable has a corresponding time-aligned energy consumption value; historical energy consumption data consists of recorded, time-series organized actual electricity consumption data in kilowatt-hours. Finally, input and output variables are generated, where the input variables are composed of the structured fusion feature matrix, and the output variables are the historical energy consumption values ​​for the corresponding time period of the input variables.

[0035] The input and output variables are divided into training sets according to the dataset partitioning method, and the training sets are modeled and trained using the linear regression method to generate an initial energy consumption baseline model. Specifically, input and output variables are combined to generate a sample set. This sample set is then segmented using a time-series-based partitioning method, preserving the original temporal structure according to the chronological order of the sample set. For example, the first 80% of the continuous time periods in the total sample set are used as the training set, while the remaining 20% ​​is retained for subsequent validation and testing. The original temporal structure refers to the arrangement of the sample set data according to the chronological order during the data collection process, preserving the sequential relationships and dynamic changes between different time points.

[0036] A linear regression model is established to model the linear relationship between input and output variables in the training set. This model constructs a functional relationship between energy consumption reference values ​​and input variables. The least squares method is used to solve for the regression coefficients corresponding to the input variables. During the training process of the linear relationship model, the regression coefficients are optimized to generate an initial energy consumption baseline model. Regression coefficients are numerical values ​​used in the linear regression model to represent the strength and direction of the relationship between input and output variables. Each set of input variables corresponds to a regression coefficient. During the modeling process, the least squares method is used to calculate the regression coefficients, ensuring that the model's predicted values ​​are as close as possible to the actually observed energy consumption values. The model's predicted values ​​refer to the predicted values ​​derived from the trained initial energy consumption baseline model based on the current input variables.

[0037] The comprehensive feature data is input into the initial energy consumption baseline model for analysis to obtain energy consumption reference values; Furthermore, the comprehensive feature vectors of the comprehensive feature data are sequentially input into the initial energy consumption benchmark model, and linear combination calculations are performed according to the regression coefficients determined during the training phase. After completing the input and calculation of all comprehensive feature vectors, the output is a set of energy consumption reference values ​​arranged by time, corresponding one-to-one with each time point in the comprehensive feature data, generating energy consumption reference values. The regression coefficients determined during the training phase refer to a set of numerical parameters used to describe the relationship between input and output variables, obtained by solving the least squares method during the training of the linear regression model.

[0038] The linear combination calculation is performed based on the regression coefficients determined during the training phase, and the expression is as follows: in, This represents a sequence of energy consumption reference values; Represents the regression coefficient; Represents the comprehensive feature vector; This represents the input variable.

[0039] S3. Based on the energy consumption reference value, carry out energy-saving renovation of the lighting renovation area, collect environmental monitoring data and lighting quality image parameters of the area after the lighting renovation, generate a lighting quality assessment after the renovation, update the initial energy consumption benchmark value model, and generate the final energy consumption benchmark value model.

[0040] The energy consumption reference value is analyzed using time series comparative analysis to identify abnormal energy consumption areas, output lighting energy-saving renovation plans, and carry out energy-saving renovations on the lighting renovation areas according to the lighting energy-saving renovation plans to generate the renovated lighting areas. Specifically, energy consumption reference values ​​are organized into time series data in chronological order and compared with historically observed real energy consumption data for each time period. By setting a deviation threshold, for example, marking time periods with energy consumption deviations exceeding 15% as abnormal energy consumption intervals, areas of excessive energy consumption or low efficiency in the lighting system can be identified. The 15% deviation threshold is mainly based on a combination of industry-standard energy-saving assessment criteria and practical engineering experience, using energy consumption fluctuations within the range of 10% to 20% as a reference indicator for operational stability. The deviation threshold is set based on the difference between the energy consumption reference values ​​and historically observed real energy consumption data, and is used to determine whether the lighting system exhibits abnormal energy consumption behavior. The value range is usually determined based on the statistical characteristics of historical error distribution; for example, the standard deviation of historical errors is selected as a reference, and the deviation threshold is set to the mean error plus 1 to 2 times the standard deviation to cover the normal operating fluctuation range. Based on the identified abnormal energy consumption areas, the contribution of time-series changes in multimodal input features concentrated in these areas is calculated to determine the main influencing factors causing energy consumption to deviate from the baseline value. A lighting energy-saving renovation plan is then output, including suggestions for adjusting lighting equipment parameters, light source replacement strategies, and environmental control optimization measures. Based on the plan, specific lighting equipment updates and parameter optimization operations are implemented in the renovation areas, such as replacing with high-efficiency LED lamps, adjusting light intensity control strategies, and optimizing color temperature consistency adjustment mechanisms, ultimately generating the post-renovation area.

