Electric heating system and heat supply method based on self-adaptive regulation and control of power grid load
Through the heating method of adaptive control of grid load, long and short-term memory network models are used to predict grid load and adjust the power of the electric heating system in real time, solving the problem of large-capacity electric heating system access to the distribution network occupying too many feeder capacity, realizing the grid burden and the efficient operation of the electric heating system.
Patent Information
- Application Number
- CN202411039435.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-11-12
AI Technical Summary
When a large-capacity electric heating system is connected to the existing distribution network, it occupies too much feeder capacity, resulting in an increased burden on the distribution network.
Adopting heating method based on adaptive control of grid load, through the collaborative work of the integrated terminal on the substation side and the user side and the load prediction cloud platform, the long and short-term memory network model is used to predict the grid load, and the operating power of the electric heating system is adjusted in real time to avoid peaks. overload of the grid during periods.
It realizes efficient operation of the electric heating system, reduces the burden on the distribution network, improves the utilization rate of the electric heating system, meets the thermal energy needs of users during peak periods, and avoids grid overload through collaborative optimization control.
Smart Images

Figure CN118944106A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power system control, and in particular relates to an electric heating system and a heating method based on adaptive control of power grid load. Background Art
[0002] With the proposal of the "dual carbon" goal, various parks have been seeking more environmentally friendly and efficient heat source supply solutions. High-power electric heating systems have become an important part of the park's heat source due to their good regulation performance and rapid response characteristics. However, the access of large-capacity electric heating systems has brought new challenges to the existing distribution network, especially during the peak period of heating demand, which may occupy too much feeder capacity, resulting in an increased burden on the distribution network.
[0003] The prior art discloses a method, device and equipment for optimizing the operation of a park electric heating system with a publication number of CN117559558A. The technical solution in the comparative document collects heating data to build a mathematical model, optimizes and solves the mathematical model to obtain an optimization result, and then optimizes and implements the optimization according to the optimization result. The prior art discloses a method and system for real-time scheduling of park electric heating that can adapt to a combination of multi-energy equipment in multiple scenarios with a publication number of CN114399172A. The technical solution in the comparative document collects the operation data of the electric heating system in multiple time periods to obtain a scheduling reference value, models the equipment components according to the scheduling reference value, and then uses a heuristic algorithm to obtain the scheduling control results for multiple equipment in each park. The present invention provides an electric heating system and a heating method based on adaptive control of power grid load data through adaptive control of power grid load data. Summary of the invention
[0004] The technical problem solved by the present invention is to overcome the problem that the access of a large-capacity electric heating system occupies too much feeder capacity of the existing distribution network, resulting in an increased burden on the distribution network.
[0005] The technical solution adopted by the present invention is as follows:
[0006] A heating method based on grid load adaptive control comprises the following steps:
[0007] S1: The fusion terminal on the substation side collects the power data of the substation and sends it to the load forecasting cloud platform and the fusion terminal on the user side;
[0008] S2: The load forecasting cloud platform receives the data and analyzes and forecasts the above data, and sends a power data restriction instruction to the user-side fusion terminal according to the forecast results;
[0009] S3: The user-side fusion terminal receives the power data and power data restriction instructions from the substation head end, collects the power data of the user-side electric heating system, and restricts the power data of the user-side electric heating system.
[0010] The power data collected, analyzed and predicted in the above steps include voltage and power data. The power data is analyzed and predicted in step S2 based on the long short-term memory network power grid load prediction model, which specifically includes the following sub-steps:
[0011] S31: Data preprocessing and preparation: recording the power grid load and cleaning the missing values or abnormal values contained in the data, wherein the cleaning includes inserting values or deleting abnormal points;
[0012] S32: Conversion to a supervised learning problem: Convert time series data into a supervised learning format;
[0013] S33: Split the dataset: Split the dataset into training set, validation set and test set; the training set is used to train the model, the validation set is used to adjust the model parameters, and the test set is used to evaluate the model performance;
[0014] S34: Build LSTM model: Use the deep learning library to build an LSTM model. The model contains multiple LSTM layers. Define the model structure according to the number of LSTM layers, the number of units in each layer, and the activation function. Add a fully connected layer as the output layer to predict the power value of the next time step.
