A method and system for optimizing and energy-saving control of operation parameters of a heating system

Abstract

The application provides a kind of heat supply system operating parameter optimization and energy-saving control method and system, belongs to heat supply system intelligent control technical field.The method comprises: establishing the digital twin model of heat supply system;Build a multi-level heat load prediction model to generate a comprehensive heat load prediction curve;The thermal inertia dynamic evaluation is carried out on the heat supply pipe network, the thermal inertia time constant of each pipe section is calculated, and the pipe network thermal inertia space-time distribution matrix is constructed;Based on the digital twin model, the comprehensive heat load prediction curve and the pipe network thermal inertia space-time distribution matrix, an optimization model is constructed;Improved multi-objective grey wolf optimization algorithm is used to solve the optimization model;A multi-dimensional energy efficiency dynamic tracking mechanism is constructed, including heat source efficiency, pipe network transmission efficiency and terminal heat utilization efficiency, and a comprehensive energy efficiency index is calculated, and the optimization model is recalculated when the comprehensive energy efficiency index is lower than the threshold or continuously decreases.The application realizes the collaborative optimization and adaptive energy-saving control of heat source-pipe network-heat exchange station-user whole chain.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for heating systems, and in particular to a method and system for optimizing and energy-saving control of operating parameters of heating systems. Background Technology

[0002] Central heating systems are a crucial component of urban energy supply, and their operational energy consumption accounts for a significant share of total building energy consumption. Improving the energy efficiency of heating system operation and achieving on-demand heating and precise regulation have always been key technical challenges in the heating industry. However, existing operation and control methods for heating systems generally suffer from technical defects such as difficulties in coordinating multi-level coupling, large inertia and long time delays leading to lag in regulation, and a disconnect between optimizing operating parameters and energy efficiency evaluation.

[0003] Heating systems comprise multiple levels, including heat sources, primary pipe networks, heat exchange stations, secondary pipe networks, and heat users. The hydraulic and thermal conditions of each level are highly coupled. Traditional control methods typically separate heat source regulation from pipe network regulation. The heat source adjusts the supply water temperature based on weather forecasts, the heat exchange station adjusts valve openings based on return water temperature, and users rely on manual adjustments. This hierarchical, independent control approach severs the inherent coupling between the various components of the system, making it difficult to achieve globally optimal operation. While existing technologies include schemes involving multi-objective optimization, their decision variables are often limited to a single level, such as optimizing only the opening of terminal valves or only the supply water temperature of the heat source, failing to establish a multi-variable collaborative optimization framework covering the entire chain of heat sources, pipe networks, heat exchange stations, and users.

[0004] Heating networks exhibit significant thermal inertia and transmission delay characteristics, with a lag of several hours or even longer between adjustments to heat source parameters and the response of users to room temperature. Existing parameter optimization methods are mostly static optimizations or rolling optimizations based on limited historical data, failing to fully exploit the energy-saving potential inherent in the thermal inertia of the heating network. This results in a severe time mismatch between heating supply and actual heat demand. While some technologies introduce a thermal inertia time constant, they are only used for single-point parameter updates of terminal valves, without establishing a spatiotemporal distribution model of thermal inertia covering the entire network. Furthermore, thermal inertia characteristics are not dynamically integrated into the search strategy of the optimization algorithm. Consequently, the population size and search step size of the optimization algorithm cannot be adaptively adjusted according to the differences in system thermal inertia. Under conditions of high thermal inertia, it is prone to getting trapped in local optima, while under conditions of low thermal inertia, the computational efficiency is low.

[0005] Current energy efficiency evaluations of heating systems often rely on single indicators such as fuel consumption at the heat source or heat loss in the heating network. They lack a multi-dimensional dynamic energy efficiency tracking system covering heat source efficiency, pipeline transmission efficiency, and end-user heat efficiency, making it impossible to identify energy efficiency weaknesses in real time and dynamically adjust optimization strategies. Furthermore, the weighting coefficients of existing optimization algorithms are typically fixed values, failing to adaptively correct for deviations between the energy efficiency levels of each stage and historical best values. This leads to the optimization algorithms gradually deviating from optimal operating conditions over long-term operation. Therefore, there is an urgent need for a heating system operation parameter optimization and energy-saving control method that can achieve collaborative optimization across the entire chain from heat source to pipeline network to heat exchange station to user, fully consider the dynamic characteristics of thermal inertia, and possess multi-dimensional adaptive energy efficiency correction capabilities. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for optimizing and controlling the operating parameters of a heating system, which solves the technical problems of existing heating systems, such as difficulties in global optimization due to multi-level independent control, mismatch between heat supply and heat demand due to long thermal inertia time delay, disconnect between optimization and evaluation due to single energy efficiency evaluation, and mismatch between optimization algorithm search strategy and system thermal inertia physical characteristics.

