Enterprise and public institution energy consumption management method and system based on B / S architecture

Abstract

The invention provides an enterprise and public institution energy consumption management method and system based on a B / S architecture, and the method comprises the steps: obtaining an original energy consumption data set through collecting the energy consumption data of time, space and intensity dimensions; aiming at the hierarchical energy consumption data set, adopting a clustering analysis algorithm to determine an association mode among all dimensions, and obtaining an energy consumption change rule; if the time dimension in the energy consumption change rule shows the peak-valley period difference, classifying the peak-valley period data based on a K-means algorithm to obtain peak-valley period energy consumption distribution; generating a trend prediction model containing seasonal changes through peak-valley period energy consumption distribution by adopting a time sequence analysis algorithm, and obtaining an energy consumption trend prediction result; and according to an energy consumption trend prediction result, generating a stratified analysis report including peak-valley period analysis and seasonal change adjustment suggestions, and obtaining structured report data.

Description

Technical Field

[0001] This invention belongs to the field of information processing technology, and in particular relates to an energy consumption management method and system for enterprise units based on B / S architecture. Background Technology

[0002] Energy management is a key area for enterprises and institutions to achieve sustainable development and cost optimization. Its importance lies in reducing energy consumption and improving resource utilization efficiency through scientific analysis and management. However, existing energy management methods have significant limitations in practical applications. Many solutions rely on localized deployment, resulting in high maintenance costs and inconvenient data access, making it difficult to meet the real-time collaboration needs of multiple departments and scenarios. In addition, existing systems often lack the ability to dynamically analyze energy consumption data in a hierarchical manner, failing to effectively address the differentiated management needs in complex scenarios, making it difficult for managers to accurately grasp the patterns of energy consumption changes.

[0003] Against this backdrop, the core challenges facing energy consumption management in enterprises and institutions lie in two interrelated technical attributes: first, multi-dimensional hierarchical modeling of energy consumption data; and second, real-time access and analysis based on the network. Energy consumption data involves multiple dimensions such as time, space, and intensity, requiring hierarchical modeling for refined management. However, existing technologies struggle to integrate this data across these dimensions and generate dynamic analysis. For example, an enterprise might need to simultaneously monitor the differences in electricity consumption across different departments during peak and off-peak hours and adjust strategies based on seasonal changes. However, existing systems cannot model these dimensions uniformly, leading to fragmented analysis results and hindering managers from quickly developing optimization plans. Furthermore, this lack of multi-dimensional modeling directly impacts the ability to access and analyze data in real time. Due to the lack of flexible network architecture support, users cannot easily access hierarchical analysis results at any time. Especially in cross-departmental collaborations, data silos are a significant problem, hindering the improvement of energy consumption management efficiency.

[0004] Therefore, how to build a network-based energy management platform, realize dynamic data analysis through multi-dimensional hierarchical modeling, and support users to access hierarchical reports and trend predictions in real time through a browser has become a key issue in the field of energy management for enterprises and institutions. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an energy consumption management method and system for enterprise units based on B / S architecture.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for energy consumption management of enterprise units based on a B / S architecture, comprising:

[0008] Obtain the raw energy consumption dataset;

[0009] Based on the original energy consumption dataset, a hierarchical energy consumption dataset is obtained;

[0010] For the hierarchical energy consumption dataset, the energy consumption variation pattern was obtained;

[0011] If the energy consumption variation pattern shows peak and valley time differences in the time dimension, then the peak and valley time data are classified based on the K-means algorithm to obtain the peak and valley time energy consumption distribution.

[0012] By analyzing the energy consumption distribution during peak and off-peak periods, we can obtain energy consumption trend prediction results.

[0013] Based on the energy consumption trend forecast results, a stratified analysis report is generated, which includes peak and off-peak period analysis and seasonal change adjustment suggestions, resulting in structured report data.

[0014] As a preferred method, raw energy consumption datasets are obtained by collecting energy consumption data in the time, space, and intensity dimensions.

[0015] As a preferred approach, based on the original energy consumption dataset, a hierarchical data processing method is used to construct a multi-layered data structure containing time, space, and intensity dimensions, resulting in a hierarchical energy consumption dataset.

[0016] As a preferred approach, clustering analysis algorithms are used for hierarchical energy consumption datasets to determine the correlation patterns between different dimensions and obtain the energy consumption change patterns.

[0017] As a preferred approach, by analyzing the energy consumption distribution during peak and off-peak periods, a time series analysis algorithm is used to generate a trend prediction model that incorporates seasonal variations, thereby obtaining energy consumption trend prediction results.

