Intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction
Through the building with established prediction models, energy consumption prediction is carried out with new buildings with strong correlation, and the coordination function and machine learning model are used to solve the problem of insufficient data in the new building, and accurate energy consumption prediction and optimized scheduling are achieved.
Patent Information
- Application Number
- CN202510469957.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-15
- Publication Date
- 2025-08-15
AI Technical Summary
The lack of sufficient data on new buildings or new low-carbon equipment makes it difficult to effectively predict energy consumption, affecting the optimization and scheduling of the energy management system.
Through the building with a predictive model, energy consumption prediction is carried out with its highly correlated new building, and the coordination function and machine learning model are used to combine real-time and delay correlation data to obtain the target prediction function and reduce the prediction error.
When new building data is insufficient, accurate energy consumption prediction is provided, which improves prediction accuracy, provides an optimized scheduling basis for energy management systems, and reduces the probability of prediction errors.
Smart Images
Figure CN120493048A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of energy consumption prediction, and in particular to an intelligent optimization low-carbon system based on multi-correlated factor energy consumption prediction. Background Art
[0002] Building energy consumption forecasting has long been a key area of low-carbon management. Currently, building energy consumption forecasting typically utilizes big data, analyzing historical energy consumption data, building operating parameters, and external environmental factors to develop mathematical models to estimate future energy demand. This approach typically relies on real-time data collected by smart meters, such as electricity, gas, and heating and cooling loads. This data is combined with meteorological information (such as temperature, humidity, and sunlight intensity), building usage characteristics (such as foot traffic and equipment operation schedules), and calendar variables (weekdays / holidays). Multi-dimensional modeling is performed using time series analysis (such as ARIMA), machine learning (such as random forests and LSTM neural networks), or deep learning algorithms. By analyzing the cyclical, trending, and random characteristics of energy consumption, forecasts are generated, providing a basis for optimizing scheduling in building energy management systems and achieving energy savings.
[0003] The problem is that training the prediction model requires a large amount of data, but for some new buildings or buildings with newly added low-carbon equipment, there is not enough reliable data to train the prediction model. Therefore, how to predict the energy consumption of new buildings or buildings with newly added low-carbon equipment in order to more scientifically schedule the energy consumption composition of new buildings is a problem that needs to be solved.
[0004] In view of this, the present invention proposes an intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction. Through the buildings for which prediction models have been established, new buildings with strong correlation with them are predicted, and a certain energy consumption prediction function is provided in the stage when the amount of new building data is insufficient. Summary of the Invention
[0005] In order to predict new buildings that have a strong correlation with buildings based on established prediction models, and to provide a certain energy consumption prediction function when the amount of new building data is insufficient, this application provides an intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction.
[0006] In the first aspect, the present application provides an intelligent optimization low-carbon system based on multi-correlated factor energy consumption prediction, which adopts the following technical solutions:
[0007] An intelligent optimization low-carbon system based on multi-correlated factor energy consumption prediction, including:
[0008] An energy consumption monitoring module, which monitors the energy consumption of Building 1 and Building 2 and obtains monitoring data;
[0009] The correlation analysis module, the data processing module obtains the coordination function based on the monitoring data of the building 1 and the building 2;
[0010] An energy consumption prediction module, comprising a preset first prediction model for predicting energy consumption of building one. The energy consumption prediction module obtains a target prediction function for building two based on the coordination function and the first prediction model. The target prediction function is used to predict the energy consumption of building two to obtain a target result. The training process of the target prediction function adopts machine learning, which can be a random forest. The specific process is not described in detail here.
[0011] A management module is used to set an optimization plan based on the target result obtained by prediction.
[0012] Through the above technical solution: a technical solution is provided for predicting building two based on the prediction results of building one, and predictions are made for new buildings with strong correlation with the buildings for which prediction models have been established. In the stage where the amount of new building data is insufficient, a certain energy consumption prediction function is provided. The target prediction function of the present invention is obtained based on the correlation data of building one and building two, and can analyze the data in building two that is highly correlated with building one, thereby achieving a certain degree of data prediction when the amount of operating data of building two is small, and providing an optimization scheduling basis for the building energy management system.
