Heating type electric load short-term prediction method based on data-knowledge fusion driving under demand response

Abstract

The invention belongs to the field of power system operation. A heating type electric load short-term prediction method based on data-knowledge fusion driving under demand response is characterized by comprising the steps that heating type electric load interaction response knowledge under demand response and other prediction influence factors serve as features, and a data-knowledge fusion driving prediction model is introduced for training; a heating type electric load prediction value which is high in prediction and close to the actual power utilization level is obtained; according to the data-knowledge fusion driving model, heat supply type electric load demand response signals, user response rate, uncertainty and other knowledge, other common prediction influence factors and the like serve as data driving model input characteristics, a data-knowledge fusion driving prediction model is established, and model training is conducted under knowledge and data dual driving. According to the method, the prediction precision can be improved, the prediction result better fits the actual power consumption level, the modeling difficulty is reduced, and the generalization and interpretability of the prediction model are improved.

Description

Technical Field

[0001] This invention belongs to the field of power system operation, and more specifically relates to a short-term forecasting method for heating-type electrical load based on data-knowledge fusion driven by demand response. Background Technology

[0002] Short-term forecasting of heating-related electricity load is of great significance for the safe and economical operation of the power system. As a crucial component of winter electricity load, fluctuations in heating electricity consumption directly affect the peak load and supply-demand balance of the power grid. By accurately forecasting future heating electricity demand, power grid dispatching departments can optimize power generation plans in advance, rationally allocate peak-shaving resources, thereby reducing the system's reserve capacity requirements, improving the utilization rate of power generation equipment, and promoting the consumption of clean energy sources such as wind power and photovoltaic power.

[0003] On the other hand, with the widespread implementation of demand response on the distribution network side, the characteristics of heating-type electrical loads are changing. Demand response, through electricity price incentives and intelligent regulation, can encourage electric heating users to shift their electricity consumption periods, thereby effectively smoothing the grid load curve and reducing the peak-to-valley load difference. This increases the complexity of heating-type electrical load characteristics. Therefore, when conducting short-term forecasting of heating-type electrical loads, it is necessary to incorporate the regulatory effect of demand response into the forecasting model to more accurately reflect the load change patterns.

[0004] In demand-response heating-related electricity load forecasting scenarios, electricity load interactions are frequent and influenced by multiple factors, including conventional factors such as climate and interactive response factors such as demand response. Furthermore, complex relationships often exist between these different influencing factors. Therefore, traditional load forecasting methods based on model-driven and data-driven forecasting have significant limitations: model-driven forecasting methods can only consider a limited number of influencing factors, are difficult to model, and have poor generalization performance; data-driven forecasting methods are highly dependent on data and lack interpretability.

[0005] Therefore, in load forecasting scenarios, considering the correlation between load and complex influencing factors, demand response, and the characteristics of multi-load synergy and complementarity are technical problems that urgently need to be solved. Summary of the Invention

[0006] The purpose of this invention is to provide a short-term forecasting method for heating-type electrical load based on data-knowledge fusion under demand response. This method can improve forecasting accuracy, make the forecast results more consistent with actual electricity consumption levels, reduce modeling difficulty, and improve the generalization and interpretability of the forecasting model.

[0007] To achieve the above objectives, the technical solution adopted by this invention is: a short-term forecasting method for heating-type electrical load based on data-knowledge fusion driven by demand response, characterized by including the following steps: By using the interactive response knowledge of heating-type electric load under demand response and other predictive influencing factors as features, a data-knowledge fusion-driven prediction model is introduced for training, resulting in a high-prediction value of heating-type electric load that is close to the actual electricity consumption level. The knowledge of interactive response to heating electricity load under demand response includes: the demand response signal extracted from the interactive response model of heating electricity load under demand response, and the user response rate and its uncertainty under different electricity price levels; Other factors affecting the forecast include: electricity price, temperature, humidity, and historical load data; The data-knowledge fusion-driven model uses knowledge such as heating-type electricity load demand response signals, user response rates and their uncertainties, along with other common predictive influencing factors, as input features to establish a data-knowledge fusion-driven prediction model, and trains the model under the dual drive of knowledge and data.

[0008] Furthermore, the demand response signal is obtained by solving the heating-type electrical load interaction response model under demand response; the heating-type electrical load interaction response model under demand response includes: optimization objective, first-order thermodynamic model of building, constraints and electrothermal conversion.

