A method and system for predicting secondary network heating based on controllable instantaneous heat energy saving rate

Abstract

This application relates to a method for predicting the temperature of a secondary heating network based on controllable instantaneous heat energy saving rate, belonging to the field of smart heating. The method includes: acquiring outdoor meteorological data and unit operation data; preprocessing the data; constructing an instantaneous heat prediction input feature set using historical time-series data of outdoor meteorological data, and establishing an instantaneous heat prediction model; constructing a secondary heating network temperature prediction input feature set using unit operation data and historical time-series data of instantaneous heat, and establishing a secondary heating network temperature prediction model; substituting current outdoor meteorological data into the instantaneous heat prediction model to obtain the predicted instantaneous heat values ​​for the unit over several hours; and substituting current unit operation data and the instantaneous heat values ​​for the unit over several hours into the secondary heating network temperature prediction model to obtain the predicted secondary heating network water temperature values ​​for several hours. The prediction model of this application can achieve precise control of the secondary heating network temperature based on changes in system operating conditions and a customized energy saving rate.

Description

Technical Field

[0001] This application relates to the field of intelligent heating control systems, and in particular to a method and system for predicting the temperature of a secondary heating network based on controllable instantaneous heat energy saving rate. Background Technology

[0002] In winter, the centralized heating process in northern cities typically involves using waste heat from steam generated by thermal power plants to heat the primary heating network to 60-130°C. This heat is then exchanged between the primary and secondary heating networks via heat exchange station units, heating the hot water in the secondary network to the supply temperature before it is delivered to users. The secondary network supply temperature is usually adjusted in real-time based on changes in outdoor temperature and the actual indoor temperature at the user's location.

[0003] With the development of IoT and AI technologies, smart heating systems are often used to regulate the secondary network water supply temperature of heat exchange station units. By forecasting the load of the heating system and considering historical data and weather forecast data of the heat exchange station for model simulation, the load forecast results and the secondary network water supply temperature are obtained.

[0004] The load forecasting model relies on historical data from the heat exchange station units for training. After the system operating conditions change, the prediction accuracy of the load forecasting model is greatly reduced. In addition, each heat exchange station unit has put forward new management requirements for energy saving rate, which the existing load forecasting model cannot meet. Summary of the Invention

[0005] In order to enable the prediction model to adapt to changes in system operating conditions and to achieve precise control of the secondary network heating temperature based on a custom energy saving rate, thereby improving the applicability and accuracy of the prediction model, this application provides a method and system for predicting secondary network heating based on a controllable instantaneous heat energy saving rate.

[0006] Firstly, this application provides a method for predicting the temperature of a secondary power grid based on controllable instantaneous heat energy saving rate, employing the following technical solution:

[0007] A method for predicting the temperature of a secondary power grid based on controllable instantaneous heat energy saving rate includes the following steps:

[0008] Acquire historical outdoor meteorological data and unit operation data;

[0009] Preprocessing of outdoor meteorological data and unit operation data;

[0010] A feature set for instantaneous heat prediction was constructed using historical time-series data of preprocessed outdoor meteorological data. A neural network model was then used for training to establish an instantaneous heat prediction model.

[0011] A secondary grid heating prediction input feature set is constructed using unit operation data and historical time series data of instantaneous heat. A neural network model is then used for training to establish a secondary grid heating prediction model.

[0012] Substitute the current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction values ​​for the unit over the next few hours;

[0013] By substituting the current unit operating data and the instantaneous heat of the unit over the next few hours into the secondary network temperature prediction model, the predicted value of the secondary network water supply temperature over the next few hours is obtained.

[0014] Since changes in system operating conditions can significantly affect the actual secondary network water supply temperature, if historical data after changes in system operating conditions are directly used as input values ​​for the load forecasting model, the system operating condition data will be mixed in with a large number of input values ​​that affect the load forecasting results. It will be difficult to reflect the changes in these data as an impact on the predicted results of the secondary network water supply temperature, resulting in a large deviation between the predicted and actual values ​​of the secondary network water supply temperature. This application employs the aforementioned technical solution, firstly, using historical time-series outdoor meteorological data to construct an input feature set for instantaneous heat prediction. A neural network model is then used to train and establish an instantaneous heat prediction model. Since outdoor meteorological data does not contain data related to system operating conditions, and the impact of outdoor meteorological data is precisely the key parameter affecting the instantaneous heat of the heat exchanger unit, the instantaneous heat prediction model established in this way can not only accurately predict the instantaneous heat value of the heat exchanger unit for several hours in the future, but also makes the instantaneous heat prediction model more focused on highly correlated key features, accelerating its convergence speed. Secondly, using unit operating data and historical time-series instantaneous heat data, a secondary network heating prediction input feature set is constructed. A neural network model is then used to train and establish a secondary network heating prediction model. System operating condition data is used as part of the unit operating data. Using unit operating data and instantaneous heat data, the input feature set of the secondary network heating prediction model is constructed, and the secondary network water supply temperature is predicted in stages. This allows the secondary network heating prediction model to adapt to changes in system operating conditions and make corresponding adjustments to the secondary network heating prediction results based on changes in system operating conditions. Even after changes in system operating conditions, precise control of the secondary network heating temperature can still be achieved.