[0013] Optionally, the feature is that the energy consumption data of building 1 and building 2 are divided into and Energy consumption units, obtain energy consumption data within n t time periods, the coordination function of an energy consumption unit of the building 2 The acquisition process includes:
[0014] For the dataset Perform screening, obtain the maximum value max and minimum value min and eliminate them based on the maximum and minimum values The target set is obtained by taking the data and is the energy consumption value of an energy consumption unit in building 2 during time t, is the energy consumption value of an energy consumption unit in building 1 during the nth time t;
[0015] Sum the target data and The smallest value among the summed values is selected as the synchronization coordination coefficient;
[0016] Obtain the probability of asynchronous coordination of the same type of energy consumption units in building 2, and collect the corresponding data sets based on the synchronous coordination data. The maximum value max and the probability of asynchronous coordination are used to obtain the asynchronous coordination coefficient;
[0017] The coordination function is obtained by performing weighted summation on the synchronous coordination coefficient and the asynchronous coordination coefficient.
[0018] Through the above technical solution: a process for obtaining a coordination function is provided. In the present invention, the coordination function is obtained based on the synchronous coordination coefficient and the asynchronous coordination coefficient. This is because the power association process includes two types: instant association and delayed association. The instant association can generate data feedback in a timely manner, while the delayed association requires a certain amount of time to generate data feedback. When making predictions through the data association of Building 1 and Building 2, if only the instant association exists, the prediction can be made very simply. However, the existence of the delayed association will cause the results predicted by the instant association to be prone to errors. The present invention reduces the probability of errors in the prediction results and improves the accuracy of the prediction results by considering the two in combination.
[0019] Optionally, the feature is that the maximum and minimum values are obtained and the The process of obtaining the target set from the data includes:
[0020] In the data set, the value range is and The data within the range is eliminated, and the number of eliminations is recorded as ;
[0021] in , , is the preset rejection index.
[0022] Optionally, the coordination function is obtained by performing weighted summation on the synchronous coordination coefficient and the asynchronous coordination coefficient. The process involves, through the formula: ; ; ;
[0023] Get coordination function ,in, and is the preset weight coefficient, =n- , Indicates the first Energy consumption data within time t, is not greater than A non-zero positive integer, is the probability of asynchronous coordination of the same type of energy consumption units in building 2, It is a dataset The maximum value of is the synchronization coordination coefficient, is the asynchronous coordination coefficient.
[0024] Optionally, the calculation process of the target prediction function Fr includes:
[0025] By formula Get the predicted value;
[0026] in is the predicted value of a certain energy consumption unit of a building at a certain time t by using a preset first prediction model;
[0027] is the coordination function of a certain energy consumption unit in building 2, It is a preset judgment function used to compare the number of remaining data in the data set after the elimination process with the preset number Compare, if the number of remaining data in the data set after the elimination process is less than The output is 0, otherwise the output is 1. It is the basic value set based on the current energy consumption unit type. .
[0028] Through the above technical solution: a calculation process of the target prediction function Fr is provided. Specifically, the present invention obtains the predicted value of Building 2 through the numerical value of the coordination function and the predicted value of the first prediction model. The coordination function of the energy consumption unit corresponding to Building 2 has a unique data set, and the source object of the data set is the energy consumption unit of Building 1. After the two establish a comparative relationship, the coordination function takes the occurrence of asynchronous coordination into consideration, that is, the larger the numerical value of the coordination function, the greater the possibility of asynchronous coordination occurring in Building 2 for which prediction data is required. At this time, the energy consumption of the energy consumption unit of Building 2 cannot be predicted through the target prediction function. On the contrary, when the numerical value of the coordination function meets the preset conditions, it indicates that the probability of asynchronous coordination occurring in Building 2 in the future is small. Data prediction at this time can reduce the probability of prediction errors and improve prediction accuracy.
[0029] Optionally, the process of predicting the energy consumption of Building 2 to obtain a target result includes:
[0030] like If the output is 0, the target result cannot be predicted;
[0031] like If the output is not 0, the output value is used as the target result.
[0032] Optionally, the process of setting an optimization solution based on the predicted target result includes:
[0033] Determine the energy consumption unit corresponding to the prediction result and obtain the alternative energy consumption equipment information of the energy consumption unit;
[0034] If there is an alternative energy-consuming device, the alternative energy-consuming device is started in advance before the predicted time t arrives.
[0035] Optionally, a fault analysis module compares the predicted value with the actual value and issues an early warning when the difference between the predicted value and the actual value exceeds a critical range, wherein the early warning content is the energy consumption unit corresponding to the predicted target.
[0036] Optionally, the startup of the alternative energy-consuming device is controlled by constructing an optimal control model based on the startup and shutdown process of the corresponding energy-consuming unit and the startup and shutdown process of the alternative energy-consuming device.