[0018] This invention also provides an enterprise / institutional unit energy management system based on a B / S architecture, comprising:

[0019] The data acquisition module is used to acquire raw energy consumption datasets;

[0020] The hierarchical processing module is used to obtain hierarchical energy consumption datasets based on the original energy consumption datasets;

[0021] The clustering analysis module is used to obtain the energy consumption change patterns for hierarchical energy consumption datasets;

[0022] The peak-valley classification module is used to classify the peak-valley data based on the K-means algorithm if the energy consumption change pattern shows the difference between peak and valley periods in the time dimension, so as to obtain the energy consumption distribution of peak and valley periods.

[0023] The trend prediction module is used to obtain energy consumption trend prediction results based on the energy consumption distribution during peak and off-peak periods;

[0024] The report generation module is used to generate hierarchical analysis reports based on energy consumption trend forecasts, including peak and off-peak period analysis and seasonal adjustment suggestions, resulting in structured report data.

[0025] As a preferred option, the data acquisition module obtains the raw energy consumption dataset by collecting energy consumption data in the dimensions of time, space, and intensity.

[0026] As a preferred approach, the hierarchical processing module uses a hierarchical data processing method to construct a multi-layered data structure containing time, space, and intensity dimensions based on the original energy consumption dataset, thereby obtaining a hierarchical energy consumption dataset.

[0027] As a preferred approach, the clustering analysis module employs a clustering analysis algorithm for the hierarchical energy consumption dataset to determine the correlation patterns between various dimensions and obtain the energy consumption change patterns.

[0028] As a preferred approach, the trend prediction module uses time series analysis algorithms to generate a trend prediction model that incorporates seasonal variations based on the energy consumption distribution during peak and off-peak periods, thereby obtaining energy consumption trend prediction results.

[0029] This invention addresses the challenge of complex time, space, and intensity-based data in energy management, where correlations are difficult to uncover. It constructs a multi-layered energy consumption dataset by integrating hierarchical data processing, cluster analysis, and time series forecasting techniques to reveal energy consumption patterns during peak and off-peak periods and seasonal variations. First, multi-dimensional energy consumption data is collected and processed hierarchically to form a structured dataset. Then, cluster analysis is used to uncover inter-dimensional correlations and identify peak and off-peak period differences. For peak and off-peak period data, the K-means algorithm is used for classification, and combined with time series analysis, a trend prediction model incorporating seasonal variations is generated. Finally, a structured analysis report is produced, offering suggestions for peak and off-peak period optimization and seasonal adjustments. This invention significantly improves the accuracy of energy consumption forecasting and management efficiency through multi-dimensional data fusion and intelligent analysis, providing a scientific basis for energy optimization. Attached Figure Description

[0030] Figure 1 This is a flowchart of the enterprise unit energy consumption management method based on B / S architecture according to the present invention. Detailed Implementation

[0031] To further understand the content of this invention, a detailed description of the invention is provided in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0032] Example 1:

[0033] like Figure 1 As shown in the figure, an embodiment of the present invention provides an energy consumption management method for enterprise units based on a B / S architecture, comprising:

[0034] Step S101: Obtain the raw energy consumption dataset by collecting energy consumption data in the time, space, and intensity dimensions.

[0035] Data collection rules are set up according to time, space, and intensity dimensions to obtain energy consumption data streams. The energy consumption data streams are sampled periodically at a preset collection frequency to obtain the raw energy consumption dataset. If the raw energy consumption dataset contains missing values, a linear interpolation algorithm is used to fill in the missing data to obtain a complete energy consumption dataset. Based on the complete energy consumption dataset, the energy consumption change trend over the time dimension is calculated to obtain energy consumption trend characteristics. K-means algorithm is applied to cluster analysis of the energy consumption trend characteristics to determine energy consumption pattern classifications. Based on the energy consumption pattern classifications, combined with the spatial and intensity dimensions, a multidimensional energy consumption distribution model is constructed to obtain energy consumption distribution characteristics. If the energy consumption distribution characteristics exceed a preset threshold, an anomaly detection algorithm is used to identify abnormal energy consumption points and obtain abnormal energy consumption records.

[0036] In one possible implementation, data collection rules are set using time, space, and intensity dimensions to obtain energy consumption data streams.

[0037] For example, in an energy management scenario for a smart building, the time dimension can be set to collect data once per hour, the spatial dimension covers various areas within the building such as offices, meeting rooms, and corridors, and the intensity dimension records the electricity consumption and air conditioning energy consumption of each area. The collection frequency can be preset to once per hour to ensure appropriate data granularity.

[0038] For example, an office records the electricity consumption of its air conditioning and lighting equipment every hour from 8 a.m. to 6 p.m., thus obtaining an energy consumption data stream.

[0039] It should be noted that if the original energy consumption dataset contains missing values, they can be filled using a linear interpolation algorithm.

[0040] For example, if data for an office at 10 AM is missing, linear interpolation can be performed based on the electricity consumption at 9 AM and 11 AM. Assuming the consumption at 9 AM is 500W and at 11 AM is 600W, the value at 10 AM would be supplemented to 550W, thus obtaining a complete energy consumption dataset. This method effectively maintains data continuity and avoids analytical bias caused by missing values.