[0041] The contribution of multimodal input features concentrated in the abnormal energy consumption region in the time series variation trend is calculated, and the expression is: in, Indicates the first The contribution of each feature; This represents the first value obtained from training the initial energy consumption baseline model. The regression coefficients corresponding to the features; Indicates the first The standard deviation of the feature; The standard deviation of energy consumption reference values; This represents the feature number in the multimodal input feature set.

[0042] Collect environmental monitoring data and lighting quality image parameters of the area after lighting renovation, generate post-renovation characteristic data, and analyze them using lighting quality analysis methods to obtain an assessment of the lighting quality after renovation. It should be noted that on-site data collection was conducted on the area after the lighting renovation to obtain three environmental monitoring data, including light intensity, air temperature and humidity, as well as four lighting quality image parameters, including brightness distribution, illuminance uniformity, color temperature consistency and glare index, to generate characteristic data after the renovation.

[0043] The lighting quality image parameters of the area after the lighting renovation are evaluated using a lighting quality analysis method. Specific evaluation content includes whether the brightness distribution meets spatial balance requirements, whether the illuminance uniformity reaches the standard threshold, whether the color temperature consistency is stable, and whether the glare index is controlled within a reasonable range. The final output is a lighting quality assessment of the renovated area, reflecting the visual comfort and lighting performance of the lighting system after the renovation. Spatial balance requirements mean that the brightness distribution should meet the specifications for spatial lighting uniformity, ensuring that there are no obviously overly bright or overly dark areas within the visual area, and the standard deviation of the brightness distribution should be controlled within 20% of the average brightness. The standard threshold refers to the minimum performance index that illuminance uniformity must achieve, used to measure whether the lighting system meets basic lighting quality requirements; the illuminance in the darkest area should not be less than 60% of the average illuminance. A reasonable range means that the glare index should be controlled within a range that does not affect visual comfort and work efficiency, avoiding visual fatigue caused by excessive glare. Based on the Unified Glare Index (UGR) evaluation system, for example, a UGR value between 16 and 22 indicates that glare is within an acceptable range, and below 19 indicates good glare control.

[0044] The improved lighting quality assessment is superimposed onto the training set using data fusion and feature mapping methods to serve as an extended training set. The initial energy consumption benchmark model is then updated based on the data distribution characteristics of the extended training set, resulting in the final energy consumption benchmark model. Furthermore, lighting quality image parameters and environmental monitoring data are extracted from the post-modification lighting quality assessment and time-aligned to form a set of structured post-modification feature data. Data fusion technology is used to concatenate the post-modification feature data with the training set in both time and feature dimensions to generate an extended training set. Feature mapping is then used to normalize the features in the extended training set, ensuring that the post-modification feature data maintains consistency with the input requirements of the initial energy consumption baseline model in terms of numerical range and distribution characteristics. Based on the data distribution characteristics of the extended training set, including mean shift, variance variation, and feature correlation changes, the regression coefficients in the initial energy consumption baseline model are retrained and optimized. The least squares method is used to update the regression coefficients corresponding to each feature in the extended training set. Cross-validation is used to control the risk of model overfitting. The extended training set is divided into multiple mutually exclusive subsets according to time sequence. For example, a five-fold cross-validation method is used, dividing the data in the extended training set into five consecutive time-segment blocks. Each time, four data blocks are selected as the training subset to update the parameters of the cross-validation control model, and the remaining data block is used as the validation subset to evaluate the predictive performance of the cross-validation control model on the extended training set data.