[0007] To achieve the above objectives, the present invention provides a method for optimizing and energy-saving control of operating parameters of a heating system, comprising the following steps: Step S1: Establish a digital twin model of the heating system; Step S2: Construct a multi-level heat load prediction model and generate a comprehensive heat load prediction curve based on the collected operational and environmental data; Step S3: Perform dynamic evaluation of the thermal inertia of the heating network, calculate the thermal inertia time constant of each pipe section, and construct a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. Step S4: Based on the digital twin model, the comprehensive heat load prediction curve, and the spatiotemporal distribution matrix of the pipeline thermal inertia, construct a multi-objective collaborative operation parameter optimization model; Step S5: Use an optimization algorithm to solve the multi-objective collaborative operation parameter optimization model. The search strategy of the optimization algorithm is dynamically adjusted according to the spatiotemporal distribution matrix of pipeline thermal inertia, and the solution set of the optimal decision variables obtained is converted into collaborative control commands and sent to the actuator. Step S6: Construct a multi-dimensional energy efficiency dynamic tracking mechanism, calculate the comprehensive energy efficiency index, and trigger the recalculation of steps S4 and S5 when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

[0008] Preferably, step S1 specifically includes: Step S11: Construct a digital twin model using a combination of mechanism modeling and data-driven approach. The digital twin model includes a heat source module, a pipe network module, a heat exchange station module, a heat user module, and a hydraulic calculation module. Step S12: Identify and dynamically optimize the roughness of the pipeline network by region: Divide the heating pipeline network into several regions. Based on the pressure and flow data collected by sensors in each region, use a genetic algorithm to identify the equivalent roughness of the pipelines in each region online, and update the identification results to the hydraulic calculation module of the digital twin model in real time.

[0009] Preferably, step S2 specifically includes: Step S21: Construct a heat source-level prediction layer and use a model combining long short-term memory network and attention mechanism to predict hourly heat load on the heat source side; Step S22: Construct a pipeline-level prediction layer and use a graph neural network to predict the heat load distribution of each node in the pipeline network; Step S23: Construct a user-level prediction layer and use a temporal convolutional network to predict the hourly heat demand of each user; Step S24: The prediction results of the heat source-level prediction layer, the pipeline-level prediction layer and the user-level prediction layer are fused through a confidence-based dynamic weight allocation mechanism to generate a comprehensive heat load prediction curve.

[0010] Preferably, step S3 specifically includes: Step S31, calculate the first... Thermal inertia time constant of pipe segment ,in, Density of the heat transfer medium; Specific heat capacity of the heat transfer medium; For the first The volume of the pipe section; For the first The overall heat transfer coefficient of the pipe section; For the first The heat dissipation area of ​​the pipe section; Step S32: Based on the thermal inertia time constant of each pipe segment Construct a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. This matrix is ​​denoted as... , its first Line number Column elements Indicates the first The spatial location of each pipe section The thermal inertia time constant at that location.

[0011] Preferably, step S4 specifically includes: Step S41: Set the optimization objectives of the multi-objective collaborative operation parameter optimization model as total system operating energy consumption, heating cost and user thermal comfort satisfaction, and the decision variables as heat source water supply temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user-end valve opening. Step S42: The objective function of the multi-objective cooperative operation parameter optimization model is: ; in, The objective function vector; For decision variable vectors; This is the total operating energy consumption function of the system; This is a function representing the operating cost of the heating system. For user thermal comfort deviation function; This represents the vector transpose operation; ; in, It is a time variable; For heat source at all times Instantaneous power; For the first The internal circulation pump of each heat exchange station is at a constant time Instantaneous power; For the first The regulating valves in each heat exchange station are at a constant time Instantaneous power; ; in, Unit fuel cost; For a moment Fuel consumption rate; The unit price of electricity; For a moment Power consumption; ; in, For the first individual users at any time The measured indoor temperature; For the first The target temperature for each user; Decision variable vector ; in: The temperature of the water supplied to the heat source; The heat source circulation flow rate; For the first Primary valve opening of each heat exchange station; For the first Frequency of the secondary circulation pump in each heat exchange station; For the first Individual user-side valve opening; Step S43: Set constraints, including heat supply and demand balance constraints, equipment operation safety constraints, hydraulic operating condition stability constraints, and grid time-of-use electricity price constraints. The grid time-of-use electricity price constraint is: during peak electricity price periods, priority should be given to using the thermal inertia of the pipeline network for heat storage regulation, and during off-peak electricity price periods, the output of the heat source should be increased.