[0009] Furthermore, the optimization objective for heating-type electricity load under demand response is to maximize the profitability of demand response management side entities such as load aggregators. in The profit that load aggregators gain by reducing heating-related electrical loads; For demand response resource purchasers in t The contract incentive price for load aggregators at all times; Pricing for contracts between load aggregators and heating-type electrical load users; For heating loads t Real-time demand response signals.

[0010] Furthermore, the first-order thermodynamic model of the building is established based on the equivalent thermal parameter method of circuit simulation, which is used to establish the relationship between heat load and indoor temperature changes. in and for t +1 time and t The indoor temperature at any given time; for t The outdoor temperature at any given time; R The equivalent thermal resistance of the building; CThe specific heat capacity of indoor air; for t The heat load at time t; e is the natural constant; It is the equivalent thermal time constant in the thermodynamic model of a building, used to measure how fast the building's temperature changes; For the time interval, take =0.1h.

[0011] Furthermore, after deforming the first-order thermodynamic model of the building, the heat load before and after reduction is obtained: in, Before the reduction t The heat load at any given time; After reduction t The heat load at any given time; This is to reduce the amount of heat load.

[0012] Furthermore, reducing the heating load will lead to a decrease in indoor temperature, reducing user comfort. Therefore, it is necessary to constrain the amount of temperature drop and heat load reduction. in, Let t be the indoor temperature at time t; T is a time series over a day, and we take T=24. This is the lower limit of indoor temperature; for t The maximum amount of heat load reduction at any given moment; This represents the maximum total reduction in heat load within a single day.

[0013] Furthermore, to reduce the impact of frequent reductions in heating load on users' normal lives, it is stipulated that a maximum of one continuous reduction in heating load will be implemented per day: in and The minimum and maximum duration of the reduction in heating-related electrical load within a day; and Let be variables 0 and 1 at times t and t+1, where 0 indicates that the reduction operation is not performed and 1 indicates that the reduction operation is performed.

[0014] Furthermore, the input terminal of the electric boiler, which responds to demand signals, is connected to the power distribution network node, and the output terminal is connected to the heating load, requiring energy conversion by the electric boiler. in, The electrothermal conversion coefficient of the electric boiler; A sequence of heating-type electrical load demand response signals for a single day; To reduce the amount of heat load; Let be the demand response signal of the heat load at time t, where (The remaining numbers 1, 2, T-1, T, etc. do not need to be explained in detail);

[0015] Furthermore, considering the impact of economic incentives on user electricity consumption behavior, and based on consumer psychology, an optimistic scenario is used. and pessimistic situations Describe the user response rate: In the formula, p For economic incentives; and These represent the dead-zone response rate under optimistic and pessimistic scenarios, respectively. and These represent the slopes of the response rate curves under optimistic and pessimistic conditions, respectively. a and b These are critical excitation and saturation excitation, respectively. This represents the user's saturation response rate.

[0016] Furthermore, considering the uncertainty of user response rate, we assume that the user response rate follows a normally distributed random variable within an interval with upper and lower limits under optimistic and pessimistic scenarios. ; This represents the actual user response rate, where p is the economic incentive; N indicates a normal distribution. This represents the expected value of the actual user response rate under economic incentive p; This represents the variance of the actual user response rate under economic incentive p.

[0017] Furthermore, the other predictive influencing factors include: t The system considers electricity price, temperature, and humidity at specific times; it employs an iterative load forecasting method to predict the load at the next time point based on the historical load at the previous four time points.

[0018] Furthermore, the knowledge of the interactive response of heating-type electrical loads under demand response and common predictive influencing factors are used as input features to introduce a data-driven model. The data-driven model includes machine learning models such as LSTM (Long Short-Term Memory) neural networks, BP (Backpropagation) neural networks, and SVM (Support Vector Machine).

[0019] Furthermore, the prediction evaluation indicators selected are the mean square error (RMSE) and the mean relative percentage error (MAPE): In the formula, RMSE represents the mean square error; MAPE represents the mean relative percentage error; and n represents the total number of loading samples. This represents the i-th load data; This represents the predicted value of the i-th load data. This represents the actual value of the i-th load data.

[0020] Furthermore, the proposed method is applied as follows: the trained data-knowledge fusion-driven prediction model is deployed to the distribution network side, and load demand response signals, user response rates and other prediction influencing factors are input into the model in parallel to obtain load prediction results with high prediction accuracy and more in line with actual electricity consumption levels, providing a data foundation for the formulation of power generation plans and dispatch plans.