[0045] During each training iteration, the regression coefficients in the initial energy consumption baseline model are updated based on the current training subset. The prediction error is obtained using the validation subset. By comparing the trend of the validation error across different training epochs, it is determined whether the initial energy consumption baseline model exhibits overfitting. Regularization parameters are then adjusted, for example, by introducing an L2 regularization term and setting the regularization coefficient to 0.01 to suppress excessive fluctuations in the regression coefficients. The validation subset refers to a small portion of data partitioned from the expanded training set during cross-validation, used to evaluate the prediction performance of the initial energy consumption baseline model on unknown data. The regularization parameter controls the model complexity of the initial energy consumption baseline model, preventing the regression coefficients from overfitting the training data. The best-performing set of regression coefficients is retained as the parameter configuration for the final energy consumption baseline model. The final energy consumption baseline model is then output. S4. The remote energy consumption detection and diagnosis center compares and analyzes the energy consumption reference values ​​of the initial energy consumption benchmark model and the final energy consumption benchmark model, and outputs the energy saving calculation results by combining the dynamic energy saving correction mechanism.

[0046] Using a unified feature input method, the modified feature data are input into the initial energy consumption benchmark model and the final energy consumption benchmark model respectively for inference and prediction, generating comparative energy consumption reference values. It should be noted that the modified feature data is input into the initial energy consumption benchmark model, and energy consumption reference values ​​for each time point are generated based on the regression coefficients determined during the training phase, forming an energy consumption reference value sequence composed of multiple time points. The modified feature data is then input into the final energy consumption benchmark model, and energy consumption reference values ​​for the corresponding time points are generated based on the updated regression coefficients from the expanded training set, forming the final energy consumption reference value sequence. The energy consumption reference value sequence and the final energy consumption reference value sequence are combined to generate a comparative energy consumption reference value.

[0047] By using the difference calculation method, the energy consumption reference values ​​are compared and analyzed over time periods to obtain preliminary energy-saving results. Furthermore, the energy consumption reference value sequence and the final energy consumption reference value sequence are compared at the same time points, with each comparison based on the same post-modification characteristic data. At each time point, the difference between the energy consumption reference value sequence and the final energy consumption reference value sequence is calculated. For example, if the initial energy consumption baseline model output is 1.8 kWh and the final energy consumption baseline model output is 1.5 kWh, the difference is 0.3 kWh. By performing the same difference operation on all time points sequentially, a set of preliminary energy saving results arranged by time is obtained, which reflects the expected reduction in energy consumption of the lighting system at different times after the implementation of energy-saving renovation measures.

[0048] Based on the actual operating conditions of lighting and the operating status of lighting equipment, a dynamic adjustment factor is constructed, and the preliminary energy-saving results are corrected by a weighted correction method to generate the corrected energy-saving results; Specifically, the changes in illuminance, air temperature fluctuations, and humidity levels under actual lighting operating conditions, as well as the aging degree of light sources, driver efficiency degradation, and cumulative operating time of lighting equipment, are extracted as key variables affecting energy consumption stability. These variables are then normalized, and principal component analysis (PCA) is used to obtain the covariance matrix among the key variables. The eigenvalues ​​of the covariance matrix and their corresponding direction vectors in the standardized data space are then calculated. Principal components with a cumulative contribution rate of over 85% are selected based on their eigenvalues ​​from largest to smallest to represent the main direction of change for the key variables. The relative contribution of each key variable to a principal component is determined based on its loading coefficient, and then converted into the weight allocation ratio for each key variable. The cumulative contribution rate refers to the sum of the proportions of variance explained by the top few principal components after sorting by the eigenvalues ​​of the covariance matrix from largest to smallest, relative to the total variance of all principal components. The contribution rate of each principal component represents its ability to represent the information of the key variables. The cumulative contribution rate is obtained by sequentially summing the contribution rates of multiple principal components. For example, the values ​​are: light source aging degree weight 0.4, driver efficiency decay weight 0.3, cumulative running time weight 0.2, and environmental temperature and humidity combined weight 0.1, ultimately constructing a dynamic adjustment factor; The dynamic adjustment factor at each time point is used as a correction coefficient and multiplied by the corresponding preliminary energy saving result. For example, if the preliminary energy saving at a certain time point is 0.3 kWh and the dynamic adjustment factor is 0.9, the corrected energy saving is 0.27 kWh. By processing the data at all time points in sequence, a complete corrected energy saving result is generated to reflect the actual energy saving effect of energy-saving retrofit measures under different operating conditions.