[0012] Preferably, step S5 specifically includes: Step S51: The improved multi-objective gray wolf optimization algorithm is adopted as the optimization algorithm; Step S52: Based on the spatiotemporal distribution matrix of the pipeline thermal inertia obtained in step S32 The population size and search step size of the improved multi-objective gray wolf optimization algorithm are dynamically adjusted as follows: Define the average thermal inertia time constant of the entire network. ,in The average thermal inertia time constant of the entire network; This represents the total number of pipeline segments. Setting a high thermal inertia threshold and low thermal inertia threshold ,when At that time, population size Search step size ;when At that time, population size Search step size ; in, This refers to the dynamically adjusted population size. This is the dynamically adjusted search step size; Used as the baseline population size; Used as the baseline search step size; , , , This is the preset adjustment coefficient; Step S53: Introduce an elite retention mechanism with Pareto optimal front into the improved multi-objective gray wolf optimization algorithm: After each generation of evolution, store the non-dominated solutions in the current population into the elite solution set. Set the maximum capacity of the elite solution set to be ,when The number of solutions exceeds At that time, calculate the congestion distance for each non-dominated solution. Delete the solution with the minimum crowding distance until ; Step S54: Introduce an adaptive weight adjustment mechanism into the improved multi-objective gray wolf optimization algorithm: Define the first... The adaptive weight coefficients for the three objective functions are: ; ; ; in, , , The first The adaptive weighting coefficients for the three objective functions; is the base of the natural logarithm; , , This is the preset sensitivity coefficient; , , These are the normalized distances from the current solution to the Pareto front in the three objective dimensions; Step S55: Convert the optimal decision variable solution set obtained from the solution into a coordinated control command and send it to the execution agency.

[0013] Preferably, step S6 specifically includes: Step S61: Calculate the heat source efficiency ,in, This refers to the actual heat output of the heat source; To input heat into the fuel; Calculate pipeline transportation efficiency ,in, Provide heat to the heat source; For heat loss in the pipeline network; Calculate the terminal thermal efficiency ,in, Provide users with effective heat utilization; To output heat to the secondary side of the heat exchange station; Step S62: Calculate the overall energy efficiency index of the system. ,in, , , The adaptive weighting coefficients are dynamically adjusted according to the following formula: ; ; ; in, This represents the historical best value for heat source efficiency. This represents the historical best value for pipeline transportation efficiency. This represents the historical best value for end-use thermal efficiency; Step S63: Set the comprehensive energy efficiency index threshold. ,when or When the control decreases for three consecutive control cycles, the recalculation of steps S4 and S5 is triggered.

[0014] Preferably, the confidence-based dynamic weight allocation mechanism in step S24 specifically includes: Step S241: Obtain the past prediction results of the heat source-level prediction layer, the pipeline-level prediction layer, and the user-level prediction layer. The average prediction error for each control cycle is denoted as . , , ; Step S242: Dynamically adjust the weight coefficients of the prediction results for each layer according to the following formula: ; ; ; in, The weighting coefficients for the heat source-level prediction results; The weighting coefficients for the pipeline-level prediction results; These are the weighting coefficients for user-level prediction results.

[0015] This invention also provides a heating system operating parameter optimization and energy-saving control system, used to execute the above-described heating system operating parameter optimization and energy-saving control method, including: The digital twin model building module is used to build a digital twin model of the heating system. It adopts a combination of mechanism modeling and data-driven approach to build a digital twin model that includes a heat source module, a pipe network module, a heat exchange station module, a heat user module, and a hydraulic calculation module. It also performs zone identification and dynamic optimization of pipe network roughness. The multi-level heat load prediction module is used to construct a multi-level heat load prediction model. It generates a comprehensive heat load prediction curve based on the collected operational and environmental data, including a heat source-level prediction layer, a pipeline-level prediction layer, and a user-level prediction layer. It also integrates the prediction results of each layer through a dynamic weight allocation mechanism based on confidence level. The thermal inertia dynamic evaluation module is used to perform thermal inertia dynamic evaluation on the heating network, calculate the thermal inertia time constant of each pipe section, and construct a network thermal inertia spatiotemporal distribution matrix covering the entire network. The multi-objective collaborative optimization module is used to construct a multi-objective collaborative operation parameter optimization model based on the digital twin model, the comprehensive heat load prediction curve and the spatiotemporal distribution matrix of the pipeline thermal inertia. The optimization objectives are the total system operating energy consumption, heating cost and user thermal comfort satisfaction. The decision variables are the heat source supply water temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user-end valve opening. The module also sets constraints including heat supply and demand balance constraints, equipment operation safety constraints, hydraulic condition stability constraints and grid time-of-use electricity price constraints. The optimization algorithm solution module is used to solve the multi-objective collaborative operation parameter optimization model using an improved multi-objective gray wolf optimization algorithm. It dynamically adjusts the population size and search step size according to the spatiotemporal distribution matrix of pipeline thermal inertia, introduces the elite retention mechanism and adaptive weight adjustment mechanism of Pareto optimal frontier, and transforms the solution set of optimal decision variables obtained by the solution into collaborative control instructions and sends them to the execution mechanism. The multi-dimensional energy efficiency dynamic tracking module is used to build a multi-dimensional energy efficiency dynamic tracking mechanism that includes heat source efficiency, pipeline transmission efficiency and end-point heat consumption efficiency. It calculates the comprehensive energy efficiency index and triggers the recalculation of the multi-objective collaborative optimization module and the optimization algorithm solution module when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