[0015] S35: Training model: Select mean square error as the loss function and optimization target; select an Adam optimization algorithm for the gradient descent process; use the training set data to train the model, and the model weights will be updated according to the gradient of the loss function; use the validation set to evaluate the model and adjust the hyperparameters according to the validation loss;
[0016] S36: Prediction and evaluation: Use the trained model to make predictions on the test set; calculate the evaluation indicators of the prediction results, using root mean square error and mean absolute error; analyze the performance of the model to determine whether further adjustments to the model structure or hyperparameters are needed;
[0017] S37: Model deployment and use: Deploy the trained model to the production environment; input the latest power load data into the model in real time or on demand to obtain short-term forecasts.
[0018] The sending of power data limitation instructions in step S2 specifically includes the following steps: when the load forecasting cloud platform predicts that the power grid is about to enter a high load state, it will send a power limitation instruction to the substation side fusion terminal, and the substation side fusion terminal will send the power limitation instruction to the user side fusion terminal. The user side fusion terminal will dynamically adjust the operating power of the electric heating system according to the received power limitation instruction.
[0019] The electric heating system based on adaptive control of power grid load includes a substation and a user side. A substation side fusion terminal is installed at the outlet of the substation. The user side includes an electric heating system. A user side fusion terminal is installed at the entrance of the electric heating system. The user side fusion terminal is connected to the substation side fusion terminal and the load prediction cloud platform. The load prediction cloud platform is connected to the substation side fusion terminal. The substation is provided with an outgoing line switch, which is connected to the substation side fusion terminal and communicates through the MUDBUS protocol. The substation side fusion terminal communicates with the user side fusion terminal through the international standard 104 protocol. The substation side fusion terminal and the user side fusion terminal communicate with the load prediction cloud platform through the MQTT protocol. The electric heating system communicates with the user side fusion terminal through the MODBUS protocol.
[0020] The present invention has the following beneficial effects:
[0021] Through the adaptive power limiting strategy of the power grid, the efficient operation of the electric heating system is achieved without occupying too much feeder capacity; the management of the power grid load is optimized and the burden on the distribution network is reduced; the utilization rate of the electric heating system is improved and the users' thermal energy demand during peak periods is met; through the combination of the fusion terminal and the cloud-based load power prediction platform, the coordinated optimization control of the electric heating system and the power grid is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a schematic diagram of the power system of the present invention;
[0023] Figure 2 This is a flow chart of the heating method of the present invention;
[0024] Figure 3 This is the flow chart of the power grid load forecasting model based on long short-term memory network.
[0025] Among them: 1. Substation; 2. Outgoing line switch; 3. User side; 4. Electric heating system; 5. Substation side fusion terminal; 6. User side fusion terminal; 7. Load prediction cloud platform. DETAILED DESCRIPTION
[0026] Embodiment 1:
[0027] A heating method based on grid load adaptive control comprises the following steps:
[0028] S1: The substation-side fusion terminal 5 collects power data of the substation 1 and sends it to the load prediction cloud platform 7 and the user-side fusion terminal 6;
[0029] S2: The load prediction cloud platform 7 receives the data and analyzes and predicts the data, and sends 6 a power data restriction instruction to the user-side fusion terminal according to the prediction result;
[0030] S3: The user-side fusion terminal 6 receives the power data and power data restriction instruction of the head end of the substation 1, and at the same time collects the power data of the electric heating system 4 in the user side 3, and restricts the power data of the electric heating system 4.
[0031] The power data collected, analyzed and predicted in the above steps include voltage and power data. The power data is analyzed and predicted in step S2 based on the long short-term memory network power grid load prediction model, which specifically includes the following sub-steps:
[0032] S31: Data preprocessing and preparation: recording the power grid load and cleaning the missing values or abnormal values contained in the data, wherein the cleaning includes inserting values or deleting abnormal points;
[0033] S32: Conversion to a supervised learning problem: Convert time series data into a supervised learning format;
[0034] S33: Split the dataset: Split the dataset into training set, validation set and test set; the training set is used to train the model, the validation set is used to adjust the model parameters, and the test set is used to evaluate the model performance;
[0035] S34: Build LSTM model: Use the deep learning library to build an LSTM model. The model contains multiple LSTM layers. Define the model structure according to the number of LSTM layers, the number of units in each layer, and the activation function. Add a fully connected layer as the output layer to predict the power value of the next time step.