[0049] The energy-saving effect analysis method is used to compare and evaluate the preliminary energy-saving results and the revised energy-saving results. Combined with the lighting energy-saving renovation plan, the energy-saving calculation results are output.

[0050] It should be noted that the preliminary energy saving results and the revised energy saving results are compared in the same time sequence. At each time point, two energy saving values ​​are obtained and the difference between them is calculated. For example, if the preliminary energy saving at a certain time point is 0.3 kWh and the revised energy saving is 0.27 kWh, then the difference at that time point is 0.03 kWh. By performing the same comparison operation on all time points, a trend curve of the energy saving difference over time is formed. Based on the characteristics of the curve, it is possible to identify in which time periods the revised energy saving result shows a significant decrease or fluctuation compared to the preliminary energy saving result. For example, if the revised energy saving result decreases by 0.03 kWh compared to the preliminary energy saving result in certain time periods, it indicates that the actual operating fluctuations affect the lighting energy saving effect. By aligning and comparing the implementation time points of equipment replacement, parameter optimization, and other measures recorded in the lighting energy-saving renovation plan with the time series of preliminary and revised energy-saving results, this study aims to identify the actual impact of the energy-saving renovation plan on the energy consumption level of the lighting system. For example, after completing the lamp replacement operation within a certain time period, the changes in energy-saving data during that time period and at several subsequent time points are observed to determine if there is a significant downward trend in energy consumption. Statistical methods are used to evaluate the correlation between energy-saving data and the time points of the energy-saving renovation plan, identifying the actual impact of the energy-saving renovation plan on the energy consumption level of the lighting system. For instance, after the driver efficiency optimization was completed, the difference between the revised energy-saving result and the preliminary energy-saving result decreased by 0.02 kWh, indicating that equipment upgrades effectively reduced the energy consumption prediction bias. Finally, based on the above comparative analysis results, energy saving calculation results were generated, and detailed records were made of the changes in energy consumption reference values ​​before and after the energy-saving renovation, the application effect of dynamic adjustment factors, and energy saving assessment data for each time period.

[0051] This embodiment also provides a computer device applicable to the energy-saving calculation method of a lighting system after building lighting energy-saving renovation, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the energy-saving calculation method of a lighting system after building lighting energy-saving renovation as proposed in the above embodiment.

[0052] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0053] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the energy-saving calculation method for a lighting system after energy-saving renovation of building lighting, as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0054] In summary, this invention achieves unified modeling of multiple factors affecting energy consumption by integrating preprocessed environmental monitoring data and lighting quality image parameters using a multimodal data fusion mechanism to generate comprehensive feature data. This improves the expressive completeness of the structured fusion feature matrix and the consistency of model input, enabling the energy consumption benchmark model to more comprehensively and accurately reflect the operating status of the lighting system. Furthermore, by updating the initial energy consumption benchmark model and generating a final energy consumption benchmark model, expanding the training set based on the modified lighting quality assessment data, and dynamically adjusting the model parameters, this invention achieves adaptive updating of the model as the state changes, enhancing the timeliness and applicability of energy consumption prediction.

