Energy-saving optimization method and system for central heating system

Abstract

The invention relates to the field of central heating, and discloses an energy-saving optimization method and system for a central heating system, sensors are arranged on a pipe network and a user side to collect data such as temperature, flow, pressure and thermal load in real time, a multivariable time sequence model is constructed after standardization processing, future heating demands and pipe network heat loss are predicted, and the energy-saving optimization method and system for the central heating system are provided. A control strategy including valve opening, pump frequency and heat exchange station outlet temperature is generated in combination with a multi-objective optimization algorithm, the system operates in the optimal energy-saving state through closed-loop control and online self-adaptive adjustment, meanwhile, the user comfort degree is considered, and a self-learning closed-loop mechanism of collection-prediction-optimization-feedback is achieved. The energy efficiency and the comfort degree of the central heating system are improved, the operation cost is reduced, and the system is suitable for a modern intelligent heating network.

Description

Technical Field

[0001] This invention relates to the field of centralized heating, and specifically to an energy-saving optimization method and system for centralized heating systems. Background Technology

[0002] Currently, centralized heating systems generally suffer from insufficient energy efficiency, high operating costs, and fluctuating user comfort. Traditional control methods often rely on fixed adjustment strategies or simple empirical rules, failing to respond in real time to changes in indoor load and environmental disturbances. This results in high heating energy consumption, and indoor temperature comfort is difficult to guarantee under sudden load changes or extreme weather conditions. Existing prediction- or optimization-based methods often consider only a single objective, lacking the ability to coordinate and optimize multiple system objectives and adaptive closed-loop regulation. They cannot achieve a dynamic balance between maximizing energy efficiency and user comfort, thus failing to meet the energy-saving requirements of modern intelligent centralized heating systems. Summary of the Invention

[0003] To achieve the above-mentioned objectives, the present invention provides the following technical solution: an energy-saving optimization method for a centralized heating system, comprising the following steps:

[0004] Step S1: By deploying sensors in the heating network and at the end-user side, real-time data on supply water temperature, return water temperature, flow rate, pressure, indoor temperature, ambient temperature, and heat load demand are collected. The collected data is then synchronized in time and missing values ​​are corrected to form a standardized dataset.

[0005] Step S2: Based on the standardized dataset formed in Step S1, a multivariate time series model is constructed using historical load data, real-time indoor and outdoor temperature and flow information to predict the future heating demand curve and heat load fluctuation trend, and to calculate the instantaneous heat loss and heating efficiency index of the pipeline network.

[0006] Step S3: Based on the predicted heating demand curve and heat loss, and combined with the pipeline network operation constraints, an optimization algorithm is used to generate a multivariate control strategy, including valve opening, pump operating frequency, and heat exchange station outlet temperature, forming a set of optimized control strategies.

[0007] Step S4: Apply the optimized control strategy set to each control terminal of the centralized heating system, monitor the system feedback parameters in real time, including the actual supply water temperature, return water temperature, flow rate and indoor temperature, and make dynamic adjustments through a closed-loop control algorithm to make the system operate in the optimal energy-saving state. At the same time, feed the real-time adjustment data back to step S1 to realize data closed-loop and iterative optimization.

[0008] Preferably, step S1 further includes:

[0009] Temperature, pressure, and flow sensors are installed in the main pipeline network, secondary heat exchange stations, and user terminals of the centralized heating system. A unified clock synchronization mechanism ensures that the timestamps of all collected data are consistent, forming the original data set. The following preprocessing operations are then performed on the original data:

[0010] Missing value correction is performed using a spatiotemporal weighted interpolation algorithm, which combines historical trends and data from adjacent measurement points to repair missing points.

[0011] Standardization processing involves normalizing the corrected data according to feature dimensions to form a standardized dataset, providing a unified data foundation for subsequent prediction and optimization.

[0012] Preferably, step S2 further includes:

[0013] Using the standardized dataset obtained in step S1, a multi-time-series input feature matrix is ​​constructed and input into an improved two-layer neural network model for heating demand prediction and energy efficiency estimation.

[0014] First, a long short-term memory network is used to model the input data in time series to extract the long-term and short-term trends of heat load. The hidden state is updated through a gating structure to achieve dynamic retention and forgetting of features at different time steps.

[0015] Secondly, an attention mechanism is superimposed on the output layer to calculate weights based on the contribution of different features to load changes, thereby enhancing load features at critical moments.