[0041] For example, based on a complete energy consumption dataset, the trend of energy consumption changes over time can be calculated, and a trend curve can be plotted by comparing energy consumption data for different time periods each day.

[0042] For example, the office area of ​​a building shows a gradual increase in electricity consumption from 9:00 AM to 12:00 PM on weekdays, a high level from 1:00 PM to 5:00 PM, and a decrease after 6:00 PM, revealing an energy consumption trend. This trend reflects the correlation between working hours and energy consumption, which helps to optimize energy scheduling.

[0043] In one possible implementation, the K-means algorithm is applied to perform cluster analysis based on energy consumption trend characteristics to determine the energy consumption pattern classification.

[0044] For example, building energy consumption data can be categorized into high, medium, and low energy consumption. A certain office area might exhibit high energy consumption during the day and low energy consumption at night. Clustering results can help identify energy usage patterns in different areas, providing a basis for energy-saving strategies.

[0045] For example, by classifying energy consumption patterns and combining spatial and intensity dimensions, a multidimensional energy consumption distribution model can be constructed.

[0046] For example, high energy consumption patterns in office areas are concentrated during working hours, while meeting rooms experience peaks at specific times, such as 2 PM to 4 PM. The model can visually display the energy consumption distribution characteristics of each area, making it easier for managers to identify high-energy-consuming areas and optimize resource allocation.

[0047] It should be noted that if the energy consumption distribution characteristics exceed the preset threshold, such as the daily electricity consumption of an office area exceeding 1000W, abnormal energy consumption points can be identified through an anomaly detection algorithm.

[0048] For example, a statistical anomaly detection method can be used to calculate the mean and standard deviation of electricity consumption. If the electricity consumption during a certain period exceeds the mean plus twice the standard deviation, it is marked as an anomaly, and an abnormal energy consumption record is generated. This method can quickly locate equipment malfunctions or improper use, improving the accuracy of energy consumption management.

[0049] In one possible implementation, abnormal energy consumption records can be used for subsequent optimization.

[0050] For example, if an office is found to have abnormally high electricity consumption at night, an inspection reveals that the problem stems from equipment being left running. Timely adjustments can save energy costs. The combined application of these methods not only improves the completeness and accuracy of energy consumption data but also optimizes energy management through pattern recognition and anomaly detection, reducing operating costs and improving building energy efficiency.

[0051] Step S102: Based on the original energy consumption dataset, a hierarchical data processing method is used to construct a multi-layered data structure containing time, space, and intensity dimensions to obtain a hierarchical energy consumption dataset.

[0052] Data in time, space, and intensity dimensions are extracted from the original energy consumption dataset. Missing and outlier values ​​are removed through data cleaning to obtain a cleaned initial dataset. A hierarchical processing method is used to segment the cleaned initial dataset along the time dimension, generating time-series subsets. Based on these time-series subsets, spatial dimensions are combined to group the data, constructing a two-dimensional data structure encompassing time and space, generating a spatiotemporal correlated dataset. For the spatiotemporal correlated dataset, an intensity dimension is introduced, and the intensity values ​​are classified using the k-means clustering algorithm, resulting in an intensity-stratified dataset. If the hierarchical relationships in the intensity-stratified dataset meet preset thresholds, the data at each level are standardized to generate a standardized hierarchical dataset; otherwise, the previous step is returned to adjust the clustering parameters and reclassify. Through a multi-layer construction method, the standardized hierarchical dataset is integrated along the time, space, and intensity dimensions to generate a multi-layered energy consumption data structure. Key features are extracted from the multi-layered energy consumption data structure, and a decision tree algorithm is used to verify the hierarchical relationships, generating the final hierarchical energy consumption dataset.

[0053] For example, in energy consumption data analysis scenarios, raw energy consumption datasets typically contain data across three dimensions: time, space, and intensity, such as electricity consumption data for a building complex. Data cleaning is a crucial step, requiring the removal of missing and outlier values. If data for a building is missing on a particular day, it can be imputed using data from adjacent time points, such as the average electricity consumption of the previous and next hour. For outliers, such as a sudden increase in electricity consumption to 10 times the normal value at a certain moment, a threshold can be set to remove them. A cleaned initial dataset is more reliable, laying the foundation for subsequent analysis.

[0054] In one possible implementation, a hierarchical processing approach splits the data along the time dimension. Assuming the data covers a year, it can be split into 12 time-series subsets by month, each containing daily electricity consumption. This split facilitates the analysis of seasonal variations, such as peak summer electricity consumption. Combining this with spatial grouping, such as by building floors or areas, generates a spatiotemporally correlated dataset. Assuming a building complex is divided into two areas, A and B, where area A is the office area and area B is the equipment area, the spatiotemporally correlated dataset can reflect the electricity consumption characteristics of different areas at different times.