[0037] The process of obtaining the target prediction function of building 2 based on the coordination function and the first prediction model includes:
[0038] Compare the value of the coordination function of a certain energy consumption unit of Building 2 with a preset confidence interval, where the endpoints of the preset confidence interval are set based on empirical data;
[0039] If the value of the coordination function falls within the confidence interval, the target prediction function Fr is calculated;
[0040] If the coordination function is outside the confidence interval, it is judged that prediction cannot be made and the target prediction function Fr=0 is output.
[0041] In summary, this application includes at least one of the following beneficial technical effects:
[0042] The present invention predicts new buildings that have a strong correlation with the buildings for which prediction models have been established, and provides a certain energy consumption prediction function when the amount of new building data is insufficient. The target prediction function of the present invention is obtained based on the correlation data of Building 1 and Building 2, and can analyze the data in Building 2 that is highly correlated with Building 1, thereby achieving a certain degree of data prediction when the amount of operating data of Building 2 is small, providing an optimization scheduling basis for the building energy management system.
[0043] The coordination function in the present invention is obtained based on the synchronous coordination coefficient and the asynchronous coordination coefficient because the power association process includes two types: instant association and delayed association. The instant association can generate data feedback in a timely manner, while the delayed association requires a certain amount of time to generate data feedback. When making predictions through the data association of Building 1 and Building 2, if only the instant association exists, the prediction can be made very simply. However, the existence of the delayed association will cause the results predicted by the instant association to be prone to errors. The present invention reduces the probability of errors in the prediction results and improves the accuracy of the prediction results by combining the two.
[0044] The present invention obtains the predicted value of Building 2 through the numerical value of the coordination function and the predicted value of the first prediction model. The coordination function of the energy consumption unit corresponding to Building 2 has a unique data set, and the source object of the data set is the energy consumption unit of Building 1. After the two establish a comparative relationship, the coordination function takes the occurrence of asynchronous coordination into consideration. That is, the larger the numerical value of the coordination function, the greater the possibility of asynchronous coordination occurring in Building 2 for which prediction data is required. At this time, the energy consumption of the energy consumption unit of Building 2 cannot be predicted through the target prediction function. On the contrary, when the numerical value of the coordination function meets the preset conditions, it indicates that the probability of asynchronous coordination occurring in Building 2 in the future is small. Data prediction at this time can reduce the probability of prediction errors and improve prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a schematic diagram of the module composition of the present invention.
[0046] Figure 2 It is a flow chart of the coordination function acquisition process. DETAILED DESCRIPTION
[0047] Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings.
[0048] Throughout this specification, reference to the terms "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
[0049] The present application embodiment discloses an intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction, referring to Figure 1 , including an energy consumption monitoring module, the monitoring module monitors the energy consumption of building 1 and building 2 and obtains monitoring data;
[0050] The correlation analysis module and the data processing module obtain the coordination function based on the monitoring data of building one and building two;
[0051] An energy consumption prediction module includes a preset first prediction model for predicting the energy consumption of building one. The energy consumption prediction module obtains a target prediction function for building two based on the coordination function and the first prediction model. The target prediction function is used to predict the energy consumption of building two and obtain a target result. The training process of the target prediction function adopts machine learning, which can be a random forest. The specific process is not described in detail here.
[0052] The management module sets the optimization plan based on the target results obtained by prediction.
[0053] In this embodiment, a technical solution is provided for predicting building two based on the prediction results of building one. Through the building for which the prediction model has been established, predictions are made for new buildings that are highly correlated with it. In the stage when the amount of new building data is insufficient, a certain energy consumption prediction function is provided. The target prediction function of the present invention is obtained based on the associated data of building one and building two, and can analyze the data in building two that is highly correlated with building one, thereby achieving a certain degree of data prediction when the amount of operating data of building two is small, providing an optimization scheduling basis for the building energy management system.
[0054] It is characterized by dividing the energy consumption data of Building 1 and Building 2 into and Energy consumption units, obtain energy consumption data for n t time periods, and the coordination function of an energy consumption unit in building 2 The acquisition process includes:
[0055] For the dataset Perform screening, obtain the maximum value max and minimum value min and eliminate them based on the maximum and minimum values The target set is obtained by taking the data and is the energy consumption value of an energy consumption unit in building 2 during time t, is the energy consumption value of an energy consumption unit in building 1 during the nth time t;
[0056] Sum the target data and The smallest value among the summed values is selected as the synchronization coordination coefficient;
[0057] Obtain the probability of asynchronous coordination of the same type of energy consumption units in building 2, and collect the corresponding data sets based on the synchronous coordination data. The maximum value max and the probability of asynchronous coordination are used to obtain the asynchronous coordination coefficient;
[0058] The coordination function is obtained by performing weighted summation on the synchronous coordination coefficient and the asynchronous coordination coefficient.