[0016] The predicted heating demand curve is finally obtained, and the instantaneous heat loss and heating efficiency index of the pipeline network are calculated accordingly.

[0017] Preferably, step S3 further includes:

[0018] Based on the predicted heating demand curve, instantaneous heat loss of the pipeline network, and heating efficiency indicators, a multi-objective optimization model is constructed to achieve a dynamic balance between energy efficiency and user comfort.

[0019] Define an optimization objective function with energy efficiency deviation and room temperature deviation as constraints, and complete data optimization by combining the constraints of the heating system. The constraints of the heating system include circulating water flow rate, outlet water temperature, and upper and lower limits of pump operating frequency.

[0020] A multi-objective genetic algorithm is used to solve the problem, including encoding, fitness calculation, crossover and mutation, selection and update steps. A dual-elite retention mechanism is used to maintain a balance between energy efficiency and comfort, and finally an optimized set of control strategies is generated.

[0021] Preferably, step S4 further includes:

[0022] The control strategy set optimized in step S3 is sent to the main control valve of the heat exchange station, the variable frequency circulating pump and the user-end control valve. The control parameters are adjusted step by step according to the time sequence to ensure the stable operation of the system.

[0023] Real-time collection of feedback data is compared with predicted data to calculate the deviation vector, and the control strategy is dynamically corrected based on the online learning model predictive control algorithm.

[0024] When encountering sudden load changes and environmental disturbances during operation, the control strategy is quickly updated to maintain energy efficiency and user comfort. At the same time, the corrected control strategy is sent back to the data acquisition module to realize a closed-loop self-learning and self-optimization of "acquisition-prediction-optimization-feedback".

[0025] This invention also provides an energy-saving optimization system for a centralized heating system, comprising the following modules:

[0026] The data acquisition and processing module is used to deploy sensors in the centralized heating network and user terminals to collect data on supply water temperature, return water temperature, flow rate, pressure, indoor temperature, ambient temperature and heat load demand in real time. It also performs missing value correction, time synchronization and standardization on the collected data to form a unified dataset, providing basic data for system prediction and optimization.

[0027] The optimization and prediction module is used to construct a multivariate time series model based on the standardized dataset provided by the data acquisition and processing module, predict the future heating demand curve and pipeline heat loss, calculate the system energy efficiency index, and generate the optimal control strategy by using a multi-objective optimization algorithm in combination with heating constraints, including valve opening, pump operating frequency and heat exchange station outlet temperature, to achieve a dynamic balance between energy efficiency and user comfort in the heating system.

[0028] The control execution and adaptive feedback module is used to distribute the control strategies generated by the optimization and prediction module to the heat exchange station, circulating pump and terminal control valve, monitor the system operating parameters in real time, make dynamic adjustments through closed-loop control algorithm, and feed back the real-time operating data to the data acquisition and processing module to realize the system's self-learning, self-optimization and closed-loop iterative operation.

[0029] The data acquisition and processing module specifically includes:

[0030] The sensor deployment unit is used to deploy all temperature, pressure, and flow sensors in the main pipeline network, secondary heat exchange stations, and user terminals. This ensures that the system can acquire real-time information on supply water temperature, return water temperature, pipeline flow, pump pressure, indoor temperature, and external ambient temperature at each node. Through a unified clock synchronization mechanism, it ensures that all collected data have a consistent timestamp, avoiding data misalignment between different sensors. This provides a high-precision and consistent data foundation for subsequent prediction and optimization of the system. At the same time, this unit supports sensor self-testing and fault alarm functions to ensure the reliability and continuity of data acquisition.

[0031] The data preprocessing unit performs missing value correction, outlier detection and repair, noise suppression, and standardization on the raw data collected by the sensors. It converts the raw data into a unified data format to form a standardized dataset. The missing value correction uses a spatiotemporal weighted interpolation algorithm combined with historical trends and adjacent measurement point data for repair. Outlier detection removes outliers by setting upper and lower thresholds and trend analysis. All processed data is normalized and standardized to meet the requirements of subsequent prediction models for consistency and uniformity of input data. At the same time, this unit stores the processing results in real time and provides an interface for the optimization and prediction modules to call.