[0015] In a specific feasible implementation, the step of substituting the current unit operating data and the instantaneous heat of the unit for the next several hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next several hours specifically includes:

[0016] Based on the predicted instantaneous heat value of the unit for the next several hours and the set energy saving rate, the target instantaneous heat value of the unit for the next several hours is obtained;

[0017] Substituting the current unit operating data and the unit's instantaneous heat target value for the next few hours into the secondary network temperature prediction model, we obtain the predicted value of the secondary network water supply temperature for the next few hours.

[0018] By adopting the above technical solution, the energy-saving rate is customized according to the management requirements of the heat exchange station units. The secondary network water supply temperature is predicted in stages by establishing an instantaneous heat prediction model and a secondary network temperature prediction model. Using the set energy-saving rate and the predicted value of the instantaneous heat prediction model, and by reversing the calculation process of the energy-saving rate, the instantaneous heat target value for several hours in the future can be derived. This process is interspersed between the operation of the instantaneous heat prediction model and the secondary network temperature prediction model, ensuring that the entire process conforms to the heating mechanism and control strategy. Based on the predicted value of the instantaneous heat prediction model and the energy-saving rate, the instantaneous heat target value for several hours in the future under the set energy-saving rate is calculated. This instantaneous heat target value is then substituted into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature under the set energy-saving rate. This achieves precise control of the secondary network water supply temperature for the customized energy-saving rate and enables corresponding control strategies for the secondary network temperature supply based on different energy-saving rates, ensuring that the energy-saving rate setting process is controllable.

[0019] In one specific implementation scheme, obtaining the target instantaneous heat value of the unit for the next several hours based on the predicted instantaneous heat value of the unit for the next several hours and the set energy saving rate specifically includes:

[0020]

[0021] in,

[0022] This represents the instantaneous heat target value for the unit in a future hour.

[0023] This is the predicted instantaneous heat output of the unit for a future hour.

[0024] Energy saving rate.

[0025] In a specific feasible implementation, the unit operating data includes the unit's primary network supply water temperature, primary network return water temperature, primary network supply water pressure, primary network return water pressure, secondary network supply water pressure, secondary network return water pressure, circulating pump frequency, and secondary network instantaneous flow rate.

[0026] By adopting the above technical solution, key factors affecting the secondary network water supply temperature—including the primary network supply water temperature, primary network return water temperature, primary network supply water pressure, primary network return water pressure, secondary network supply water pressure, secondary network return water pressure, circulating pump frequency, and instantaneous flow rate of the secondary network—are used as input data for the secondary network temperature prediction model, ensuring the model's accuracy. Changes in system operating conditions, such as the instantaneous flow rate of the secondary network, can significantly impact the secondary network water supply temperature. The secondary network temperature prediction model can adapt to these changes and predict the secondary network water supply temperature accordingly, achieving precise control of the secondary network heating temperature.

[0027] In one specific implementation scheme, when the collected value of the instantaneous flow rate of the secondary network exceeds 10% of the average value of the instantaneous flow rate of the secondary network collected in the previous several hours, the validity of the current instantaneous flow rate of the secondary network is determined based on the change of the circulation pump frequency. If it is valid, it is retained; otherwise, the average value of the instantaneous flow rate of the secondary network collected in the previous several hours is used as the current instantaneous flow rate value of the secondary network.

[0028] By adopting the above technical solution, after the heating system undergoes secondary network balancing system modification, heating is generally provided through a small flow rate and large temperature difference. At this time, the system operating conditions change, especially the instantaneous flow rate of the secondary network, which fluctuates significantly. In practice, the instantaneous flow rate data of the secondary network is usually collected by the ultrasonic heat meter of the heating unit's secondary network. Considering the large fluctuations in flow measurement by ultrasonic heat meters, directly using the collected raw instantaneous flow rate value of the secondary network in the secondary network temperature prediction model would cause data noise interference and unstable secondary network temperature prediction. Therefore, the average value of the instantaneous flow rate collected in the previous few hours is compared with the current instantaneous flow rate value. When the instantaneous flow rate of the secondary network changes significantly, the current instantaneous flow rate is verified. The validity of the currently collected instantaneous flow rate value is determined by the relationship between the instantaneous flow rate of the secondary network and the frequency of the circulating pump, thus ensuring the accuracy of the instantaneous flow rate data and further improving the accuracy of the secondary network temperature prediction model.

[0029] In a specific feasible implementation, the following steps are also included: classifying the heat exchange station units into distributed pump systems and electrically controlled valve systems according to their type; the operating data of the heat exchange station units in the distributed pump system also includes the frequency of the distributed pumps; the operating data of the heat exchange station units in the electrically controlled valve system also includes the opening degree of the electrically controlled valves.

[0030] By adopting the above technical solution, the heat exchange station units are divided into a distributed pump system and an electric regulating valve system. After classifying the heat exchange station units, the frequency of the distributed pumps or the opening degree of the electric regulating valves are added to the unit operation data according to their types. The important factors affecting the secondary network water supply temperature are fully considered for the operation process of different types of heat exchange station units, thereby improving the accuracy of the secondary network temperature prediction model.