[0055] Specifically, after introducing the intensity dimension, the k-means clustering algorithm is applied to classify electricity intensity. Assuming intensity values ​​are divided into high, medium, and low categories, clustering can be performed based on the electricity consumption range (e.g., 0-50kWh for low, 50-100kWh for medium, and above 100kWh for high), generating a stratified intensity dataset. If the hierarchical relationship does not conform to the preset threshold, such as an excessively high proportion of high-intensity data, the k-value or initial cluster centers can be adjusted, and reclassification can be performed. This method ensures that the classification results are reasonable and reflect the true energy consumption pattern.

[0056] For example, standardization transforms data from different levels into a unified dimension, facilitating comparison. Suppose a dataset at a certain level has an electricity consumption range of 20-80 kWh; this can be mapped to the 0-1 range through linear normalization. Standardizing hierarchical datasets facilitates subsequent integration and analysis. When constructing a multi-layered energy consumption data structure, the time, space, and intensity dimensions can be integrated into a three-dimensional data model. For instance, using time as the x-axis, space as the y-axis, and intensity as the z-axis, a three-dimensional energy consumption distribution view can be generated, visually displaying energy consumption patterns.

[0057] In one possible implementation, a decision tree algorithm is used to verify the hierarchical relationship.

[0058] For example, decision trees can generate classification rules based on time (e.g., weekdays or weekends), space (e.g., area A or area B), and intensity (e.g., high or low) to determine which factors dominate energy consumption changes. Hypothesis testing reveals that area A has a high proportion of high-intensity electricity consumption on weekdays, allowing for further analysis of its equipment operating patterns. This method ensures that the final hierarchical energy consumption dataset accurately reflects multidimensional characteristics, facilitating subsequent optimization of energy management.

[0059] Understandably, each step of the above method closely revolves around the core needs of energy consumption data analysis, progressing step by step from data cleaning to final verification to ensure data integrity and analytical accuracy, providing a reliable basis for energy consumption management.

[0060] Step S103: For the hierarchical energy consumption dataset, a clustering analysis algorithm is used to determine the correlation patterns between each dimension and obtain the energy consumption change patterns.

[0061] A hierarchical energy consumption dataset is obtained. Data preprocessing is performed to clean missing and outlier values, resulting in a standardized energy consumption dataset. K-means clustering is used to cluster the standardized energy consumption dataset, determining preliminary groupings among dimensions and obtaining clustering results. Based on the clustering results, the correlation coefficient matrix between dimensions within each cluster is calculated to determine dimensional association patterns. If the significance of a dimensional association pattern is higher than a preset threshold, that association pattern is retained, resulting in a set of effective association patterns. Using this set of effective association patterns, time series analysis is employed to extract energy consumption change trends among dimensions, revealing energy consumption change patterns. Based on these patterns, principal component analysis is used to extract the main influencing dimensions, identifying key energy consumption drivers. Using these key energy consumption drivers, a multidimensional model of energy consumption change patterns is generated, yielding the final energy consumption change pattern.

[0062] In one possible implementation, data preprocessing is a crucial step when acquiring a hierarchical energy consumption dataset. The raw energy consumption data may contain missing values ​​and outliers; for example, in a factory's electricity consumption records, data for a particular day might be empty or display abnormally high values. During cleaning, missing values ​​can be imputed using the median, and outliers can be detected using box plots.

[0063] For example, in a factory's monthly electricity consumption record, the data for a certain day is 10,000 kWh, while the normal range is 1,000-2,000 kWh. This data can be replaced with the nearest mean. After cleaning, the dataset is standardized according to the time, space, and intensity dimensions to ensure that the data units are consistent across all dimensions, facilitating subsequent analysis.

[0064] For example, when using the K-means clustering algorithm to cluster a standardized energy consumption dataset, the data can be divided along a time dimension, such as by hour, date, or month. Suppose an industrial park has 10 factories, and the clustering goal is to divide the factories into three categories based on their energy consumption patterns. The clustering results might show that some factories have lower energy consumption at night and higher energy consumption during the day, while other factories have stable energy consumption throughout the day. This grouping reflects differences in production rhythms and provides a basis for subsequent analysis.

[0065] In one possible implementation, when calculating the correlation coefficient matrix between dimensions within a cluster, the temporal, spatial, and intensity-related relationships can be analyzed.

[0066] For example, the energy intensity of factories within a cluster is highly correlated with the operating time of their production equipment, with a correlation coefficient of 0.85, indicating that equipment operating time is a key factor in energy consumption changes. If the correlation coefficient is higher than a preset threshold of 0.7, this correlation pattern is retained, forming a set of valid correlation patterns. This method helps identify the core dimensions affecting energy consumption.