[0032] The optimization and prediction module specifically includes:

[0033] The multivariate prediction unit is used to construct a multivariate time series input matrix using the standardized dataset provided by the data acquisition and processing module, and to predict future heating demand and heat load through an improved two-layer neural network model. First, the LSTM unit is used to perform time series modeling on historical and real-time data to extract the long-term and short-term trends of heat load. At the same time, an attention mechanism is superimposed on the output layer to enhance the feature weights at key moments. The prediction results include the future heating demand curve, instantaneous heat loss of the pipeline network and heating efficiency indicators, providing accurate reference for subsequent optimization control.

[0034] The multi-objective optimization unit is used to construct a multi-objective optimization model based on the prediction results provided by the multivariate prediction unit and the constraints of the heating system. The optimization objectives are to maximize energy efficiency and minimize room temperature comfort. A multi-objective genetic algorithm is used to solve the model. During the optimization process, the control parameters, such as valve opening, pump frequency, and heat exchange station outlet temperature, are first encoded into chromosome vectors. Then, fitness calculation, crossover and mutation, selection and updating are performed. A dual elite retention mechanism is adopted to ensure the balance between the two objectives of energy efficiency and comfort. After optimization, the optimal control strategy set is output, and the control execution module is provided with the basis for the next adjustment. At the same time, dynamic weight adjustment is supported to adapt to seasonal and building load changes.

[0035] The control execution and adaptive feedback module specifically includes:

[0036] The control execution unit is used to send the control strategy generated by the optimization and prediction module to the main control valve of the heat exchange station, the variable frequency circulating pump and the user terminal control valve, and to gradually adjust each control parameter according to the time sequence. During the execution process, it combines real-time sensor data to ensure a stable system response and avoid pipeline shock and temperature fluctuation caused by sudden changes in control parameters.

[0037] The adaptive feedback unit monitors system feedback data in real time, including supply and return water temperature, flow rate, pressure, and user-end room temperature. It forms a feedback dataset and compares it with predicted data, calculates the deviation vector, and dynamically corrects the control strategy through a closed-loop control algorithm combined with an online-learned adaptive gain matrix. This enables the system to quickly and adaptively adjust under conditions of sudden load changes and rapid drops in outdoor temperature. Simultaneously, it transmits the operating data back to the data acquisition and processing module to update training samples and standardized datasets, forming a closed-loop self-evolution mechanism of "acquisition-prediction-optimization-feedback," enabling the system to have long-term self-learning and self-optimization capabilities.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0039] This invention achieves comprehensive perception of the operating status of the heating system by deploying sensors in the heating pipeline network and at the user end and performing data standardization processing, providing high-precision basic data for prediction and optimization, thereby improving the overall energy efficiency and operational reliability of the system.

[0040] By employing multivariate time series prediction and multi-objective optimization algorithms, this invention can achieve a dynamic balance between maximizing energy efficiency and user comfort, generate optimal control strategies in real time, significantly reduce heat loss, and save energy consumption.

[0041] This invention employs a closed-loop adaptive feedback and online learning control mechanism to achieve iterative optimization of "acquisition-prediction-optimization-feedback", enabling the system to adapt to load fluctuations and environmental disturbances, maintain heating stability, and possess long-term self-learning and self-optimization capabilities. Attached Figure Description

[0042] Figure 1 A flowchart illustrating the method steps provided in this application;

[0043] Figure 2 A schematic diagram of the system modules provided in this application. Detailed Implementation

[0044] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0045] refer to Figure 1 This invention provides an energy-saving optimization method for a centralized heating system, comprising the following steps:

[0046] Step 1: Collect water supply temperature in real time by installing sensors in the heating network and at the end-user side. Return water temperature ,flow ,pressure Indoor temperature Ambient temperature and heat load demand data The collected data is then synchronized with the time data and missing values ​​are corrected to form a standardized dataset. ;

[0047] Step 2: Based on the dataset Using historical load data, real-time indoor and outdoor temperature and flow information, a multivariate time series model is constructed to predict future heating demand curves. And the trend of heat load fluctuation, and calculate the instantaneous heat loss of the pipeline network. and heating efficiency indicators ;

[0048] Step 3: Based on the heating demand curve and heat loss Based on the constraints of pipeline network operation, an optimization algorithm is used to generate a multivariate control strategy, including valve opening. Pump operating frequency Heat exchange station outlet temperature To form an optimized set of control strategies ;

[0049] Step 4: Optimize the set of control strategies It is used in various control terminals of centralized heating systems to monitor system feedback parameters in real time, including the actual water supply temperature. Return water temperature ,flow and indoor temperature The system is dynamically adjusted through a closed-loop control algorithm to operate in the optimal energy-saving state, while real-time adjustment data is fed back to step one to achieve data closure and iterative optimization.