[0031] In one specific implementation scheme, the outdoor meteorological data includes outdoor temperature for the past two hours, outdoor temperature for the past one hour, current outdoor temperature, outdoor temperature for the next one hour, outdoor temperature for the next two hours, outdoor temperature for the next three hours, outdoor temperature for the next four hours, outdoor temperature for the next five hours, outdoor temperature for the next six hours, real-time light intensity, real-time outdoor wind speed, real-time relative humidity, current month, and current hour.

[0032] By adopting the above technical solution, outdoor temperature, light intensity, outdoor wind speed, relative humidity, and the current month and hour are key factors affecting instantaneous heat. Using these as input data for the instantaneous heat prediction model can accurately reflect their relationship with changes in instantaneous heat. Among these, changes in outdoor temperature have a significant impact on instantaneous heat. Considering the impact of outdoor temperature at different times on instantaneous heat, data from the past two hours, past one hour, current time, and the next one, two, three, four, five, and six hours—all of which have a significant impact on instantaneous heat—are used as input data for the instantaneous heat prediction model, effectively improving the accuracy of the instantaneous heat prediction model.

[0033] In a specific feasible implementation plan, the following steps are also included: continuously adding the outdoor meteorological data completed in the current heating season to the instantaneous heat prediction input feature set, continuously adding the unit operation data completed in the current heating season to the secondary network heating prediction input feature set, and regularly conducting rolling training of new instantaneous heat prediction models and new secondary network heating prediction models.

[0034] By adopting the above technical solution, when the operating conditions and energy-saving rate requirements of the heat exchange station unit system change, the instantaneous heat prediction model and the secondary network heating prediction model trained solely on historical data before the change will no longer meet the control needs after the change. By continuously adding completed outdoor meteorological data and unit operation data from the current heating season to the input feature set of the instantaneous heat prediction, and continuously adding completed unit operation data to the input feature set of the secondary network heating prediction, the amount of data in both sets gradually increases. After periodically retraining the heat prediction model and the secondary network heating prediction model, the new models more closely reflect the control requirements after the changes in system operating conditions and energy-saving rate during the current heating season, thus improving the applicability and accuracy of the heat prediction model and the secondary network heating prediction model.

[0035] In a specific feasible implementation, the following steps are also included: taking outdoor meteorological data for a fixed time period backward from the current time as the input feature set data for instantaneous heat prediction, taking unit operation data for a fixed time period backward from the current time as the input feature set data for secondary grid heating prediction, and periodically conducting rolling training of new instantaneous heat prediction models and new secondary grid heating prediction models.

[0036] By adopting the above technical solution, after changes in the operating conditions and energy-saving rate requirements of the heat exchange station unit system, outdoor meteorological data for a fixed time period backwards from the current time is taken as the instantaneous heat prediction input feature set data, and unit operation data for a fixed time period backwards from the current time is taken as the secondary network heating prediction input feature set data. While keeping the data volume of the instantaneous heat prediction input feature set and the secondary network heating prediction input feature set constant, the ratio of data after changes in system operating conditions and energy-saving rate to the instantaneous heat prediction input feature set and the secondary network heating prediction input feature set gradually increases. After periodically retraining the heat prediction model and the secondary network heating prediction model, the new heat prediction model and the new secondary network heating prediction model are more closely aligned with the control requirements after changes in system operating conditions and energy-saving rate during this heating season, thus improving the applicability and accuracy of the heat prediction model and the secondary network heating prediction model.

[0037] Secondly, this application provides a secondary grid temperature prediction system based on controllable instantaneous heat energy saving rate, which adopts the following technical solution:

[0038] A secondary grid temperature prediction system based on controllable instantaneous heat energy saving rate includes:

[0039] The data acquisition module is used to acquire historical outdoor meteorological data and unit operation data;

[0040] The data processing module is used to preprocess outdoor meteorological data and unit operation data;

[0041] The instantaneous heat model construction module is used to construct an input feature set for instantaneous heat prediction using historical time-series data of preprocessed outdoor meteorological data, and to train and establish an instantaneous heat prediction model using a neural network model.

[0042] The secondary network heating model construction module uses unit operation data and historical time-series data of instantaneous heat to construct a secondary network heating prediction input feature set, and uses a neural network model for training to establish a secondary network heating prediction model;

[0043] The instantaneous heat prediction module is used to input the current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction value of the unit for the next several hours.

[0044] The secondary network temperature prediction module inputs the current unit operating data and the instantaneous heat of the unit for the next few hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next few hours.

[0045] Thirdly, this application provides a computer-readable storage medium, which adopts the following technical solution:

[0046] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed by the above-described method for predicting the temperature of a secondary grid supply system based on controllable instantaneous heat energy saving rate.