[0016] Therefore, the present invention employs the above-mentioned method and system for optimizing and saving energy in heating system operating parameters, and the beneficial technical effects are as follows: (1) This invention establishes a full-chain digital twin model including heat source, pipeline network, heat exchange station and heat user, and constructs a multi-objective collaborative optimization model with heat source water supply temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user valve opening as decision variables. This realizes the collaborative control between heat source-pipeline network-heat exchange station-user, solves the problem of global optimization difficulty caused by hierarchical independent control in the prior art, and reduces the total operating energy consumption and heating cost of the system.

[0017] (2) This invention quantitatively calculates the thermal inertia time constant of each pipe section of the heating network and constructs a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. It embeds this matrix into the dynamic adjustment strategy of the population size and search step size of the multi-objective gray wolf optimization algorithm, so that the search strategy of the optimization algorithm matches the thermal inertia physical characteristics of the heating system. This effectively solves the time mismatch problem between the heat supply and the actual heat demand, and avoids the defects of easily getting trapped in local optima under the condition of large thermal inertia and low computational efficiency under the condition of small thermal inertia.

[0018] (3) This invention constructs a multi-dimensional dynamic energy efficiency tracking mechanism that includes heat source efficiency, pipeline transmission efficiency and terminal heat efficiency. It corrects the comprehensive energy efficiency index in real time through adaptive weights and automatically triggers the optimization model to recalculate when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases. At the same time, it adopts a dynamic weight allocation mechanism based on confidence to integrate multi-level heat load prediction results, realizes closed-loop adaptive correction of operating parameter optimization and energy efficiency evaluation, and solves the technical problems of single energy efficiency evaluation and disconnect between optimization and evaluation in the prior art. Attached Figure Description

[0019] Figure 1 This is a flowchart of a method for optimizing and energy-saving control of operating parameters of a heating system according to the present invention; Figure 2 Flowchart for improving the multi-objective gray wolf optimization algorithm; Figure 3 This is a flowchart for multi-dimensional dynamic energy efficiency tracking and recalculation. Detailed Implementation

[0020] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0022] Example 1 This embodiment uses a centralized heating system in a city as an example to describe the method of the present invention in detail. The heating system covers a heating area of ​​approximately 500,000 square meters, including one gas-fired boiler heat source, three heat exchange stations, a primary pipeline with a total length of approximately 8 kilometers, a secondary pipeline with a total length of approximately 15 kilometers, and serves approximately 4,000 heat users.

[0023] Reference Figures 1-3 A method for optimizing operating parameters and controlling energy conservation in a heating system includes the following steps: Step S1: Establish a digital twin model of the heating system.

[0024] Step S11: Construct a digital twin model using a combination of mechanism modeling and data-driven approach. The digital twin model includes a heat source module, a pipeline network module, a heat exchange station module, a heat user module, and a hydraulic calculation module.

[0025] (a) Heat source module.

[0026] Based on the thermodynamic characteristics of gas-fired boilers, a dynamic heat output model is established. The relationship between the boiler's instantaneous heat supply and the gas flow rate, return water temperature, and supply water temperature is expressed by the following heat balance equation: ; in, For the boiler at all times Instantaneous heat supply (kW). The boiler thermal efficiency is taken as 0.92. The mass flow rate of the gas is (kg / s). The lower heating value of the gas is taken as 35.5 MJ / kg. Meanwhile, the relationship between the heating capacity and the supply and return water temperatures, and the circulation flow rate, is as follows: ; in, The specific heat capacity of the heat transfer medium is 4187 J / (kg·℃) The density of the heat transfer medium is 983 kg / m³. 3 ), Circulating volumetric flow rate (m³) 3 / s), and These are the supply and return water temperatures (°C), respectively.