[0059] In this embodiment, a process for obtaining a coordination function is provided. In the present invention, the coordination function is obtained based on the synchronous coordination coefficient and the asynchronous coordination coefficient because the power association process includes two types: instant association and delayed association. Instant association can generate data feedback in a timely manner, while delayed association requires a certain amount of time to generate data feedback. When making predictions through the data association of Building 1 and Building 2, if only instant association exists, predictions can be made very simply. However, the existence of delayed association will cause the results predicted by instant association to be prone to errors. The present invention reduces the probability of errors in the prediction results and improves the accuracy of the prediction results by combining the two.
[0060] It is characterized by obtaining the maximum and minimum values and eliminating them based on the maximum and minimum values The process of obtaining the target set from the data includes:
[0061] In the data set, the value range is and The data within the range is eliminated, and the number of eliminations is recorded as ;
[0062] in , , is the preset rejection index, which is a constant.
[0063] The coordination function is obtained by weighted summing the synchronous coordination coefficient and the asynchronous coordination coefficient. The process involves, through the formula: ; ; ;
[0064] Get coordination function ,in, and is the preset weight coefficient, which is set based on historical data and is a constant. is the number of data in the target set, =n- , Indicates the first Energy consumption data within time t, is not greater than A non-zero positive integer, is the probability of asynchronous coordination of the same type of energy consumption units in building 2, It is a dataset The maximum value of is the synchronization coordination coefficient, is the asynchronous coordination coefficient.
[0065] The calculation process of the target prediction function Fr includes:
[0066] By formula Get the predicted value;
[0067] in is the predicted value of a certain energy consumption unit of a building at a certain time t by using a preset first prediction model;
[0068] is the coordination function of a certain energy consumption unit in building 2, It is a preset judgment function used to compare the number of remaining data in the data set after the elimination process with the preset number Compare, if the number of remaining data in the data set after the elimination process is less than The output is 0, otherwise the output is 1. It is the basic value set based on the current energy consumption unit type. , is a constant.
[0069] In this embodiment, a calculation process of the target prediction function Fr is provided. Specifically, the present invention obtains the predicted value of Building 2 through the numerical value of the coordination function and the predicted value of the first prediction model. The coordination function of the energy consumption unit corresponding to Building 2 has a unique data set, and the source object of the data set is the energy consumption unit of Building 1. After the two establish a comparative relationship, the coordination function takes the occurrence of asynchronous coordination into consideration. That is, the larger the numerical value of the coordination function, the greater the possibility of asynchronous coordination occurring in Building 2 for which prediction data is required. At this time, the energy consumption of the energy consumption unit of Building 2 cannot be predicted through the target prediction function. On the contrary, when the numerical value of the coordination function meets the preset conditions, it indicates that the probability of asynchronous coordination occurring in Building 2 in the future is small. Data prediction at this time can reduce the probability of prediction errors and improve prediction accuracy.
[0070] The process of predicting the energy consumption of Building 2 and obtaining the target results includes:
[0071] like If the output is 0, the target result cannot be predicted;
[0072] like If the output is not 0, the output value is used as the target result.
[0073] The process of setting up an optimization plan based on the target results obtained by prediction includes:
[0074] Determine the energy consumption unit corresponding to the prediction result and obtain the alternative energy consumption equipment information of the energy consumption unit;
[0075] If there is an alternative energy-consuming device, the alternative energy-consuming device is started in advance before the predicted time t arrives.
[0076] Fault analysis module, the fault analysis module compares the predicted value with the actual value, and issues an early warning when the difference between the predicted value and the actual value exceeds the critical range. The early warning content is the corresponding energy consumption unit of the predicted target. The critical range is a preset value based on empirical data and is a constant.
[0077] The startup of alternative energy-consuming equipment is controlled by constructing an optimal control model based on the start-stop process of the corresponding energy-consuming unit and the start-stop process of the alternative energy-consuming equipment. The optimal control model is established through the optimal system operating parameters or timetable to ensure that the start-stop consumption is minimized. For example, when it is predicted that the electricity price will rise during the peak electricity consumption period at time t, the management module automatically starts the ice storage device one hour in advance to make ice with low-valley electricity, and at the same time delays the operation of the fresh air unit in non-essential areas; or when the temperature needs to be raised, if there will be sufficient solar energy supply at noon, the electric curtains will be started first to adjust the shading rate to reduce the cooling load, and the high-efficiency magnetic levitation chiller will be started in conjunction to replace the traditional centrifugal unit.