[0050] In step one, all temperature, pressure, and flow sensors are installed in the main pipeline network, secondary heat exchange stations, and user terminals of the centralized heating system. A unified clock synchronization mechanism ensures that the timestamps of all collected data are consistent, forming the original data set. :

[0051] ;

[0052] in, Indicates the water supply temperature. This indicates the return water temperature, which is derived from the pipe network temperature sensor.

[0053] The instantaneous flow rate is derived from the flow meter.

[0054] The pipeline pressure is derived from a pressure sensor.

[0055] For the user's indoor temperature, The outdoor ambient temperature is obtained from real-time data feedback from the user terminal and the weather station, respectively.

[0056] The real-time heat load is calculated by a heat meter.

[0057] and the original collected dataset Perform the following preprocessing operations:

[0058] Missing value correction employs a spatiotemporal weighted interpolation algorithm, combining historical trends and data from adjacent measurement points to repair missing points.

[0059] ;

[0060] in, This represents the current time point where missing values ​​exist. For the time index of the historical moment involved in the interpolation, The weighting coefficient represents For missing moments The weighted influence coefficient satisfies =1;

[0061] Standardization involves normalizing the corrected data according to its feature dimensions to obtain a standardized dataset. :

[0062] ;

[0063] in, , These are the historical mean and standard deviation, respectively.

[0064] In step two, the standardized dataset obtained in step one is used... Establish a multi-time series input feature matrix :

[0065] ;

[0066] An improved two-layer neural network model is used to input this matrix to achieve energy consumption prediction and energy efficiency estimation of a centralized heating system under dynamic load conditions. First, LSTM is used to perform time series modeling on the input data. Each LSTM unit consists of an input gate, a forget gate, and an output gate. By controlling the update of the gate structure weight parameter matrix, the dynamic retention and forgetting of features at different time steps are achieved, thereby extracting the long-term and short-term trends of heat load and hiding the state. The updated formula is as follows:

[0067] ;

[0068] in, , This is the weight matrix. For bias terms, It is a non-linear activation function.

[0069] Secondly, to improve the model's response to complex environmental disturbances (such as sudden temperature changes or fluctuations in pipeline pressure), an attention mechanism layer is superimposed on the LSTM output layer. The attention mechanism layer calculates attention weights based on the contribution of different features to load changes. The formula is:

[0070] ;

[0071] ;

[0072] in, As an intermediate variable in attention calculation, For LSTM in the first The hidden state at each point in time;

[0073] This mechanism enables feature enhancement at critical moments (such as peak load periods), ultimately resulting in a heating demand curve. for:

[0074] ;

[0075] Instantaneous heat loss of the pipe network is calculated based on predicted values ​​and environmental parameters. :

[0076] ;

[0077] in, The heat transfer coefficient of the pipe section is... The exposed area of ​​the pipe section;

[0078] Finally, to evaluate the overall operating efficiency of the system, based on the heating demand curve... Instantaneous heat loss of the pipeline network Calculate heating efficiency index :

[0079] .

[0080] In step three, based on the heating demand curve Instantaneous heat loss in pipeline network and heating efficiency indicators Real-time analysis of indicators such as energy efficiency and user comfort is used to construct a multi-objective optimization model to achieve a dynamic balance between energy efficiency and user comfort in the heating system. First, the optimization objective function is defined:

[0081] ;

[0082] in, This reflects the system's energy efficiency deviation. Represents the absolute value of the room temperature deviation. To set the temperature, and To adjust the weights, they can be updated adaptively based on season, building density, or policy guidance;

[0083] At the same time, the following conditions are met under the constraints:

[0084] ;

[0085] in, For circulating water flow rate, For the outlet temperature, For pump frequency, , These are the upper and lower limits of the circulating water flow rate. , These are the upper and lower limits of the outlet water temperature. , These are the upper and lower limits of the pump frequency;

[0086] To avoid traditional single-objective algorithms getting trapped in local optima, a multi-objective genetic algorithm is used for solving the problem. The multi-objective genetic algorithm is based on a dual-elite retention mechanism, which simultaneously retains individuals with the best energy efficiency and those with the best comfort level to ensure the diversity of the search space. The optimization steps include:

[0087] (1) In the coding stage, the set of optimized control strategies will be used. Convert to a chromosome vector;

[0088] (2) Fitness calculation to optimize the objective function To serve as evaluation indicators, a normalization strategy is used to achieve a unified evaluation of objectives with different dimensions.