[0021] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: This invention establishes an interactive response model for heating-type electrical loads under demand response conditions, considering the impact of complex interactive response factors on load changes in demand response scenarios, and extracts the demand response signal for heating-type electrical loads. This invention also establishes a user response rate and its uncertainty model, representing the user response rate through optimistic and pessimistic scenarios, and using a normally distributed random variable to represent the uncertainty of the user response rate under a certain economic incentive, thus more closely reflecting the actual user response level. This invention incorporates two types of load knowledge—the demand response signal for heating-type electrical loads and the user response rate and its uncertainty—along with common load influencing factors such as electricity prices, temperature and humidity, and historical load data, as features into a data-driven model, establishing a data-knowledge fusion-driven model. This improves the model's prediction accuracy and generalization ability, and reduces its dependence on data. Thus, this invention, through a data-knowledge fusion-driven method, uses demand response signals and user response rates as load knowledge, introducing them along with other influencing factors as input features into the data-driven model, improving prediction accuracy, making the prediction results more closely reflect actual electricity consumption levels, reducing modeling difficulty, and improving the generalization and interpretability of the prediction model. Attached Figure Description

[0022] Figure 1A framework diagram of a short-term heating-type electrical load forecasting method based on data-knowledge fusion driven by demand response provided in an embodiment of the present invention.

[0023] Figure 2 A demand response management model diagram provided for embodiments of the present invention.

[0024] Figure 3 This is a graph illustrating the user response rate and its uncertainty patterns, provided for embodiments of the present invention.

[0025] Figure 4 A flowchart for data-knowledge fusion-driven heating power load prediction provided in an embodiment of the present invention.

[0026] Figure 5 shows the prediction results for a certain day in the test set provided by the embodiment of the present invention; (a) is the load prediction curve of the LSTM model; (b) is the load prediction curve of the BP model; and (c) is the load prediction curve of the SVM model. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0028] See Figure 1 This invention provides a short-term forecasting method for heating-type electrical load based on data-knowledge fusion driven by demand response, comprising: By combining two types of load knowledge—heating-type electricity load demand response signal, user response rate and its uncertainty—with common predictive influencing factors such as electricity price, temperature and humidity, and historical load data as input features, the data-knowledge fusion-driven model is introduced in parallel to obtain the data-knowledge fusion-driven model. Specifically, the heating-type electricity load demand response signal refers to the optimization objective in the heating-type electricity load demand response interaction model, which aims to maximize the profitability of load aggregators while ensuring user comfort. t A 0-1 signal indicating whether to implement heating-type electricity load reduction operations at any given time. The demand response management mode includes... Figure 2 As shown The user response rate and its uncertainty are described under a certain economic incentive level. Specifically, based on consumer psychology, response curves for optimistic and pessimistic scenarios are obtained to describe user response rates. The actual user response rate is a random variable following a normal distribution within an interval bounded by the optimistic and pessimistic scenarios, thus describing the uncertainty of the user response rate. The user response rate pattern considering uncertainty is as follows: Figure 3 As shown.

[0029] The data-knowledge fusion-driven model includes: knowledge extraction, knowledge input, and data-driven modeling. Specifically, the heating-type electrical load demand response signal is extracted based on the interactive response model of heating-type electrical load under demand response, and the actual user response rate is extracted based on the user response rate and its uncertainty model, thus completing knowledge extraction. These two types of load knowledge, along with common load forecasting influencing factors, are used as input features and input into the data-driven model in parallel, completing knowledge input. Appropriate hyperparameters are selected for model training, completing data-driven modeling. The specific forecasting process is as follows: Figure 4 As shown.

[0030] The following example uses contract data between load aggregators and users in a certain market to illustrate the above method. The dataset is divided into training and test sets with a ratio of 0.9:0.1.

[0031] In the demand response model for heating-type electrical load interaction, the equivalent thermal resistance of the building is set. R =18℃ / kW, Indoor air specific heat capacity C =0.525kWh / ℃, t Maximum heat load reduction at any time =10kW, maximum total daily heat load reduction =30kWh, lower limit of indoor temperature =15℃, minimum duration of slashing and maximum sustained reduction time The electrothermal conversion coefficient of the electric boiler is calculated for 1 hour and 4 hours respectively. =0.90.

[0032] In the user response rate and its uncertainty model, a dead-zone response rate is defined under the optimistic scenario. The dead zone response rate is 10% under pessimistic conditions. -5%, user saturation response rate 95%, critical excitation a and saturation excitation b These are 1 times and 1.25 times the contracted electricity price, respectively.