[0047] In summary, this application includes at least one of the following beneficial technical effects:

[0048] By performing tiered prediction of the secondary network water supply temperature, an instantaneous heat prediction model is first established using outdoor meteorological data, and then a secondary network heating prediction model is established using unit operation data. This enables the secondary network heating prediction model to adapt to changes in system operating conditions and to achieve precise control of the secondary network heating temperature after changes in system operating conditions, thereby improving the applicability and accuracy of the prediction model.

[0049] By customizing the energy saving rate, the instantaneous heat target value is calculated based on the energy saving rate and the predicted value of the instantaneous heat prediction model. Then, the predicted value of the secondary network water supply temperature is obtained based on the instantaneous heat target value, so as to achieve precise control of the secondary network water supply temperature for the custom energy saving rate, which is in line with the heating mechanism and control ideas.

[0050] The outdoor meteorological data and unit operation data completed in the current heating season are continuously added to the corresponding input feature set, so that the instantaneous heat prediction model and the secondary network heating prediction model are incrementally rolled and learned regularly, so that the new heat prediction model and the new secondary network heating prediction model are closer to the control requirements after the changes in system operating conditions and energy saving rate in this heating season.

[0051] By taking outdoor meteorological data and unit operation data over a fixed period of time backward from the current time as the corresponding input feature set data, the instantaneous heat prediction model and the secondary network heating prediction model are periodically subjected to quantitative rolling learning, so that the new heat prediction model and the new secondary network heating prediction model are closer to the control requirements after the changes in system operating conditions and energy saving rate in this heating season. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the secondary network temperature prediction method based on controllable instantaneous heat energy saving rate, as described in this application embodiment. Detailed Implementation

[0053] The following is in conjunction with the appendix Figure 1 This application will be described in further detail.

[0054] This application discloses a method for predicting the temperature of a secondary power grid based on controllable instantaneous heat energy saving rate.

[0055] Example 1

[0056] A method for predicting the temperature of a secondary power grid based on controllable instantaneous heat energy saving rate includes the following steps:

[0057] S1: Acquire historical outdoor meteorological data and unit operation data, specifically including:

[0058] S101: Classify the heat exchange station units into distributed pump systems and electrically controlled valve systems based on their type.

[0059] S102: For each type of heat exchange station unit data, correlation analysis is used to quantitatively analyze the factors affecting instantaneous heat and secondary network water supply temperature. Based on the analysis results, the heat exchange station data is divided into outdoor meteorological data, which has a more significant impact on instantaneous heat, and unit operation data, which has a more significant impact on secondary network water supply temperature.

[0060] S103: Construct a multiple linear regression model, use data analysis software to find the data with the highest correlation to instantaneous heat from a large amount of outdoor meteorological data, and use it as the input feature set data of the instantaneous heat prediction model.

[0061] In this application, the outdoor meteorological data most correlated with instantaneous heat include outdoor temperature over the past two hours, outdoor temperature over the past one hour, current outdoor temperature, outdoor temperature over the next one hour, outdoor temperature over the next two hours, outdoor temperature over the next three hours, outdoor temperature over the next four hours, outdoor temperature over the next five hours, outdoor temperature over the next six hours, real-time light intensity, real-time outdoor wind speed, real-time relative humidity, current month, and current hour.

[0062] S104: Construct a multiple linear regression model, and use data analysis software to find the data with the highest correlation to the secondary network hot water supply temperature from a large number of unit operation data, and use it as the input feature set data of the secondary network temperature prediction model.

[0063] In this application, the unit operating data most correlated with the secondary network water supply temperature include the unit primary network water supply temperature, the unit primary network return water temperature, the unit primary network water supply pressure, the unit primary network return water pressure, the unit secondary network water supply pressure, the unit secondary network return water pressure, the circulation pump frequency, and the unit secondary network instantaneous flow rate.

[0064] Since the instantaneous flow rate data of the secondary network is usually collected by the ultrasonic heat meter of the heating unit's secondary network, and considering the large fluctuations in flow measurement by the ultrasonic heat meter, directly using the original instantaneous flow rate data in the secondary network temperature prediction model would cause data noise interference and instability in the secondary network temperature prediction. Therefore, when the collected instantaneous flow rate value of the secondary network exceeds 10% of the average value of the collected instantaneous flow rate values ​​of the secondary network in the previous several hours, the validity of the current collected instantaneous flow rate value is determined based on the change in the circulation pump frequency. If valid, it is retained; otherwise, the average value of the collected instantaneous flow rate values ​​of the secondary network in the previous several hours is used as the current instantaneous flow rate value.

[0065] Specifically, when the collected instantaneous flow rate of the secondary network exceeds 10% of the average instantaneous flow rate of the secondary network collected in the previous four hours, the instantaneous flow rate of the secondary network at the current moment is verified. If the ratio of the instantaneous flow rate of the secondary network at the current moment to the instantaneous flow rate of the secondary network at a previous moment is equal to the cube of the ratio of the circulation pump frequency at the current moment to the circulation pump frequency at the same time in the previous moment, then the collected instantaneous flow rate of the secondary network at the current moment is valid data; otherwise, the average value of the instantaneous flow rate of the secondary network collected in the previous four hours is taken as the instantaneous flow rate value of the secondary network at the current moment.