[0067] For example, when extracting energy consumption trends using time series analysis based on a set of effective correlation patterns, the sliding window method can be used to analyze the monthly energy consumption data of a factory. The results may show that energy consumption exhibits periodic peaks in winter, related to heating demand. This trend analysis helps predict future energy demand.

[0068] In one possible implementation, principal component analysis can be used to extract the main influencing dimensions, thereby reducing the dimensionality of multi-dimensional data such as time, space, and intensity.

[0069] For example, analysis revealed that equipment uptime and production load accounted for 80% of the variance in energy consumption variation, identifying these as key energy consumption drivers. These factors can be used to optimize production scheduling and reduce energy consumption.

[0070] For example, when generating a multidimensional model of energy consumption changes, a predictive model that includes time, space, and intensity can be constructed by combining key driving factors.

[0071] For example, a factory uses models to predict energy consumption during peak hours and adjusts its production schedule in advance. Such models provide data support for energy management and help in developing energy-saving strategies.

[0072] Step S104: If the energy consumption change pattern shows peak and valley time differences in the time dimension, then classify the peak and valley time data based on the K-means algorithm to obtain the peak and valley time energy consumption distribution.

[0073] Energy consumption data and corresponding timestamps are obtained from a data source to generate a time-series dataset with time dimension markers, resulting in structured energy consumption data. The structured energy consumption data is preprocessed, and a standardization method is used to eliminate dimensional differences, resulting in normalized energy consumption data. Based on the normalized energy consumption data, the K-means algorithm is applied for cluster analysis to determine the classification boundaries of peak and valley periods, obtaining initial clustering results. If the intra-cluster variance of the initial clustering results exceeds a preset threshold, the K value is adjusted and the K-means algorithm is re-executed to obtain optimized clustering results. By optimizing the clustering results, the time period range and corresponding energy consumption characteristics of each cluster are extracted to generate peak and valley period divisions. Based on the peak and valley period divisions, energy consumption statistics for each period are calculated to generate energy consumption distribution patterns. For the energy consumption distribution patterns, a visualization tool is used to plot time-series energy consumption curves, obtaining energy consumption distribution maps for peak and valley periods.

[0074] For example, when acquiring energy consumption data and timestamps, data can be extracted from smart meters or industrial IoT devices. Suppose a factory's meters record energy consumption hourly, with data including electricity consumption, voltage, and power factor, and timestamps accurate to the minute. When generating a time-series dataset, the data can be organized by day or hour, ensuring each data point corresponds to a timestamp, forming structured energy consumption data.

[0075] For example, data for a particular day may show that energy consumption is higher from 8 a.m. to 10 a.m. and lower from 3 p.m. to 5 p.m., reflecting peak and trough production.

[0076] In one possible implementation, when preprocessing structured energy consumption data, the Z-score normalization method is used to convert the electricity consumption and power factor of different dimensions into dimensionless data with a mean of 0 and a standard deviation of 1.

[0077] For example, if a device consumes 500 kWh of electricity with a power factor of 0.9, its standardized data allows for comparison of trends across different dimensions. This normalized energy consumption data facilitates subsequent analysis and avoids interference from dimensional differences in clustering results.

[0078] Specifically, when using the K-means algorithm for cluster analysis, the initial K value can be set according to the time characteristics of the energy consumption data.

[0079] For example, suppose a day is divided into three periods: peak, flat, and trough, with an initial K value of 3. After clustering, the data points are divided into three groups, corresponding to the morning peak, the midday flat period, and the nighttime trough. If the intra-cluster variance is too large, such as energy consumption fluctuations exceeding 30% within a cluster, K can be adjusted to 4, and the clustering can be re-implemented to further subdivide the evening secondary peak period. After optimizing the clustering results, the time range of each cluster can be extracted, such as the peak period being 8:00-11:00, with energy consumption characteristics of high electricity consumption and low power factor.

[0080] For example, when calculating energy consumption statistics for different time periods, the average electricity consumption during peak hours is 600 kWh with a standard deviation of 50 kWh, reflecting the concentration of energy consumption during peak hours. The average electricity consumption during off-peak hours is 200 kWh with a standard deviation of 20 kWh, showing that energy consumption is stable during off-peak hours. These statistics form an energy consumption distribution pattern, clearly demonstrating the electricity consumption characteristics of different time periods.

[0081] In one possible implementation, visualization tools such as line charts are used to plot time-series energy consumption curves, with the time axis representing 24 hours and the vertical axis representing normalized energy consumption values.

[0082] For example, the curve shows that energy consumption rises rapidly from 8:00 to 11:00, remains stable from 14:00 to 16:00, and declines after 22:00. This peak-valley energy consumption distribution map visually presents energy consumption changes and makes it easy to identify electricity consumption patterns.