[0027] (ii) Pipeline module.

[0028] A distributed parameter model is established based on the hydraulic-thermal coupling equations. The primary and secondary pipe networks are each divided into several pipe segments, and the hydraulic characteristics of each segment are described using the Darcy-Weisbach equations. For each pipe segment... (in (Numbering the nodes at both ends of the pipe section), the pressure drop calculation formula is: ; in, For pipe section The pressure drop (Pa). This is the friction coefficient. The length of the pipe section is (m). The inner diameter of the pipe section is (m). Flow velocity (m / s). Friction coefficient. With pipe absolute roughness The relationship is expressed using the Colebrook-White formula: ; in, For pipe section The Reynolds number. The thermodynamic characteristics of the pipe section are described by a one-dimensional heat conduction equation, considering heat loss along the pipe: ; in, and These are the inlet and outlet water temperatures (°C) of the pipe section, respectively. Soil temperature (°C). The overall heat transfer coefficient of the pipe section (W / (m)) 2 ·℃)), Pipe section volumetric flow rate (m³) 3 / s).

[0029] (III) Heat exchange station module.

[0030] A dynamic heat transfer model is established based on the heat transfer equations of a plate heat exchanger. The heat transfer equations of the heat exchanger are as follows: ; in, Heat exchanger capacity (kW). The overall heat transfer coefficient (W / (m)) 2 ·℃)), For heat exchange area (m) 2 ), Logarithmic mean temperature difference (°C): ; in, , The primary side supply and return water temperature, , The secondary side supply and return water temperatures are used. The relationship between the primary side flow rate and valve opening is expressed using the regulating valve flow characteristic equation: ; in, Valve opening degree (%) This is the valve characteristic index (2.5 for equal percentage characteristics). The maximum primary flow rate (m³) when the valve is fully open 3 / s).

[0031] (iv) Hot user module.

[0032] A room temperature response model is established based on the building's thermal dynamics (equivalent heat capacity-thermal resistance model, i.e., RC model). A first-order RC model is used to describe room temperature changes: ; in, The equivalent heat capacity of the building (J / ℃). Equivalent thermal resistance of the building (°C / W). and The indoor and outdoor temperatures (°C) are respectively. This refers to the terminal heating power (W). The terminal heating power is related to the secondary side supply water temperature. The relationship between valve opening degree and valve opening degree is as follows: ; in, The overall heat transfer coefficient of the radiator (W / ℃). The valve opening degree (%) at the user end.

[0033] (v) Hydraulic calculation module.

[0034] The hydraulic calculation module uses the nodal pressure method (loop adjustment method) to calculate the flow distribution and pressure distribution of the pipe network. For systems containing... Each node For each pipe segment of the pipeline network, establish the nodal flow balance equation and the loop pressure drop balance equation: ; ; in, These are elements of the node-pipe segment association matrix (values ​​can be +1, -1, or 0). For the first Volumetric flow rate of a pipe section (m³) 3 / s), For nodes Net injection flow (m 3 / s), These are elements of the loop-pipe segment correlation matrix (values ​​can be +1, -1, or 0). For the first Pressure drop (Pa) of the pipe section. Let be the number of independent loops. The Hardy-Cross iterative method is used to solve the above equations, with a convergence accuracy set to a voltage drop error of less than 0.1%.

[0035] Step S12: Identify and dynamically optimize the roughness of the pipeline network by region: Divide the heating pipeline network into several regions. Based on the pressure and flow data collected by sensors in each region, use a genetic algorithm to identify the equivalent roughness of the pipelines in each region online, and update the identification results to the hydraulic calculation module of the digital twin model in real time.

[0036] In this embodiment, the heating network is divided into four zones, with pipes in each zone having similar materials, laying conditions, and service life. The basic parameters of the pipes in each zone are shown in Table 1.

[0037] Table 1 Initial parameters for pipeline roughness zoning identification

[0038] A genetic algorithm is used to identify the equivalent roughness of the pipeline in each region online. The design of the genetic algorithm is as follows: (1) Decision variables: equivalent roughness of each region The search range is 0.1mm to 1.5mm.