[0078] The process of obtaining the target prediction function of building 2 based on the coordination function and the first prediction model includes:
[0079] Compare the value of the coordination function of a certain energy consumption unit of Building 2 with a preset confidence interval, where the endpoints of the preset confidence interval are set based on empirical data;
[0080] If the value of the coordination function falls within the confidence interval, the target prediction function Fr is calculated;
[0081] If the coordination function is outside the confidence interval, it is judged that prediction cannot be made and the target prediction function Fr=0 is output.
[0082] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. An intelligent optimization low-carbon system based on multi-correlated factor energy consumption prediction, characterized in that: include: An energy consumption monitoring module, which monitors the energy consumption of Building 1 and Building 2 and obtains monitoring data; The correlation analysis module, the data processing module obtains the coordination function based on the monitoring data of the building 1 and the building 2; An energy consumption prediction module, wherein the energy consumption prediction module includes a preset first prediction model for predicting energy consumption of building one, and the energy consumption prediction module obtains a target prediction function for building two based on the coordination function and the first prediction model, and the target prediction function is used to predict the energy consumption of building two to obtain a target result; A management module is used to set an optimization plan based on the target result obtained by prediction.
2. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The energy consumption data of Building 1 and Building 2 are divided into and Energy consumption units, obtain energy consumption data within n t time periods, the coordination function of an energy consumption unit of the building 2 The acquisition process includes: For the dataset Perform screening, obtain the maximum value max and minimum value min and eliminate them based on the maximum and minimum values The target set is obtained by taking the data and is the energy consumption value of an energy consumption unit in building 2 during time t, is the energy consumption value of an energy consumption unit in building 1 during the nth time t; Sum the target data and The smallest value among the summed values is selected as the synchronization coordination coefficient; Obtain the probability of asynchronous coordination of the same type of energy consumption units in building 2, and collect the corresponding data sets based on the synchronous coordination data. The maximum value max and the probability of asynchronous coordination are used to obtain the asynchronous coordination coefficient; The coordination function is obtained by performing weighted summation on the synchronous coordination coefficient and the asynchronous coordination coefficient.
3. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 2 is characterized in that: Get the maximum and minimum values and remove them based on the maximum and minimum values The process of obtaining the target set from the data includes: In the data set, the value range is and The data within the range is eliminated, and the number of eliminations is recorded as ; in , , is the preset rejection index.
4. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 2 is characterized in that: The coordination function is obtained by weighted summing the synchronous coordination coefficient and the asynchronous coordination coefficient. The process involves, through the formula: ； ； ； Get coordination function ,in, and is the preset weight coefficient, =n- , Indicates the first Energy consumption data within time t, is not greater than A non-zero positive integer, is the probability of asynchronous coordination of the same type of energy consumption units in building 2, It is a dataset The maximum value of is the synchronization coordination coefficient, is the asynchronous coordination coefficient.
5. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The calculation process of the target prediction function Fr includes: By formula Get the predicted value; in is the predicted value of a certain energy consumption unit of a building at a certain time t by using a preset first prediction model; is the coordination function of a certain energy consumption unit in building 2, It is a preset judgment function used to compare the number of remaining data in the data set after the elimination process with the preset number Compare, otherwise the output is 1, It is the basic value set based on the current energy consumption unit type. .
6. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The process of predicting the energy consumption of Building 2 and obtaining the target results includes: like If the output is 0, the target result cannot be predicted; like If the output is not 0, the output value is used as the target result.
7. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The process of setting up an optimization plan based on the target results obtained by prediction includes: Determine the energy consumption unit corresponding to the prediction result and obtain the alternative energy consumption equipment information of the energy consumption unit; If there is an alternative energy-consuming device, the alternative energy-consuming device is started in advance before the predicted time t arrives.
8. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The fault analysis module compares the predicted value with the actual value and issues an early warning when the difference between the predicted value and the actual value exceeds a critical range. The early warning content is the energy consumption unit corresponding to the predicted target.
9. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 7 is characterized in that: The startup of the alternative energy consuming device is controlled by constructing an optimal control model based on the startup and shutdown process of the corresponding energy consuming unit and the startup and shutdown process of the alternative energy consuming device.
10. The intelligent optimization low-carbon system based on multi-correlation factor energy consumption prediction according to claim 1 is characterized in that: The process of obtaining the target prediction function of building 2 based on the coordination function and the first prediction model includes: Compare the value of the coordination function of a certain energy consumption unit of building 2 with the preset confidence interval; If the value of the coordination function falls within the confidence interval, the target prediction function Fr is calculated; If the coordination function is outside the confidence interval, it is judged that prediction cannot be made and the target prediction function Fr=0 is output.