[0083] It should be noted that peak-valley time periods can be used to optimize electricity dispatching.

[0084] For example, factories can adjust their production plans based on the distribution map, scheduling high-energy-consuming equipment to operate during off-peak hours to reduce costs. This analytical approach, driven by data, clearly reveals energy consumption characteristics and improves energy management efficiency.

[0085] Step S105: Based on the energy consumption distribution during peak and valley periods, a time series analysis algorithm is used to generate a trend prediction model that includes seasonal variations, and the energy consumption trend prediction results are obtained.

[0086] Historical energy consumption data is acquired and segmented according to peak and valley periods to obtain the energy consumption distribution during these periods. Based on this distribution, a time series decomposition method is used to extract seasonal variations and long-term trends, resulting in decomposed time series components. These components are then used to train a model using the ARIMA algorithm, generating a prediction model that incorporates seasonal variations. If the residuals of the prediction model exceed a preset threshold, the model parameters are adjusted, and the model is retrained to obtain an optimized prediction model. Based on this optimized model, recent energy consumption data is input to generate future energy consumption trend predictions. By comparing the prediction results with historical data, the prediction error is calculated, yielding error analysis results. Based on the error analysis results, a sliding window method is used to update the prediction model, resulting in a dynamically adjusted energy consumption trend prediction model.

[0087] For example, acquiring historical energy consumption data and segmenting it is the basis for analyzing energy consumption distribution during peak and off-peak periods.

[0088] For example, in an industrial park, electricity consumption data from the past year is collected, including hourly consumption and timestamps. Data observation reveals higher consumption between 9:00 AM and 11:00 AM and between 2:00 PM and 4:00 PM on weekdays, and lower consumption between midnight and 6:00 AM. These periods can be initially labeled as peak and off-peak times. During segmented processing, the data is grouped by hour, and the average energy consumption for each period is calculated. The average peak consumption is found to be 5000 kWh, and the average off-peak consumption is 1500 kWh, forming a preliminary energy consumption distribution. Time series decomposition methods can extract seasonal variations and long-term trends.

[0089] Specifically, the classic additive decomposition method is used to split energy consumption data into three parts: trend, seasonality, and residual.

[0090] For example, the decomposition revealed a clear daily periodicity in weekday energy consumption, with stable peak and off-peak periods each day. Long-term trends showed slightly higher energy consumption in summer, possibly related to increased air conditioning use. This decomposition helps understand the periodic patterns of energy consumption, providing clear input for subsequent modeling. The ARIMA algorithm was used to train a predictive model incorporating seasonal variations.

[0091] In one embodiment, the ARIMA(1,1,1)(1,1,1,24) model is selected, taking into account a 24-hour seasonal cycle. The model is trained using energy consumption data from the first six months, and the validation set is the data from the following month. After training, the model can capture the fluctuations during peak and trough periods relatively well.

[0092] For example, the predicted peak energy consumption at 9 AM on a certain workday is 5100 kWh, which is close to the actual value of 5050 kWh. If the model residual is large, for example, the prediction error exceeds 10%, the parameters are adjusted, such as increasing the AR or MA order, and the model is retrained to improve accuracy. The optimized prediction model is then used to generate energy consumption trend predictions for future periods.

[0093] For example, by inputting energy consumption data from the past week, the model predicts that peak-hour energy consumption will remain stable at 4800 to 5200 kWh for the next 7 days, while off-peak consumption will be between 1400 and 1600 kWh. This prediction can help the industrial park rationally plan its electricity usage. Prediction error analysis calculates the root mean square error by comparing the predicted and actual values.

[0094] For example, a prediction error of 5% indicates high model reliability. The sliding window method is used to dynamically update the prediction model.

[0095] For example, obtain the latest energy consumption data weekly, update the training dataset, and refit the ARIMA model. This approach ensures that the model adapts to recent changes in energy consumption patterns, such as energy savings due to equipment upgrades.

[0096] For example, after adding high-efficiency equipment in a certain month, the energy consumption during off-peak hours dropped to 1300 kWh. The model captures this change in a timely manner through a sliding window to maintain prediction accuracy.

[0097] It should be noted that dynamically adjusted models are better able to adapt to short-term fluctuations and long-term trends in energy consumption.

[0098] For example, during holiday periods, peak hours may disappear, and the model can adjust its predictions promptly through sliding window updates. This approach improves the robustness of predictions, ensuring reliable energy consumption trend analysis across different scenarios.

[0099] Step S106: Based on the energy consumption trend prediction results, generate a hierarchical analysis report, including peak and valley period analysis and seasonal change adjustment suggestions, to obtain structured report data.