[0039] (2) Fitness function: ; ; in, This represents the total number of pressure measurement points. For the first The measured pressure values ​​at each measuring point This is the simulated pressure value calculated by the hydraulic calculation module based on the current roughness parameters.

[0040] (3) Genetic algorithm parameters: population size 50, real number encoding, roulette wheel selection, simulated binary crossover (SBX) probability of 0.85, polynomial mutation probability of 0.1, termination condition is that the optimal fitness change is less than 10-5 for 20 consecutive generations or the maximum number of iterations is reached 100.

[0041] (4) The identification results are shown in Table 2.

[0042] Table 2. Pipeline roughness zoning identification results

[0043] The identification results are updated in real time to the roughness parameter table of the hydraulic calculation module, which improves the accuracy of hydraulic simulation by about 12%.

[0044] Step S2: Construct a multi-level heat load prediction model and generate a comprehensive heat load prediction curve based on the collected operational and environmental data.

[0045] Step S21: Construct a heat source-level prediction layer and use a model combining long short-term memory network and attention mechanism to predict hourly heat load on the heat source side.

[0046] The structural parameters of the model combining long short-term memory networks and attention mechanisms are as follows: Input features: outdoor temperature sequence of the past 72 hours, historical heat load sequence, and weather forecast data (outdoor temperature, wind speed, and solar radiation intensity for the next 24 hours); Input time step: 72 (hours); Number of LSTM layers: 2; Number of neurons in the first LSTM layer: 128; Number of neurons in the second LSTM layer: 64; Dropout ratio: 0.2 (to prevent overfitting); Attention mechanism: Additive Attention, attention dimension 32; Output layer: Fully connected layer, number of neurons 24 (corresponding to hourly heat load for the next 24 hours); Activation function: tanh for LSTM layers, linear activation for the output layer; Loss function: Mean squared error (MSE); Optimizer: Adam, learning rate 0.001; Batch size: 32; Number of training epochs: 200 epochs (20 epochs for early stopping).

[0047] Step S22: Construct a pipeline-level prediction layer and use a graph neural network to predict the heat load distribution of each node in the pipeline network.

[0048] The parameters of the graph neural network structure are as follows: Graph structure: 3 nodes (corresponding to 3 heat exchange stations), edge number equal to the number of pipe connections; Node features: historical heat load (72 time points), primary side supply water temperature, and return water temperature for each heat exchange station; Edge features: pipe length, pipe diameter, and heat loss coefficient; GNN layers: 3; Hidden dimensions per layer: 128→64→32; Aggregation function: Mean Aggregation; Update function: Gated Recurrent Unit (GRU); Output layer: Each node outputs the hourly heat load for the next 24 hours; Loss function: Mean Squared Error (MSE); Optimizer: Adam, learning rate 0.0005; Batch size: 16; Number of training rounds: 150.

[0049] Step S23: Construct a user-level prediction layer and use a temporal convolutional network to predict the hourly heat demand of each user.

[0050] The parameters of the temporal convolutional network structure are as follows: Input features: Indoor temperature, valve opening, and outdoor temperature for each user over the past 48 hours; Input time step: 48; Number of dilated convolutional layers: 4 layers, with dilation rates of 1, 2, 4, and 8 respectively; Number of kernels per layer: 64; Kernel size: 3; Residual connections: Residual blocks are added after each layer; Output layer: Global average pooling + fully connected layer (24 neurons); Activation function: ReLU; Loss function: Mean Absolute Error (MAE); Optimizer: Adam, learning rate 0.001; Batch size: 128; Number of training epochs: 100.

[0051] Step S24: The prediction results of the heat source-level prediction layer, the pipeline-level prediction layer and the user-level prediction layer are fused through a confidence-based dynamic weight allocation mechanism to generate a comprehensive heat load prediction curve.

[0052] The confidence-based dynamic weight allocation mechanism in step S24 specifically includes: Step S241: Obtain the past prediction results of the heat source-level prediction layer, the pipeline-level prediction layer, and the user-level prediction layer. The average prediction error over 30 control cycles is denoted as follows: , , In this embodiment, the values ​​are 4.2%, 5.8%, and 7.5%, respectively.

[0053] Step S242: Dynamically adjust the weight coefficients of the prediction results for each layer according to the following formula: ; ; ; in, The weighting coefficients for the heat source-level prediction results; The weighting coefficients for the pipeline-level prediction results; These are the weighting coefficients for user-level prediction results.

[0054] Step S3: Perform dynamic evaluation of thermal inertia of the heating network, calculate the thermal inertia time constant of each pipe section, and construct a spatiotemporal distribution matrix of network thermal inertia covering the entire network.