[0033] To analyze the impact of two types of knowledge—demand response signals and user response rates taking uncertainty into account—on load forecasting, and to explore the generalization of data-knowledge fusion-driven methods, three data-driven models were selected: Long Short-Term Memory (LSTM) neural network, Backpropagation (BP) neural network, and Support Vector Machine (SVM). Three forecasting scenarios were set up for each data model: Scenario 1: Predict the load at the next moment without considering demand response signals, user response rates and their uncertainties; Scenario 2: Considering demand response signals, but ignoring user response rates and their uncertainties, predict the load at the next moment; Scenario 3: Considering demand response signals, user response rates and their uncertainties, predict the load at the next moment.

[0034] The parameter settings for each data-driven model are as follows: LSTM model hyperparameters: number of hidden neurons: 50, batch size: 50, maximum number of iterations: 200, initial learning rate: 0.007, learning rate decreases to 0.0035 after 100 iterations. BP model hyperparameters: number of iterations: 200, error threshold: The learning rate is 0.01. SVM model hyperparameters: kernel type is radial basis function, penalty factor is 5.

[0035] The mean square error (RMSE) and mean relative percentage error (MAPE) are used as prediction evaluation indicators.

[0036] The prediction results for a specific day in the test set were selected for comparative analysis, as shown in Figure 5.

[0037] The entire test set was selected, and the RMSE and MAPE of each model under different prediction scenarios were calculated. The results are shown in Table 1.

[0038] Table 1. RMSE and MAPE under different forecasting scenarios for heating electrical loads In the LSTM, BP, and SVM models, incorporating heating-related electrical load demand response signals reduces prediction errors. Further reducing errors by adding user response rates that account for uncertainty. In the LSTM model, compared to scenario one, scenario two shows a reduction of RMSE of 0.5793 kW, a decrease of MAPE of 0.1036%, and a 6.06% improvement in prediction accuracy. Considering user response rates that account for uncertainty in scenario two, scenario three shows a reduction of RMSE of 0.4976 kW, a decrease of MAPE of 0.0496%, and a further 3.09% improvement in prediction accuracy compared to scenario two. The prediction analysis for the BP and SVM models follows the same logic. Under the BP model, compared to scenario one, scenario two shows a reduction of RMSE of 0.0614 kW, a decrease of MAPE of 0.1173%, and a 7.91% improvement in prediction accuracy; scenario three shows a reduction of RMSE of 0.5613 kW, a decrease of MAPE of 0.0534%, and a further 3.91% improvement in prediction accuracy compared to scenario two. Under the SVM model, compared with scenario one, scenario two saw a reduction of 0.1060kW in RMSE, a reduction of 0.0172% in MAPE, and an improvement of 1.26% in prediction accuracy; compared with scenario two, scenario three saw a reduction of 0.5225kW in RMSE, a reduction of 0.1004% in MAPE, and a further improvement of 7.46% in prediction accuracy.

[0039] Based on the above prediction results, the following conclusions can be drawn: In Scenario 2, after incorporating the knowledge of heating-type electricity load demand response signals, the load prediction error is reduced during peak and off-peak periods compared to Scenario 1; Scenario 3, based on Scenario 2, adds knowledge of user response rates that take into account uncertainty, which is closer to the actual electricity consumption behavior of users, corrects the prediction curve, and further reduces the error between the predicted load value and the actual load value.

[0040] Therefore, with the addition of load knowledge, the overall prediction error of the three data-driven models, namely LSTM neural network, BP neural network and support vector machine (SVM), all show a decreasing trend, indicating that the data-knowledge fusion-driven method has good generalization ability in heating-type electrical load prediction.

[0041] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A short-term forecasting method for heating-type electrical load based on data-knowledge fusion driven by demand response, characterized in that... Including the following steps: By using the interactive response knowledge of heating-type electric load under demand response and other predictive influencing factors as features, a data-knowledge fusion-driven prediction model is introduced for training, resulting in a high-prediction value of heating-type electric load that is close to the actual electricity consumption level. The knowledge of interactive response to heating electricity load under demand response includes: the demand response signal extracted from the interactive response model of heating electricity load under demand response, and the user response rate and its uncertainty under different electricity price levels; Other factors affecting the forecast include: electricity price, temperature, humidity, and historical load data; The data-knowledge fusion-driven model uses heating-type electricity load demand response signals, user response rates and their uncertainties, along with other common predictive influencing factors, as input features to establish a data-knowledge fusion-driven prediction model, and trains the model under the dual drive of knowledge and data.