[0089] (3) Crossover and mutation, using adaptive crossover probability With the probability of mutation And dynamically adjust with the number of iterations:

[0090] ;

[0091] ;

[0092] in, This represents the current iteration number. This represents the maximum number of iterations.

[0093] (4) Selection and updating: The uniformity of the solution set is maintained by non-dominated sorting and crowding distance calculation;

[0094] After multiple iterations, the set of optimized control strategies generated by the algorithm converges is denoted as . , Including valve opening Pump operating frequency and heat exchange station outlet temperature .

[0095] In step four, based on the optimization results of step three, the system will optimize the set of control strategies. The control terminals sent to the centralized heating network, including the main control valves of the heat exchange station, the variable frequency circulating pump controller, and the user terminal control valves, are updated step by step by the execution layer according to the time sequence to ensure the stability of the entire system response.

[0096] During control execution, the sensor network collects real-time data on supply and return water temperature, flow rate, pressure, and user-end temperature to form a feedback dataset.

[0097] ;

[0098] The system establishes a mapping table between predicted and actual operational data in the scheduling center and calculates the deviation vector. Based on this, the accuracy and response rate of the system control execution can be determined using the following formula:

[0099] ;

[0100] in, To predict heat load data;

[0101] To achieve continuous adaptive adjustment, an online learning-based model predictive control algorithm is adopted. This algorithm uses a system state prediction model as its core and calculates the rolling optimal value of the future control variable within each sampling period.

[0102] ;

[0103] in, The adaptive gain matrix is ​​continuously updated based on the historical trend of the control error. This matrix is ​​iteratively optimized using gradient descent.

[0104] ;

[0105] The learning rate can be dynamically adjusted to balance response speed and system stability.

[0106] During operation, when nonlinear disturbances such as sudden changes in heat load or a sharp drop in outdoor temperature are detected, the system quickly updates the gain matrix. To achieve pre-adjustment and avoid a sudden drop in energy efficiency or a decrease in user comfort, in addition, a modified set of control strategies is implemented. It will be sent back to the data acquisition module to update the training samples and the standardized dataset. This forms a closed-loop self-evolutionary mechanism of "collection-prediction-optimization-feedback", enabling the system to have self-learning and self-optimization capabilities in long-term operation.

[0107] refer to Figure 2 This invention provides an energy-saving optimization system for a centralized heating system, comprising the following modules:

[0108] The data acquisition and processing module is used to deploy sensors in the centralized heating network and user terminals to collect data on supply water temperature, return water temperature, flow rate, pressure, indoor temperature, ambient temperature and heat load demand in real time. It also performs missing value correction, time synchronization and standardization on the collected data to form a unified dataset, providing basic data for system prediction and optimization.

[0109] The optimization and prediction module is used to construct a multivariate time series model based on the standardized dataset provided by the data acquisition and processing module, predict the future heating demand curve and pipeline heat loss, calculate the system energy efficiency index, and generate the optimal control strategy by using a multi-objective optimization algorithm in combination with heating constraints, including valve opening, pump operating frequency and heat exchange station outlet temperature, to achieve a dynamic balance between energy efficiency and user comfort in the heating system.

[0110] The control execution and adaptive feedback module is used to distribute the control strategies generated by the optimization and prediction module to the heat exchange station, circulating pump and terminal control valve, monitor the system operating parameters in real time, make dynamic adjustments through closed-loop control algorithm, and feed back the real-time operating data to the data acquisition and processing module to realize the system's self-learning, self-optimization and closed-loop iterative operation.

[0111] The data acquisition and processing module specifically includes:

[0112] The sensor deployment unit is used to deploy all temperature, pressure, and flow sensors in the main pipeline network, secondary heat exchange stations, and user terminals. This ensures that the system can acquire real-time information on supply water temperature, return water temperature, pipeline flow, pump pressure, indoor temperature, and external ambient temperature at each node. Through a unified clock synchronization mechanism, it ensures that all collected data have a consistent timestamp, avoiding data misalignment between different sensors. This provides a high-precision and consistent data foundation for subsequent prediction and optimization of the system. At the same time, this unit supports sensor self-testing and fault alarm functions to ensure the reliability and continuity of data acquisition.