[0066] In addition, the operating data of the heat exchange station units of the distributed pump system that are most correlated with the secondary network water supply temperature also include the frequency of the distributed pumps, and the operating data of the heat exchange station units of the electric regulating valve system that are most correlated with the secondary network water supply temperature also include the opening degree of the electric regulating valve.

[0067] S2: Preprocess outdoor meteorological data and unit operation data.

[0068] Specifically, this involves removing duplicate data from outdoor meteorological data and unit operation data, detecting and repairing errors in outdoor meteorological data and unit operation data, using interpolation methods to fill in missing values ​​in outdoor meteorological data and unit operation data, and converting the data into a unified format.

[0069] ETL tools are used to transform and integrate data, combining and standardizing outdoor meteorological data and unit operation data from different data sources, eliminating problems such as inconsistent formats and naming conventions in the data.

[0070] S3: Construct an input feature set for instantaneous heat prediction using historical time-series data of preprocessed outdoor meteorological data, train a neural network model, and establish an instantaneous heat prediction model.

[0071] Specifically, the preprocessed outdoor meteorological data is divided into training set, validation set and test set in a ratio of 7:2:1.

[0072] The input feature set for the instantaneous heat prediction model is constructed using preprocessed historical time-series data of the past two hours, past one hour, current outdoor temperature, future one hour, future two hours, future three hours, future four hours, future five hours, and future six hours, as well as real-time light intensity, real-time outdoor wind speed, real-time relative humidity, and the current month and current hour.

[0073] The neural network model for instantaneous heat prediction uses a fully connected layer for the input layer to receive the input feature set, an LSTM layer for the hidden layer, a 24-hour sliding time window, and a non-linear activation function for the output layer to output the instantaneous heat value. The loss function is MAE.

[0074] S4: Construct a secondary network heating prediction input feature set using unit operation data and historical time series data of instantaneous heat, train a neural network model, and establish a secondary network heating prediction model.

[0075] Specifically, the preprocessed outdoor meteorological data is divided into training set, validation set and test set in a ratio of 7:2:1.

[0076] The input feature set of the secondary network temperature prediction model is constructed using the historical time series data of the pre-processed primary network supply water temperature, primary network return water temperature, primary network supply water pressure, primary network return water pressure, secondary network supply water pressure, secondary network return water pressure, circulating pump frequency, secondary network instantaneous flow rate, and distribution pump frequency or electric regulating valve opening.

[0077] The neural network model for predicting secondary network temperature uses a fully connected layer to receive the input feature set in the input layer, a bidirectional LSTM layer in the hidden layer, a 24-hour sliding time window, and a nonlinear activation function to output the secondary network water supply temperature in the output layer. The loss function is Pinball loss (α=0.5).

[0078] S5: Substitute the current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction values ​​for the unit over the next few hours. In this embodiment, the output values ​​of the instantaneous heat prediction model are the instantaneous heat prediction values ​​for the unit over the next one, two, three, four, five, and six hours.

[0079] S6: Based on the predicted instantaneous heat output of the unit over the next few hours and the set energy saving rate, the target instantaneous heat output of the unit over the next few hours is obtained, specifically:

[0080]

[0081] in,

[0082] This represents the instantaneous heat target value for the unit in a future hour.

[0083] This is the predicted instantaneous heat output of the unit for a future hour.

[0084] Energy saving rate.

[0085] The energy saving rate is generally determined based on the management requirements of the heat exchange station.

[0086] S7: Substitute the current unit operating data and the unit's instantaneous heat target values ​​for the next several hours into the secondary network temperature prediction model to obtain the predicted values ​​of the secondary network water supply temperature for the next several hours. In this embodiment, the output values ​​of the secondary network temperature prediction model are the predicted values ​​of the secondary network water supply temperature for the next one, two, three, four, five, and six hours.

[0087] S8: Continuously add the outdoor meteorological data completed in the current heating season to the instantaneous heat prediction input feature set, and regularly conduct rolling training of the new instantaneous heat prediction model; continuously add the unit operation data completed in the current heating season to the secondary network heating prediction input feature set, and regularly conduct rolling training of the new secondary network heating prediction model.

[0088] The implementation principle of a secondary network temperature prediction method based on controllable instantaneous heat energy saving rate in this application embodiment is as follows:

[0089] First, a set of input features for instantaneous heat prediction is constructed using historical time-series outdoor meteorological data. A neural network model is then used to train and establish an instantaneous heat prediction model. This model can not only accurately predict the instantaneous heat value of the heat exchange unit for several hours in the future, but also make the instantaneous heat prediction model more focused on key features with high correlation, thus accelerating its convergence speed.

[0090] Then, by utilizing the historical time-series data of unit operation data and instantaneous heat, a secondary network heating prediction input feature set is constructed. A neural network model is used for training and a secondary network heating prediction model is established. System operating condition data is used as part of the unit operation data. The input feature set of the secondary network heating prediction model is constructed using the unit operation data and instantaneous heat data to perform graded prediction of the secondary network water supply temperature. This enables the secondary network heating prediction model to adapt to changes in system operating conditions and make corresponding adjustments to the secondary network heating prediction results according to changes in system operating conditions. It can also achieve precise control of the secondary network heating temperature after changes in system operating conditions.