[0100] Energy consumption data is acquired and preprocessed to extract features from the time series, resulting in a cleaned dataset. Time series analysis is then performed using the ARIMA model to predict energy consumption trends, yielding forecast results. If the forecasts show periodic fluctuations, peak and trough periods are divided based on timestamps to obtain time-period distribution. The correlation between energy consumption patterns and seasonal characteristics is analyzed within this time-period distribution to identify seasonal variation patterns. Based on these seasonal patterns, adjustment suggestions are generated, resulting in a set of optimization strategies. The forecast results, time-period distribution, and optimization strategy set are integrated using a structured data format to produce an analysis report. Finally, the analysis report data is converted into a visualization format using a pre-defined template, resulting in a structured analysis report.

[0101] For example, energy consumption data is acquired and preprocessed to extract features from the time series, resulting in a cleaned dataset.

[0102] For example, in the energy management scenario of a power company, historical energy consumption data may come from smart meters, containing hourly electricity consumption, timestamps, and weather information. During preprocessing, outliers are first removed. For instance, if electricity consumption suddenly increases to 10,000 kWh in a certain hour, far exceeding the normal range of 1,000-2,000 kWh, it may be due to equipment malfunction and needs to be deleted or interpolated to complete the data. Next, features such as average daily energy consumption, differences between weekends and weekdays, and holiday identifiers are extracted to generate a clean dataset containing time-series features, providing a reliable foundation for subsequent analysis. Through time-series analysis, the ARIMA model is used to predict the trend of energy consumption data, yielding the prediction results.

[0103] Specifically, the ARIMA model uses historical energy consumption trends and periodicity to predict electricity consumption for the coming week.

[0104] For example, based on data from the past year, the model identifies the peak daily electricity consumption during the summer as around 2 PM, at approximately 1500 kWh. The prediction shows an average daily electricity consumption of 1200 kWh for the next 7 days, with an error range of ±5%. This prediction provides a reference for power dispatch and optimizes resource allocation. If the prediction shows periodic fluctuations, peak and off-peak periods are divided according to timestamps to obtain the time distribution.

[0105] For example, analysis of forecast data reveals that daily electricity consumption peaks between 8:00 AM and 11:00 AM and between 1:00 PM and 4:00 PM, at 1400 kWh and 1600 kWh respectively, while the low point is between 1:00 AM and 4:00 AM, at only 500 kWh. After this division, peak periods are 8:00-11:00 AM and 1:00-4:00 PM, and valley periods are 0:00-4:00 AM. This distribution clearly reflects electricity consumption patterns, facilitating the development of time-of-use pricing strategies. Furthermore, analyzing the correlation between energy consumption patterns and seasonal characteristics based on the time-of-use distribution reveals seasonal variation patterns.

[0106] For example, electricity consumption during peak summer hours is 20% higher than in winter due to frequent air conditioning use. Analysis reveals that daily peak consumption occurs during hot summer days, with peak consumption increasing by approximately 50 kWh for every 1°C increase in temperature. In winter, electricity consumption increases slightly during off-peak hours at night due to heating demand. This pattern helps in precisely adjusting power supply strategies. Based on seasonal variations, adjustment suggestions are generated, resulting in a set of optimized strategies.

[0107] Specifically, in response to peak electricity demand in the summer, it is recommended to increase electricity prices by 10% during peak hours to encourage users to use electricity during off-peak periods; at the same time, discounts should be increased during off-peak hours to encourage the use of charging devices at night.

[0108] For example, electric vehicle charging stations can offer a 20% discount in the early morning to reduce grid pressure. These strategies, when combined, provide guidance for energy conservation and emission reduction. Using a structured data format, the forecast results, time-period distribution, and optimization strategy set are integrated to obtain the analytical report data.

[0109] For example, predicted daily electricity consumption, peak-valley time divisions, and electricity price adjustment suggestions can be stored in JSON format, including fields such as "Date," "Peak Electricity Consumption," "Valuation Electricity Consumption," and "Suggested Electricity Price." This structured data facilitates automatic system processing and cross-departmental sharing. Using preset templates, the analysis report data can be converted into a visual format to obtain a structured analysis report.

[0110] For example, bar charts are used to display daily peak and off-peak electricity consumption comparisons, line charts present the predicted trend for the next 7 days, and tables list electricity price adjustment recommendations. The report is output in PDF format, including charts and text descriptions, intuitively presenting energy consumption patterns and facilitating management decision-making. This visualization method enhances the report's readability and usability.

[0111] Example 2:

[0112] This invention provides an enterprise / institutional unit energy consumption management system based on a B / S architecture, comprising:

[0113] The data acquisition module is used to obtain raw energy consumption datasets by collecting energy consumption data in the dimensions of time, space, and intensity.

[0114] The hierarchical processing module is used to construct a multi-layered data structure containing time, space, and intensity dimensions based on the original energy consumption dataset using hierarchical data processing methods, thereby obtaining a hierarchical energy consumption dataset.