[0036] S35: Training model: Select mean square error as the loss function and optimization target; select an Adam optimization algorithm for the gradient descent process; use the training set data to train the model, and the model weights will be updated according to the gradient of the loss function; use the validation set to evaluate the model and adjust the hyperparameters according to the validation loss;
[0037] S36: Prediction and evaluation: Use the trained model to make predictions on the test set; calculate the evaluation indicators of the prediction results, using root mean square error and mean absolute error; analyze the performance of the model to determine whether further adjustments to the model structure or hyperparameters are needed;
[0038] S37: Model deployment and use: Deploy the trained model to the production environment; input the latest power load data into the model in real time or on demand to obtain short-term forecasts.
[0039] The sending of power data limitation instructions in step S2 specifically includes the following steps: when the load prediction cloud platform 7 predicts that the power grid is about to enter a high load state, it will send a power limitation instruction to the substation side fusion terminal 5, and the substation side fusion terminal 5 will send the power limitation instruction to the user side fusion terminal 6. The user side fusion terminal 6 will dynamically adjust the operating power of the electric heating system 4 according to the received power limitation instruction.
[0040] Specifically, the sensors on the substation fusion terminal collect the voltage and power data at the head end of the substation feeder and send them to the load prediction cloud platform and the user-side fusion terminal; the load prediction cloud platform analyzes and predicts the above data and sends a power limitation instruction to the user-side fusion terminal via the substation fusion terminal based on the prediction results; the user-side fusion terminal collects the power data of the user-side electric heating system and adaptively adjusts the power of the electric heating system after receiving the power limitation instruction.
[0041] Specifically, the load forecasting cloud platform establishes a power grid load forecasting model based on the long short-term memory network. During the data preprocessing and preparation stage, the power grid load is recorded at a point every 15 minutes. After the time series data is converted into a supervised learning format, the load values of the past several 15 minutes are used as the learning basis. The learning target is the load value of the next 15 minutes. On this basis, a sliding window is created, and the data of the past 24 hours is used to predict the load of the next 15 minutes. The data set is divided into a training set, a validation set, and a test set, which are used for training the model, adjusting parameters, and evaluating performance, respectively. An LSTM model is constructed, and a fully connected layer is added as the output layer to predict the power value of the next time step. After training, predicting, and evaluating the above model, the model is applied to the production environment, and the latest power load data can be input as needed to obtain short-term predictions.
[0042] Embodiment 2:
[0043] like Figure 1 As shown, the electric heating system based on adaptive control of power grid load includes a substation 1 and a user side 3. A substation side fusion terminal 5 is installed at the outlet of the substation 1. The user side 3 includes an electric heating system 4. A user side fusion terminal 6 is installed at the entrance of the electric heating system 4. The user side fusion terminal 6 is connected to the substation side fusion terminal 5 and the load prediction cloud platform 7. The load prediction cloud platform 7 is connected to the substation side fusion terminal 5; the substation 1 is provided with an outgoing line switch 2, the outgoing line switch 2 is connected to the substation side fusion terminal 5 and communicates through the MUDBUS protocol; the substation side fusion terminal 5 communicates with the user side fusion terminal 6 through the international standard 104 protocol; the substation side fusion terminal 5 and the user side fusion terminal 6 communicate with the load prediction cloud platform 7 through the MQTT protocol; the electric heating system 4 communicates with the user side fusion terminal 6 through the MODBUS protocol.
[0044] In practical applications, firstly, corresponding fusion terminals are installed at the substation 1 and the user side 3. The substation-side fusion terminal 5 is installed at the head end of the feeder to collect the power and voltage data of the feeder through sensors, and send the data to the load prediction cloud platform 7 through the wireless communication module. At the same time, the substation-side fusion terminal 5 receives control instructions from the load prediction cloud platform 7.
[0045] The user-side fusion terminal 6 receives data from the substation 1 and monitors the power consumption of the electric heating system 4. When the load forecasting cloud platform 7 issues a power limit instruction according to the load forecasting result, the user-side fusion terminal 6 will automatically adjust the power output of the electric heating system 4 to reduce the impact on the power grid while ensuring the heating demand in the park.
[0046] In this way, the electric heating system 4 can achieve intelligent coordination with the grid load, which not only ensures the thermal energy supply of the park, but also avoids excessive pressure on the grid, achieving the dual goals of efficient energy utilization and environmental protection.