[0091] Based on the predicted values ​​of the instantaneous heat forecasting model and the energy saving rate, the instantaneous heat target value for several hours in the future is calculated under the set energy saving rate. The instantaneous heat target value is then substituted into the secondary network heating forecasting model to obtain the predicted value of the secondary network water supply temperature under the set energy saving rate. This enables precise control of the secondary network water supply temperature for the customized energy saving rate. This process is interspersed between the operation of the instantaneous heat forecasting model and the secondary network heating forecasting model, making the entire process consistent with the heating mechanism and control ideas, and ensuring that the energy saving rate setting process is controllable.

[0092] By continuously adding completed outdoor meteorological data from the current heating season to the instantaneous heat forecast input feature set, and continuously adding completed unit operation data from the current heating season to the secondary network heating forecast input feature set, the data volume of the instantaneous heat forecast input feature set and the secondary network heating forecast input feature set is continuously expanded. This results in a gradual increase in the data on changes in system operating conditions and energy efficiency in the instantaneous heat forecast input feature set and the secondary network heating forecast input feature set. After periodic rolling training of the heat forecast model and the secondary network heating forecast model, the new heat forecast model and the new secondary network heating forecast model are made closer to the control requirements after changes in system operating conditions and energy efficiency in this heating season, thereby improving the applicability and accuracy of the heat forecast model and the secondary network heating forecast model.

[0093] Figure 1 This is a flowchart illustrating a secondary grid temperature prediction method based on controllable instantaneous heat energy saving rate, as described in an embodiment of this application. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows; unless explicitly stated otherwise, there is no strict order requirement for the execution of these steps, and they can be executed in other orders; and Figure 1At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0094] Example 2

[0095] The difference between this embodiment and Embodiment 1 is that:

[0096] S8: Take outdoor meteorological data for a fixed time period backward from the current time as the input feature set data for instantaneous heat prediction, and regularly conduct rolling training of the new instantaneous heat prediction model; take unit operation data for a fixed time period backward from the current time as the input feature set data for secondary network heating prediction, and regularly conduct rolling training of the new secondary network heating prediction model.

[0097] Outdoor meteorological data for a fixed period of time backwards from the current time is used as the input feature set for instantaneous heat prediction, and unit operation data for a fixed period of time backwards from the current time is used as the input feature set for secondary grid heating prediction. While keeping the data volume of the instantaneous heat prediction input feature set and the secondary grid heating prediction input feature set constant, the data in the instantaneous heat prediction input feature set and the secondary grid heating prediction input feature set after changes in system operating conditions and energy saving rates are gradually increased. After periodic rolling training of the heat prediction model and the secondary grid heating prediction model, the new heat prediction model and the new secondary grid heating prediction model are made closer to the control requirements after changes in system operating conditions and energy saving rates in this heating season, thereby improving the applicability and accuracy of the heat prediction model and the secondary grid heating prediction model.

[0098] This application also discloses a secondary network temperature prediction system based on controllable instantaneous heat energy saving rate.

[0099] A secondary grid temperature prediction system based on controllable instantaneous heat energy saving rate includes:

[0100] The data acquisition module is used to acquire historical outdoor meteorological data and unit operation data.

[0101] The heat exchange station units are classified into distributed pump systems and electrically controlled valve systems based on their type.

[0102] For each type of heat exchange station unit data, correlation analysis was used to quantitatively analyze the factors affecting instantaneous heat and secondary network water supply temperature. Based on the analysis results, the heat exchange station data were divided into outdoor meteorological data, which has a more significant impact on instantaneous heat, and unit operation data, which has a more significant impact on secondary network water supply temperature.

[0103] A multiple linear regression model was constructed, and data analysis software was used to find the data with the highest correlation to instantaneous heat from a large amount of outdoor meteorological data, which was then used as the input feature set data for the instantaneous heat prediction model.

[0104] In this application, the outdoor meteorological data most correlated with instantaneous heat include outdoor temperature over the past two hours, outdoor temperature over the past one hour, current outdoor temperature, outdoor temperature over the next one hour, outdoor temperature over the next two hours, outdoor temperature over the next three hours, outdoor temperature over the next four hours, outdoor temperature over the next five hours, outdoor temperature over the next six hours, real-time light intensity, real-time outdoor wind speed, real-time relative humidity, current month, and current hour.

[0105] A multiple linear regression model was constructed, and data analysis software was used to find the data with the highest correlation to the temperature of hot water supplied by the secondary network from a large number of unit operation data. This data was then used as the input feature set data for the secondary network temperature prediction model.

[0106] In this application, the unit operating data most correlated with the secondary network water supply temperature include the unit primary network water supply temperature, the unit primary network return water temperature, the unit primary network water supply pressure, the unit primary network return water pressure, the unit secondary network water supply pressure, the unit secondary network return water pressure, the circulation pump frequency, and the unit secondary network instantaneous flow rate.