[0113] The data preprocessing unit performs missing value correction, outlier detection and repair, noise suppression, and standardization on the raw data collected by the sensors. It converts the raw data into a unified data format to form a standardized dataset. The missing value correction uses a spatiotemporal weighted interpolation algorithm combined with historical trends and adjacent measurement point data for repair. Outlier detection removes outliers by setting upper and lower thresholds and trend analysis. All processed data is normalized and standardized to meet the requirements of subsequent prediction models for consistency and uniformity of input data. At the same time, this unit stores the processing results in real time and provides an interface for the optimization and prediction modules to call.

[0114] The optimization and prediction module specifically includes:

[0115] The multivariate prediction unit is used to construct a multivariate time series input matrix using the standardized dataset provided by the data acquisition and processing module, and to predict future heating demand and heat load through an improved two-layer neural network model. First, the LSTM unit is used to perform time series modeling on historical and real-time data to extract the long-term and short-term trends of heat load. At the same time, an attention mechanism is superimposed on the output layer to enhance the feature weights at key moments. The prediction results include the future heating demand curve, instantaneous heat loss of the pipeline network and heating efficiency indicators, providing accurate reference for subsequent optimization control.

[0116] The multi-objective optimization unit is used to construct a multi-objective optimization model based on the prediction results provided by the multivariate prediction unit and the constraints of the heating system. The optimization objectives are to maximize energy efficiency and minimize room temperature comfort. A multi-objective genetic algorithm is used to solve the model. During the optimization process, the control parameters, such as valve opening, pump frequency, and heat exchange station outlet temperature, are first encoded into chromosome vectors. Then, fitness calculation, crossover and mutation, selection and updating are performed. A dual elite retention mechanism is adopted to ensure the balance between the two objectives of energy efficiency and comfort. After optimization, the optimal control strategy set is output, and the control execution module is provided with the basis for the next adjustment. At the same time, dynamic weight adjustment is supported to adapt to seasonal and building load changes.

[0117] The control execution and adaptive feedback module specifically includes:

[0118] The control execution unit is used to send the control strategy generated by the optimization and prediction module to the main control valve of the heat exchange station, the variable frequency circulating pump and the user terminal control valve, and to gradually adjust each control parameter according to the time sequence. During the execution process, it combines real-time sensor data to ensure a stable system response and avoid pipeline shock and temperature fluctuation caused by sudden changes in control parameters.

[0119] The adaptive feedback unit monitors system feedback data in real time, including supply and return water temperature, flow rate, pressure, and user-end room temperature. It forms a feedback dataset and compares it with predicted data, calculates the deviation vector, and dynamically corrects the control strategy through a closed-loop control algorithm combined with an online-learned adaptive gain matrix. This enables the system to quickly adapt and adjust under disturbances such as sudden load changes and rapid drops in outdoor temperature. Simultaneously, the unit transmits the operating data back to the data acquisition and processing module to update training samples and standardized datasets, forming a closed-loop self-evolution mechanism of "acquisition-prediction-optimization-feedback," which enables the system to have long-term self-learning and self-optimization capabilities.

[0120] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the embodiments of the present invention can be combined with each other.

[0121] Obviously, the embodiments described above are merely some embodiments of the present invention, not all embodiments. The accompanying drawings show preferred embodiments of the present invention, but do not limit the patent scope of the present invention. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of this invention.

Claims

1. An energy-saving optimization method for a centralized heating system, characterized in that, Includes the following steps: Step S1: By deploying sensors in the heating network and at the end-user side, real-time data on supply water temperature, return water temperature, flow rate, pressure, indoor temperature, ambient temperature, and heat load demand are collected. The collected data is then synchronized in time and missing values ​​are corrected to form a standardized dataset. Step S2: Based on the standardized dataset formed in Step S1, a multivariate time series model is constructed using historical load data, real-time indoor and outdoor temperature and flow information to predict the future heating demand curve and heat load fluctuation trend, and to calculate the instantaneous heat loss and heating efficiency index of the pipeline network. Step S3: Based on the predicted heating demand curve and heat loss, and combined with the pipeline network operation constraints, an optimization algorithm is used to generate a multivariate control strategy, including valve opening, pump operating frequency, and heat exchange station outlet temperature, forming a set of optimized control strategies. Step S4: Apply the optimized control strategy set to each control terminal of the centralized heating system, monitor the system feedback parameters in real time, including the actual supply water temperature, return water temperature, flow rate and indoor temperature, and make dynamic adjustments through a closed-loop control algorithm to make the system operate in the optimal energy-saving state. At the same time, feed the real-time adjustment data back to step S1 to realize data closed-loop and iterative optimization.