[0055] Step S31, calculate the first... Thermal inertia time constant of pipe segment Taking a typical pipe segment in a primary pipeline network as an example, (983kg / m 3 () represents the density of the heat transfer medium; (4187 J / (kg.℃)) is the specific heat capacity of the heat transfer medium; (120m) 3 ) is the first The volume of the pipe section; (0.35W / (m) 2 .℃)) is the first The overall heat transfer coefficient of the pipe section; (850m) 2 ) is the first The heat dissipation area of ​​the pipe section.

[0056] Similarly, the thermal inertia time constant of each pipe section was calculated, and the results are shown in Table 3.

[0057] Table 3 Thermal inertia time constants for some pipe sections

[0058] Step S32: Based on the thermal inertia time constant of each pipe segment Construct a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. This matrix is ​​denoted as... , its first Line number Column elements Indicates the first The spatial location of each pipe section The thermal inertia time constant at the point. In this embodiment, the matrix dimension is 7×3 (7 pipe segments, each pipe segment taking 3 spatial location points). For example, the thermal inertia time constants of pipe segment P1-01 at the inlet, midpoint, and outlet are 14.8h, 15.1h, and 15.3h, respectively.

[0059] Step S4: Based on the digital twin model, the comprehensive heat load prediction curve, and the spatiotemporal distribution matrix of the pipeline thermal inertia, construct a multi-objective collaborative operation parameter optimization model.

[0060] Step S41: Set the optimization objectives of the multi-objective collaborative operation parameter optimization model as total system operating energy consumption, heating cost, and user thermal comfort satisfaction. The decision variables are heat source water supply temperature (65℃~85℃) and heat source circulation flow rate (200m³ / h). 3 / h~400m 3 / h), heat exchange station valve opening (0~100%), heat exchange station circulating pump frequency (30Hz~50Hz), and user-end valve opening (0~100%).

[0061] Step S42: The objective function of the multi-objective cooperative operation parameter optimization model is: ; in, The objective function vector; For decision variable vectors; This is the total operating energy consumption function of the system; This is a function representing the operating cost of the heating system. For user thermal comfort deviation function; This represents the vector transpose operation; ; in, It is a time variable; For heat source at all times Instantaneous power; For the first The internal circulation pump of each heat exchange station is at a constant time Instantaneous power; For the first The regulating valves in each heat exchange station are at a constant time Instantaneous power; ; in, Unit fuel cost; For a moment Fuel consumption rate; The unit price of electricity; For a moment Power consumption; in, For the first individual users at any time The measured indoor temperature; For the first The target temperature for each user; Decision variable vector ; in: The temperature of the water supplied to the heat source; The heat source circulation flow rate; For the first Primary valve opening of each heat exchange station; For the first Frequency of the secondary circulation pump in each heat exchange station; For the first Individual user-side valve opening; Step S43: Set constraints, which include: Heat supply and demand balance constraint: Heat supply from heat source = sum of heat load of each heat exchange station + heat loss of pipeline network; Equipment operation safety constraints: water supply temperature ≤ 90℃, circulating pump frequency ≤ 50Hz; Hydraulic stability constraints: Pressure differential fluctuation at each node of the pipeline network ≤ 10%; Time-of-use pricing constraints for the power grid: During peak hours (10:00-15:00, 18:00-21:00), priority is given to utilizing the thermal inertia storage of the pipeline network, while during off-peak hours (23:00-7:00 the next day), the output of heat sources is increased.

[0062] Step S5: Use an optimization algorithm to solve the multi-objective collaborative operation parameter optimization model. The search strategy of the optimization algorithm is dynamically adjusted according to the spatiotemporal distribution matrix of pipeline thermal inertia, and the optimal decision variable solution set obtained is converted into collaborative control commands and sent to the actuator.

[0063] Step S51: The improved multi-objective gray wolf optimization algorithm is adopted as the optimization algorithm.

[0064] Step S52: Based on the spatiotemporal distribution matrix of the pipeline thermal inertia obtained in step S32 The population size and search step size of the improved multi-objective gray wolf optimization algorithm are dynamically adjusted as follows: Define the average thermal inertia time constant of the entire network. ,in The average thermal inertia time constant of the entire network; This represents the total number of pipeline segments. Setting a high thermal inertia threshold and low thermal inertia threshold ,when At that time, population size Search step size ;when At that time, population size Search step size ; in, This refers to the dynamically adjusted population size. This is the dynamically adjusted search step size; The baseline population size is set to 50. The baseline search step size is set to 0.2; , , , The preset adjustment coefficient is set to 0.5.