[0107] Since the instantaneous flow rate data of the secondary network is usually collected by the ultrasonic heat meter of the heating unit's secondary network, and considering the large fluctuations in flow measurement by the ultrasonic heat meter, directly using the original instantaneous flow rate data in the secondary network temperature prediction model would cause data noise interference and instability in the secondary network temperature prediction. Therefore, when the collected instantaneous flow rate value of the secondary network exceeds 10% of the average value of the collected instantaneous flow rate values ​​of the secondary network in the previous several hours, the validity of the current collected instantaneous flow rate value is determined based on the change in the circulation pump frequency. If valid, it is retained; otherwise, the average value of the collected instantaneous flow rate values ​​of the secondary network in the previous several hours is used as the current instantaneous flow rate value.

[0108] Specifically, when the collected instantaneous flow rate of the secondary network exceeds 10% of the average instantaneous flow rate of the secondary network collected in the previous four hours, the instantaneous flow rate of the secondary network at the current moment is verified. If the ratio of the instantaneous flow rate of the secondary network at the current moment to the instantaneous flow rate of the secondary network at a previous moment is equal to the cube of the ratio of the circulation pump frequency at the current moment to the circulation pump frequency at the same time in the previous moment, then the collected instantaneous flow rate of the secondary network at the current moment is valid data; otherwise, the average value of the instantaneous flow rate of the secondary network collected in the previous four hours is taken as the instantaneous flow rate value of the secondary network at the current moment.

[0109] In addition, the operating data of the heat exchange station units of the distributed pump system that are most correlated with the secondary network water supply temperature also include the frequency of the distributed pumps, and the operating data of the heat exchange station units of the electric regulating valve system that are most correlated with the secondary network water supply temperature also include the opening degree of the electric regulating valve.

[0110] The data processing module is used to preprocess outdoor meteorological data and unit operation data.

[0111] Specifically, this involves removing duplicate data from outdoor meteorological data and unit operation data, detecting and repairing errors in outdoor meteorological data and unit operation data, using interpolation methods to fill in missing values ​​in outdoor meteorological data and unit operation data, and converting the data into a unified format.

[0112] ETL tools are used to transform and integrate data, combining and standardizing outdoor meteorological data and unit operation data from different data sources, eliminating problems such as inconsistent formats and naming conventions in the data.

[0113] The instantaneous heat model construction module is used to construct an input feature set for instantaneous heat prediction using historical time-series data of preprocessed outdoor meteorological data, and to train and establish an instantaneous heat prediction model using a neural network model.

[0114] The neural network model for instantaneous heat prediction uses a fully connected layer for the input layer to receive the input feature set, an LSTM layer for the hidden layer, a 24-hour sliding time window, and a non-linear activation function for the output layer to output the instantaneous heat value. The loss function is MAE.

[0115] The secondary network heating model construction module uses unit operation data and historical time-series data of instantaneous heat to construct a secondary network heating prediction input feature set, and uses a neural network model for training to establish a secondary network heating prediction model.

[0116] The neural network model for predicting secondary network temperature uses a fully connected layer to receive the input feature set in the input layer, a bidirectional LSTM layer in the hidden layer, a 24-hour sliding time window, and a nonlinear activation function to output the secondary network water supply temperature in the output layer. The loss function is Pinball loss (α=0.5).

[0117] The instantaneous heat prediction module is used to input current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction values ​​for the unit over the next several hours. In this embodiment, the output values ​​of the instantaneous heat prediction model are the instantaneous heat prediction values ​​for the unit over the next one, two, three, four, five, and six hours.

[0118] The custom energy-saving rate module is used to obtain the target instantaneous heat value of the unit for the next several hours based on the predicted instantaneous heat value of the unit for the next several hours and the set energy-saving rate.

[0119] Specifically,

[0120]

[0121] in,

[0122] This represents the instantaneous heat target value for the unit in a future hour.

[0123] This is the predicted instantaneous heat output of the unit for a future hour.

[0124] Energy saving rate.

[0125] The secondary network temperature prediction module is used to input the current unit operating data and the unit's instantaneous heat target value for the next several hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next several hours. In this embodiment, the output value of the secondary network temperature prediction model is the predicted value of the secondary network water supply temperature for the next one, two, three, four, five, and six hours.

[0126] The data and model update module is used to continuously add the outdoor meteorological data completed in the current heating season to the input feature set of instantaneous heat prediction, and to regularly conduct rolling training of the new instantaneous heat prediction model; it also continuously adds the unit operation data completed in the current heating season to the input feature set of secondary network heating prediction, and to regularly conduct rolling training of the new secondary network heating prediction model.

[0127] In another embodiment, the data and model update module is used to take outdoor meteorological data for a fixed time period backward from the current time as the input feature set data for instantaneous heat prediction, and periodically perform rolling training of the new instantaneous heat prediction model; and to take unit operation data for a fixed time period backward from the current time as the input feature set data for secondary network heating prediction, and periodically perform rolling training of the new secondary network heating prediction model.

[0128] This application also discloses a computer-readable storage medium.