Claims
1. A heating method based on adaptive control of power grid load, characterized in that: The following steps are involved: S1: The substation-side fusion terminal (5) collects power data from the substation (1) and sends it to the load prediction cloud platform (7) and the user-side fusion terminal (6); S2: The load prediction cloud platform (7) receives the data and performs analysis and prediction on the data, and sends a power data restriction instruction to the user-side fusion terminal (6) according to the prediction result; S3: The user-side fusion terminal (6) receives the power data and power data restriction instruction from the head end of the substation (1), and at the same time collects the power data of the electric heating system (4), and restricts the power data of the electric heating system (4).
2. The heating method based on grid load adaptive control according to claim 1 is characterized in that: The electric power data includes voltage and power data.
3. The heating method based on grid load adaptive control according to claim 1 is characterized in that: The analysis and prediction of the power data in step S2 is performed based on the long short-term memory network power grid load prediction model, which specifically includes the following sub-steps: S31: Data preprocessing and preparation: recording the power grid load and cleaning the missing values or abnormal values contained in the data, wherein the cleaning includes inserting values or deleting abnormal points; S32: Conversion to a supervised learning problem: Convert time series data into a supervised learning format; S33: Split the dataset: Split the dataset into training set, validation set and test set; the training set is used to train the model, the validation set is used to adjust the model parameters, and the test set is used to evaluate the model performance; S34: Build LSTM model: Use the deep learning library to build an LSTM model. The model contains multiple LSTM layers. Define the model structure according to the number of LSTM layers, the number of units in each layer, and the activation function. Add a fully connected layer as the output layer to predict the power value of the next time step. S35: Training model: Select mean square error as the loss function and optimization target; select an Adam optimization algorithm for the gradient descent process; use the training set data to train the model, and the model weights will be updated according to the gradient of the loss function; use the validation set to evaluate the model and adjust the hyperparameters according to the validation loss; S36: Prediction and evaluation: Use the trained model to make predictions on the test set; calculate the evaluation indicators of the prediction results, using root mean square error and mean absolute error; analyze the performance of the model to determine whether further adjustments to the model structure or hyperparameters are needed; S37: Model deployment and use: Deploy the trained model to the production environment; input the latest power load data into the model in real time or on demand to obtain short-term forecasts.
4. The heating method based on grid load adaptive control according to claim 1 or 2, characterized in that: The sending of the power data restriction instruction in step S2 specifically includes the following steps: when the load prediction cloud platform (7) predicts that the power grid is about to enter a high load state, it will send a power restriction instruction to the substation side fusion terminal (5), and the substation side fusion terminal (5) will send the power restriction instruction to the user side fusion terminal (6), and the user side fusion terminal (6) will dynamically adjust the operating power of the electric heating system (4) according to the received power restriction instruction.
5. An electric heating system based on grid load adaptive control, based on the heating method based on grid load adaptive control according to any one of claims 1 to 4, characterized in that: The invention comprises a substation (1) and a user side (3), wherein a substation side fusion terminal (5) is installed at the outlet of the substation (1), the user side (3) comprises an electric heating system (4), a user side fusion terminal (6) is installed at the inlet of the electric heating system (4), the user side fusion terminal (6) is connected to the substation side fusion terminal (5) and a load prediction cloud platform (7), and the load prediction cloud platform (7) is connected to the substation side fusion terminal (5).
6. The electric heating system based on grid load adaptive control according to claim 5 is characterized in that: The substation (1) is provided with an outgoing line switch (2), and the outgoing line switch (2) is connected to a substation-side fusion terminal (5) and communicates via a MUDBUS protocol.
7. The electric heating system based on grid load adaptive control according to claim 5 is characterized in that: The substation-side fusion terminal (5) communicates with the user-side fusion terminal (6) via the international standard 104 protocol.
8. The electric heating system based on grid load adaptive control according to claim 5 is characterized in that: The substation-side fusion terminal (5) and the user-side fusion terminal (6) communicate with the load prediction cloud platform (7) via the MQTT protocol.
9. The electric heating system based on grid load adaptive control according to claim 5, characterized in that: The electric heating system (4) communicates with the user-side fusion terminal (6) via the MODBUS protocol.
Citation Information
Patent Citations
Park electric heating real-time scheduling method and system capable of adapting to multi-energy equipment combination in multiple scenes
CN114399172A
Park electric heating system operation optimization method, device and equipment
CN117559558A
Cited By
Self-adaptive power regulation and control system and regulation and control method thereof
CN118998810A
Electric heating collaboration method for directional transportation of multi-scale electric heat energy flow
CN119358966A