2. The energy-saving optimization method for a centralized heating system according to claim 1, characterized in that, Step S1 further includes: Temperature, pressure, and flow sensors are installed in the main pipeline network, secondary heat exchange stations, and user terminals of the centralized heating system. A unified clock synchronization mechanism ensures that the timestamps of all collected data are consistent, forming the original data set. The following preprocessing operations are then performed on the original data: Missing value correction is performed using a spatiotemporal weighted interpolation algorithm, which combines historical trends and data from adjacent measurement points to repair missing points. Standardization processing involves normalizing the corrected data according to feature dimensions to form a standardized dataset, providing a unified data foundation for subsequent prediction and optimization.

3. The energy-saving optimization method for a centralized heating system according to claim 1, characterized in that, Step S2 further includes: Using the standardized dataset obtained in step S1, a multi-time-series input feature matrix is ​​constructed and input into an improved two-layer neural network model for heating demand prediction and energy efficiency estimation. First, a long short-term memory network is used to model the input data in time series to extract the long-term and short-term trends of heat load. The hidden state is updated through a gating structure to achieve dynamic retention and forgetting of features at different time steps. Secondly, an attention mechanism is superimposed on the output layer to calculate weights based on the contribution of different features to load changes, thereby enhancing load features at critical moments. The predicted heating demand curve is finally obtained, and the instantaneous heat loss and heating efficiency index of the pipeline network are calculated accordingly.

4. The energy-saving optimization method for a centralized heating system according to claim 1, characterized in that, Step S3 further includes: Based on the predicted heating demand curve, instantaneous heat loss of the pipeline network, and heating efficiency indicators, a multi-objective optimization model is constructed to achieve a dynamic balance between energy efficiency and user comfort. Define an optimization objective function with energy efficiency deviation and room temperature deviation as constraints, and complete data optimization by combining the constraints of the heating system. The constraints of the heating system include circulating water flow rate, outlet water temperature, and upper and lower limits of pump operating frequency. A multi-objective genetic algorithm is used to solve the problem, including encoding, fitness calculation, crossover and mutation, selection and update steps. A dual-elite retention mechanism is used to maintain a balance between energy efficiency and comfort, and finally an optimized set of control strategies is generated.

5. The energy-saving optimization method for a centralized heating system according to claim 1, characterized in that, Step S4 further includes: The control strategy set optimized in step S3 is sent to the main control valve of the heat exchange station, the variable frequency circulating pump and the user-end control valve. The control parameters are adjusted step by step according to the time sequence to ensure the stable operation of the system. Real-time collection of feedback data is compared with predicted data to calculate the deviation vector, and the control strategy is dynamically corrected based on the online learning model predictive control algorithm. When encountering sudden load changes and environmental disturbances during operation, the control strategy is quickly updated to maintain energy efficiency and user comfort. At the same time, the corrected control strategy is sent back to the data acquisition module to realize a closed-loop self-learning and self-optimization of "acquisition-prediction-optimization-feedback".

6. An energy-saving optimization system for a centralized heating system, characterized in that, Includes the following modules: The data acquisition and processing module is used to deploy sensors in the centralized heating network and user terminals to collect data on supply water temperature, return water temperature, flow rate, pressure, indoor temperature, ambient temperature and heat load demand in real time. It also performs missing value correction, time synchronization and standardization on the collected data to form a unified dataset, providing basic data for system prediction and optimization. The optimization and prediction module is used to construct a multivariate time series model based on the standardized dataset provided by the data acquisition and processing module, predict the future heating demand curve and pipeline heat loss, calculate the system energy efficiency index, and generate the optimal control strategy by using a multi-objective optimization algorithm in combination with heating constraints, including valve opening, pump operating frequency and heat exchange station outlet temperature, to achieve a dynamic balance between energy efficiency and user comfort in the heating system. The control execution and adaptive feedback module is used to distribute the control strategies generated by the optimization and prediction module to the heat exchange station, circulating pump and terminal control valve, monitor the system operating parameters in real time, make dynamic adjustments through closed-loop control algorithm, and feed back the real-time operating data to the data acquisition and processing module to realize the system's self-learning, self-optimization and closed-loop iterative operation.