[0055] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for calculating the energy savings of a lighting system after energy-saving renovation of building lighting, characterized in that: include, The environmental monitoring data and lighting quality image parameters were collected and preprocessed. The specific steps are as follows: Deploy multiple sensors in the lighting area to collect environmental monitoring data at regular intervals; Lighting quality image parameters, including brightness distribution, illuminance uniformity, color temperature consistency, and glare index, are extracted from image data captured in standard lighting scenarios. Brightness distribution is recorded in candela per square meter, and illuminance uniformity is expressed as a percentage, calculated from the average illuminance difference in key areas of the image. Color temperature consistency is recorded in Kelvin and obtained through statistical analysis of the color temperature values ​​of multiple sampling points in the image; The glare index is calculated according to the standard recommended method and is quantified based on the contrast between the brightness of the bright areas in the image and the background brightness. The preprocessed environmental monitoring data and lighting quality image parameters are integrated using a multimodal data fusion mechanism to generate comprehensive feature data. An initial energy consumption benchmark model is constructed to perform regression analysis on the comprehensive feature data and output energy consumption reference values. Based on energy consumption reference values, energy-saving renovations are carried out in the lighting renovation area. Environmental monitoring data and lighting quality image parameters of the area after the renovation are collected to generate a lighting quality assessment after the renovation. The initial energy consumption baseline model is then updated to generate a final energy consumption baseline model. The specific steps are as follows: Analyze energy consumption reference values, identify areas with abnormal energy consumption, output lighting energy-saving renovation plans, and carry out energy-saving renovations on the lighting renovation areas according to the plans, generating the renovated lighting areas. Collect environmental monitoring data and lighting quality image parameters of the area after the lighting renovation, generate post-renovation feature data, analyze them, and obtain an assessment of the lighting quality after the renovation. Lighting quality image parameters and environmental monitoring data were extracted from the post-modification lighting quality assessment and time-aligned to form structured post-modification feature data. By using data fusion technology, the modified feature data and the training set are concatenated in both the time and feature dimensions to generate an expanded training set. The feature mapping method is used to normalize the features in the extended training set so that the modified feature data are consistent with the input requirements of the initial energy consumption baseline model in terms of numerical range and distribution characteristics. Based on the mean shift, variance change, and feature correlation change of the extended training set, the regression coefficients in the initial energy consumption benchmark model are retrained and optimized. The least squares method is used to update the regression coefficients corresponding to each feature in the extended training set. Cross-validation is used to control the risk of model overfitting. The best-performing set of regression coefficients is retained as the parameter configuration of the final energy consumption benchmark model, and the final energy consumption benchmark model is output. The remote energy consumption monitoring and diagnostic center compares and analyzes the energy consumption reference values ​​of the initial energy consumption baseline model and the final energy consumption baseline model, and outputs the energy saving calculation results by combining a dynamic energy saving correction mechanism. The specific steps are as follows: The modified feature data are input into the initial energy consumption baseline model and the final energy consumption baseline model respectively for inference and prediction, generating comparative energy consumption reference values. The comparative energy consumption reference values ​​are then compared and analyzed time-by-time using the difference calculation method to obtain preliminary energy saving results. The changes in light intensity, air temperature fluctuations, and humidity levels under actual lighting operating conditions, as well as the aging degree of light sources, driver efficiency degradation, and cumulative operating time of lighting equipment, are extracted as key variables affecting energy consumption stability and normalized. Principal component analysis is used to obtain the covariance matrix among key variables, and the eigenvalues ​​of the covariance matrix and the corresponding direction vectors in the standardized data space are solved. The covariance matrix is ​​sorted from largest to smallest according to the eigenvalues. The relative contribution of each key variable to the principal component is determined based on the loading coefficients of each key variable in each principal component, and converted into the weight allocation ratio of each key variable to construct a dynamic adjustment factor. The dynamic adjustment factor at each time point is used as a correction coefficient, and multiplied with the corresponding preliminary energy-saving result item by item to generate the corrected energy-saving result. The preliminary energy-saving results and the revised energy-saving results are compared and evaluated, and the energy-saving calculation results are output in conjunction with the lighting energy-saving renovation plan.

2. The method for calculating energy savings of a lighting system after energy-saving renovation of building lighting as described in claim 1, characterized in that: The steps for constructing an initial energy consumption benchmark model and performing regression analysis on comprehensive characteristic data to output energy consumption reference values ​​are as follows: Extract the structured fusion feature matrix from the comprehensive feature data as the input variable, and extract the energy consumption values ​​corresponding to the structured fusion feature matrix from the historical real energy consumption data to generate the output variable; The input and output variables are divided into training sets, and the training sets are modeled and trained using linear regression to generate an initial energy consumption baseline model. The comprehensive feature data is input into the initial energy consumption baseline model for analysis to obtain energy consumption reference values.

3. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the energy-saving calculation method for the lighting system after building lighting energy-saving renovation as described in any one of claims 1 to 2.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the energy-saving calculation method for the lighting system after energy-saving renovation of building lighting as described in any one of claims 1 to 2.

Images