2. The method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response as described in claim 1, characterized in that, The demand response signal is obtained by solving the heating-type electrical load interactive response model under demand response; the heating-type electrical load interactive response model under demand response includes: optimization objective, first-order thermodynamic model of building, constraints and electrothermal conversion.

3. The method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response, as described in claim 1, is characterized in that... The optimization objective for heating-type electricity load under demand response is to maximize the profit of the load aggregator's demand response management side. in The profit that load aggregators gain by reducing heating-related electrical loads; For demand response resource purchasers in t The contract incentive price for load aggregators at all times; Pricing for contracts between load aggregators and heating-type electrical load users; For heating loads t Real-time demand response signals.

4. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response, as described in claim 2, is characterized in that... The first-order thermodynamic model of the building is established based on the equivalent thermal parameter method of circuit simulation, which is used to establish the relationship between heat load and indoor temperature changes. in and for t +1 time and t The indoor temperature at any given time; for t The outdoor temperature at any given time; R The equivalent thermal resistance of the building; C Specific heat capacity of indoor air; for t The heat load at time t; e is the natural constant; It is the equivalent thermal time constant in the thermodynamic model of a building, used to measure how fast the building's temperature changes; For the time interval, take =0.1h; After deformation of the first-order thermodynamic model of the building, the heat load before and after reduction is obtained: in, Before the reduction t The heat load at any given time; After reduction t The heat load at any given time; This is to reduce the amount of heat load.

5. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response, as described in claim 1, is characterized in that... Reductions in heating-related electrical loads can lead to lower indoor temperatures and reduced user comfort; therefore, the amount of temperature drop and heat load reduction must be controlled. in, Let t be the indoor temperature at time t; T is a time series over a day, and we take T=24. This is the lower limit of indoor temperature; for t The maximum amount of heat load reduction at any given moment; This represents the maximum total reduction in heat load within a single day. To reduce the impact of frequent reductions in heating-related electrical loads on users' daily lives, it is stipulated that a maximum of one continuous reduction in heating load should be implemented per day. in and The minimum and maximum duration of the reduction in heating-related electrical load within a day; and Let be variables 0 and 1 at times t and t+1, where 0 indicates that the reduction operation is not performed and 1 indicates that the reduction operation is performed.

6. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response, as described in claim 1, is characterized in that... The input terminal of the electric boiler that responds to demand signals is connected to the power distribution network node, and the output terminal is connected to the heating load. Energy conversion of the electric boiler is required. in, The electrothermal conversion coefficient of the electric boiler; A sequence of heating-related electrical load demand response signals for a single day; To reduce the amount of heat load; Let be the demand response signal of the heat load at time t, where .

7. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion under demand response, as described in claim 1, is characterized in that... Considering the impact of economic incentives on user electricity consumption behavior, and based on consumer psychology, an optimistic scenario is used. and pessimistic situations Describe the user response rate: In the formula, p For economic incentives; and These represent the dead-zone response rate under optimistic and pessimistic scenarios, respectively. and These represent the slopes of the response rate curves under optimistic and pessimistic conditions, respectively. a and b These are critical excitation and saturation excitation, respectively. This represents the user's saturation response rate.

8. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion driven by demand response, as described in claim 1, is characterized in that... Considering the uncertainty of user response rate, we assume that the user response rate follows a normal distribution random variable within an interval with upper and lower limits under optimistic and pessimistic scenarios. ; This represents the actual user response rate, where p is the economic incentive; N indicates a normal distribution. This represents the expected value of the actual user response rate under economic incentive p; This represents the variance of the actual user response rate under economic incentive p.

9. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion driven by demand response, as described in claim 1, is characterized in that... Other predictive influencing factors include: t The system considers electricity price, temperature, and humidity at specific times; it employs an iterative load forecasting method to predict the load at the next time point based on the historical load at the previous four time points.

10. A method for short-term forecasting of heating-type electrical load based on data-knowledge fusion driven by demand response, as described in claim 1, is characterized in that... The knowledge of interactive response of heating-type electrical load under demand response and common predictive influencing factors are used as input features to introduce a data-driven model; the data-driven model includes: LSTM neural network, BP neural network and SVM machine learning model; The prediction evaluation indicators selected are the mean square error (RMSE) and the mean relative percentage error (MAPE). In the formula, RMSE represents the mean square error; MAPE represents the mean relative percentage error; and n represents the total number of load samples. This represents the i-th load data; This represents the predicted value of the i-th load data. This represents the actual value of the i-th load data.

Images