7. The energy-saving optimization system for a centralized heating system according to claim 6, characterized in that, The data acquisition and processing module specifically includes: The sensor deployment unit is used to deploy all temperature, pressure, and flow sensors in the main pipeline network, secondary heat exchange stations, and user terminals. This ensures that the system can acquire real-time information on supply water temperature, return water temperature, pipeline flow, pump pressure, indoor temperature, and external ambient temperature at each node. Through a unified clock synchronization mechanism, it ensures that all collected data have a consistent timestamp, avoiding data misalignment between different sensors. This provides a high-precision and consistent data foundation for subsequent prediction and optimization of the system. At the same time, this unit supports sensor self-testing and fault alarm functions to ensure the reliability and continuity of data acquisition. The data preprocessing unit performs missing value correction, outlier detection and repair, noise suppression, and standardization on the raw data collected by the sensors. It converts the raw data into a unified data format to form a standardized dataset. The missing value correction uses a spatiotemporal weighted interpolation algorithm combined with historical trends and adjacent measurement point data for repair. Outlier detection removes outliers by setting upper and lower thresholds and trend analysis. All processed data is normalized and standardized to meet the requirements of subsequent prediction models for consistency and uniformity of input data. At the same time, this unit stores the processing results in real time and provides an interface for the optimization and prediction modules to call.

8. The energy-saving optimization system for a centralized heating system according to claim 6, characterized in that, The optimization and prediction module specifically includes: The multivariate prediction unit is used to construct a multivariate time series input matrix using the standardized dataset provided by the data acquisition and processing module, and to predict future heating demand and heat load through an improved two-layer neural network model. First, the LSTM unit is used to perform time series modeling on historical and real-time data to extract the long-term and short-term trends of heat load. At the same time, an attention mechanism is superimposed on the output layer to enhance the feature weights at key moments. The prediction results include the future heating demand curve, instantaneous heat loss of the pipeline network and heating efficiency indicators, providing accurate reference for subsequent optimization control. The multi-objective optimization unit is used to construct a multi-objective optimization model based on the prediction results provided by the multivariate prediction unit and the constraints of the heating system. The optimization objectives are to maximize energy efficiency and minimize room temperature comfort. A multi-objective genetic algorithm is used to solve the model. During the optimization process, the control parameters, such as valve opening, pump frequency, and heat exchange station outlet temperature, are first encoded into chromosome vectors. Then, fitness calculation, crossover and mutation, selection and updating are performed. A dual elite retention mechanism is adopted to ensure the balance between the two objectives of energy efficiency and comfort. After optimization, the optimal control strategy set is output, and the control execution module is provided with the basis for the next adjustment. At the same time, dynamic weight adjustment is supported to adapt to seasonal and building load changes.

9. The energy-saving optimization system for a centralized heating system according to claim 6, characterized in that, The control execution and adaptive feedback module specifically includes: The control execution unit is used to send the control strategy generated by the optimization and prediction module to the main control valve of the heat exchange station, the variable frequency circulating pump and the user terminal control valve, and to gradually adjust each control parameter according to the time sequence. During the execution process, it combines real-time sensor data to ensure a stable system response and avoid pipeline shock and temperature fluctuation caused by sudden changes in control parameters. The adaptive feedback unit monitors system feedback data in real time, including supply and return water temperature, flow rate, pressure, and user-end room temperature. It forms a feedback dataset and compares it with predicted data, calculates the deviation vector, and dynamically corrects the control strategy through a closed-loop control algorithm combined with an online-learned adaptive gain matrix. This enables the system to quickly adapt and adjust under conditions of sudden load changes and rapid drops in outdoor temperature. Simultaneously, the unit transmits the operating data back to the data acquisition and processing module to update training samples and standardized datasets, forming a closed-loop self-evolution mechanism of "acquisition-prediction-optimization-feedback," which enables the system to have long-term self-learning and self-optimization capabilities.

Images