[0129] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed, such as the above-described method for predicting the temperature of a secondary grid supply system based on controllable instantaneous heat energy saving rate. The computer-readable storage medium includes, for example, various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0130] This specific embodiment is merely an explanation of the present invention and is not intended to limit the invention. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A method for predicting the temperature of a secondary power grid based on controllable instantaneous heat energy saving rate, characterized in that, Includes the following steps: Acquire historical outdoor meteorological data and unit operation data; Preprocessing of outdoor meteorological data and unit operation data; A feature set for instantaneous heat prediction was constructed using historical time-series data of preprocessed outdoor meteorological data. A neural network model was then used for training to establish an instantaneous heat prediction model. A secondary grid heating prediction input feature set is constructed using unit operation data and historical time series data of instantaneous heat. A neural network model is then used for training to establish a secondary grid heating prediction model. Substitute the current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction values ​​for the unit over the next few hours; Substitute the current unit operating data and the instantaneous heat of the unit for the next few hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next few hours. The step of substituting current unit operating data and instantaneous heat from the unit over the next few hours into the secondary network temperature prediction model to obtain predicted values ​​for the secondary network water supply temperature over the next few hours specifically includes: Based on the predicted instantaneous heat value of the unit for the next several hours and the set energy saving rate, the target instantaneous heat value of the unit for the next several hours is obtained; Substitute the current unit operating data and the unit's instantaneous heat target value for the next few hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next few hours. The process of obtaining the target instantaneous heat value of the unit for the next several hours based on the predicted instantaneous heat value of the unit over the next few hours and the set energy saving rate specifically includes: ， in, This represents the instantaneous heat target value for the unit in a future hour. This is the predicted instantaneous heat output of the unit for a future hour. Energy saving rate.

2. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate as described in claim 1, characterized in that, The unit's operating data includes the primary network supply water temperature, the primary network return water temperature, the primary network supply water pressure, the primary network return water pressure, the secondary network supply water pressure, the secondary network return water pressure, the circulation pump frequency, and the instantaneous flow rate of the secondary network.

3. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate as described in claim 2, characterized in that, When the collected value of the instantaneous flow rate of the secondary network exceeds 10% of the average value of the instantaneous flow rate of the secondary network collected in the previous several hours, the validity of the current instantaneous flow rate of the secondary network is determined based on the change of the circulation pump frequency. If it is valid, it is retained; otherwise, the average value of the instantaneous flow rate of the secondary network collected in the previous several hours is used as the current instantaneous flow rate value of the secondary network.

4. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate as described in claim 1, characterized in that, It also includes the following steps: Based on the type of heat exchange station units, they are divided into distributed pump systems and electrically controlled valve systems. The operating data of heat exchange station units in distributed pump systems also includes the frequency of the distributed pumps; the operating data of heat exchange station units in electrically controlled valve systems also includes the opening degree of the electrically controlled valves.

5. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate as described in claim 1, characterized in that, The outdoor meteorological data includes outdoor temperature for the past two hours, outdoor temperature for the past one hour, current outdoor temperature, outdoor temperature for the next one hour, outdoor temperature for the next two hours, outdoor temperature for the next three hours, outdoor temperature for the next four hours, outdoor temperature for the next five hours, outdoor temperature for the next six hours, real-time light intensity, real-time outdoor wind speed, real-time relative humidity, current month, and current hour.

6. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate according to claim 1, characterized in that, It also includes the following steps: The outdoor meteorological data completed in the current heating season is continuously added to the input feature set for instantaneous heat prediction, and the unit operation data completed in the current heating season is continuously added to the input feature set for secondary network heating prediction. In addition, new instantaneous heat prediction models and new secondary network heating prediction models are regularly trained.

7. The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate according to claim 1, characterized in that, It also includes the following steps: taking outdoor meteorological data for a fixed period of time backward from the current time as the input feature set data for instantaneous heat prediction, taking unit operation data for a fixed period of time backward from the current time as the input feature set data for secondary grid heating prediction, and periodically conducting rolling training of new instantaneous heat prediction models and new secondary grid heating prediction models.

8. A secondary grid temperature prediction system based on controllable instantaneous heat energy saving rate, characterized in that, The method for predicting secondary grid temperature supply based on controllable instantaneous heat energy saving rate as described in any one of claims 1 to 7 includes: The data acquisition module is used to acquire historical outdoor meteorological data and unit operation data; The data processing module is used to preprocess outdoor meteorological data and unit operation data; The instantaneous heat model construction module is used to construct an input feature set for instantaneous heat prediction using historical time-series data of preprocessed outdoor meteorological data, and to train and establish an instantaneous heat prediction model using a neural network model. The secondary network heating model construction module uses unit operation data and historical time-series data of instantaneous heat to construct a secondary network heating prediction input feature set, and uses a neural network model for training to establish a secondary network heating prediction model; The instantaneous heat prediction module is used to input the current outdoor meteorological data into the instantaneous heat prediction model to obtain the instantaneous heat prediction value of the unit for the next several hours. The secondary network temperature prediction module inputs the current unit operating data and the instantaneous heat of the unit for the next few hours into the secondary network temperature prediction model to obtain the predicted value of the secondary network water supply temperature for the next few hours.

Images