[0115] The clustering analysis module is used to determine the correlation patterns between different dimensions and obtain the energy consumption change patterns for hierarchical energy consumption datasets using clustering analysis algorithms.

[0116] The peak-valley classification module is used to classify the peak-valley data based on the K-means algorithm if the energy consumption change pattern shows the difference between peak and valley periods in the time dimension, so as to obtain the energy consumption distribution of peak and valley periods.

[0117] The trend prediction module is used to generate a trend prediction model that includes seasonal variations by using time series analysis algorithms based on the energy consumption distribution during peak and off-peak periods, and to obtain the energy consumption trend prediction results.

[0118] The report generation module is used to generate hierarchical analysis reports based on energy consumption trend forecasts, including peak and off-peak period analysis and seasonal adjustment suggestions, resulting in structured report data.

[0119] The above embodiments are merely one of the preferred embodiments of the present invention and should not be used to limit the scope of protection of the present invention. Any modifications or refinements made to the main design concept and spirit of the present invention that are not of substantial significance, but solve the same technical problem as the present invention, should be included within the scope of protection of the present invention.

Claims

1. A method for energy consumption management of enterprise units based on B / S architecture, characterized in that, include: Obtain the raw energy consumption dataset; Based on the original energy consumption dataset, a hierarchical energy consumption dataset is obtained; For the hierarchical energy consumption dataset, the energy consumption variation pattern was obtained; If the energy consumption variation pattern shows peak and valley time differences in the time dimension, then the peak and valley time data are classified based on the K-means algorithm to obtain the peak and valley time energy consumption distribution. By analyzing the energy consumption distribution during peak and off-peak periods, we can obtain energy consumption trend prediction results. Based on the energy consumption trend forecast results, a stratified analysis report is generated, which includes peak and valley period analysis and seasonal change adjustment suggestions, resulting in structured report data.

2. The enterprise / institutional energy consumption management method based on B / S architecture as described in claim 1, characterized in that, The raw energy consumption dataset is obtained by collecting energy consumption data in terms of time, space, and intensity.

3. The enterprise / institutional unit energy consumption management method based on B / S architecture as described in claim 2, characterized in that, Based on the original energy consumption dataset, a hierarchical data processing method is used to construct a multi-layered data structure containing time, space, and intensity dimensions, resulting in a hierarchical energy consumption dataset.

4. The enterprise / institutional unit energy consumption management method based on B / S architecture as described in claim 3, characterized in that, For the hierarchical energy consumption dataset, a clustering analysis algorithm is used to determine the correlation patterns between each dimension and obtain the energy consumption change patterns.

5. The enterprise / institutional energy consumption management method based on B / S architecture as described in claim 4, characterized in that, By analyzing energy consumption distribution during peak and off-peak periods, a time series analysis algorithm is used to generate a trend prediction model that incorporates seasonal variations, thus obtaining energy consumption trend prediction results.

6. An enterprise / institutional unit energy consumption management system based on a B / S architecture, characterized in that, include: The data acquisition module is used to acquire raw energy consumption datasets; The hierarchical processing module is used to obtain hierarchical energy consumption datasets based on the original energy consumption datasets; The clustering analysis module is used to obtain the energy consumption change patterns for hierarchical energy consumption datasets; The peak-valley classification module is used to classify the peak-valley data based on the K-means algorithm if the energy consumption change pattern shows the difference between peak and valley periods in the time dimension, so as to obtain the energy consumption distribution of peak and valley periods. The trend prediction module is used to obtain energy consumption trend prediction results based on the energy consumption distribution during peak and off-peak periods; The report generation module is used to generate hierarchical analysis reports based on energy consumption trend forecasts, including peak and off-peak period analysis and seasonal adjustment suggestions, resulting in structured report data.

7. The enterprise / institutional unit energy management system based on B / S architecture as described in claim 6, characterized in that, The data acquisition module collects energy consumption data in terms of time, space, and intensity to obtain the raw energy consumption dataset.

8. The enterprise / institutional unit energy consumption management system based on B / S architecture as described in claim 7, characterized in that, The hierarchical processing module uses a hierarchical data processing method to construct a multi-layered data structure containing time, space, and intensity dimensions based on the original energy consumption dataset, thus obtaining a hierarchical energy consumption dataset.

9. The enterprise / institutional unit energy consumption management system based on B / S architecture as described in claim 8, characterized in that, The clustering analysis module uses a clustering analysis algorithm to determine the correlation patterns between different dimensions of the hierarchical energy consumption dataset, thereby obtaining the energy consumption change patterns.

10. The enterprise / institutional energy management system based on B / S architecture as described in claim 9, characterized in that, The trend prediction module uses time series analysis algorithms to generate a trend prediction model that includes seasonal variations by analyzing the energy consumption distribution during peak and off-peak periods, thus obtaining the energy consumption trend prediction results.

Images