[0065] Step S53: Introduce an elite retention mechanism with Pareto optimal front into the improved multi-objective gray wolf optimization algorithm: After each generation of evolution, store the non-dominated solutions in the current population into the elite solution set. Set the maximum capacity of the elite solution set to be (Set to 20), when The number of solutions exceeds At that time, calculate the congestion distance for each non-dominated solution. Delete the solution with the minimum crowding distance until .

[0066] Step S54: Introduce an adaptive weight adjustment mechanism into the improved multi-objective gray wolf optimization algorithm: Define the first... The adaptive weight coefficients for the three objective functions are: ; ; ; in, , , The first The adaptive weight coefficients for the three objective functions; is the base of the natural logarithm; , , The preset sensitivity coefficients are all set to 1.0; , , These represent the normalized distances from the current solution to the Pareto front in the three objective dimensions.

[0067] Step S55: Convert the optimal decision variable solution set obtained from the solution into a coordinated control command and send it to the execution agency.

[0068] After 80 iterations, a Pareto optimal solution set was obtained, one of which is a compromise optimal solution set shown in Table 4.

[0069] Table 4 Comparison of operating parameters before and after optimization

[0070] Step S6: Construct a multi-dimensional energy efficiency dynamic tracking mechanism, calculate the comprehensive energy efficiency index, and trigger the recalculation of steps S4 and S5 when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

[0071] Step S61: Calculate the heat source efficiency ,in, =285000kWh is the actual heat output of the heat source; =310,000 kWh is the heat input for fuel; Calculate pipeline transportation efficiency ,in, =285000kWh of heat source provides heat; =28500 kWh is the heat loss from the pipeline network; Calculate the terminal thermal efficiency ,in, =228000 kWh is the effective heat consumption for users; =256500kWh is the heat output from the secondary side of the heat exchange station.

[0072] Step S62: Calculate the overall energy efficiency index of the system. ,in, , , The adaptive weighting coefficients are dynamically adjusted according to the following formula: ; ; ; in, (The value is 0.93) represents the historical best value for heat source efficiency; (The value is 0.92) represents the historical best value for pipeline transportation efficiency; (The value is 0.91) is the historical best value for end-use thermal efficiency.

[0073] Step S63: Set the comprehensive energy efficiency index threshold. (Set to 0.9), when or When the control decreases for three consecutive control cycles, the recalculation of steps S4 and S5 is triggered.

[0074] Example 2 An optimization and energy-saving control system for operating parameters of a heating system, comprising: The digital twin model building module is used to build a digital twin model of the heating system. It adopts a combination of mechanism modeling and data-driven approach to build a digital twin model that includes a heat source module, a pipe network module, a heat exchange station module, a heat user module, and a hydraulic calculation module. It also performs zone identification and dynamic optimization of pipe network roughness. The multi-level heat load prediction module is used to construct a multi-level heat load prediction model. It generates a comprehensive heat load prediction curve based on the collected operational and environmental data, including a heat source-level prediction layer, a pipeline-level prediction layer, and a user-level prediction layer. It also integrates the prediction results of each layer through a dynamic weight allocation mechanism based on confidence level. The thermal inertia dynamic evaluation module is used to perform thermal inertia dynamic evaluation on the heating network, calculate the thermal inertia time constant of each pipe section, and construct a network thermal inertia spatiotemporal distribution matrix covering the entire network. The multi-objective collaborative optimization module is used to construct a multi-objective collaborative operation parameter optimization model based on the digital twin model, the comprehensive heat load prediction curve and the spatiotemporal distribution matrix of the pipeline thermal inertia. The optimization objectives are the total system operating energy consumption, heating cost and user thermal comfort satisfaction. The decision variables are the heat source supply water temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user-end valve opening. The module also sets constraints including heat supply and demand balance constraints, equipment operation safety constraints, hydraulic condition stability constraints and grid time-of-use electricity price constraints. The optimization algorithm solution module is used to solve the multi-objective collaborative operation parameter optimization model using an improved multi-objective gray wolf optimization algorithm. It dynamically adjusts the population size and search step size according to the spatiotemporal distribution matrix of pipeline thermal inertia, introduces the elite retention mechanism and adaptive weight adjustment mechanism of Pareto optimal frontier, and transforms the solution set of optimal decision variables obtained by the solution into collaborative control instructions and sends them to the execution mechanism. The multi-dimensional energy efficiency dynamic tracking module is used to build a multi-dimensional energy efficiency dynamic tracking mechanism that includes heat source efficiency, pipeline transmission efficiency and end-point heat consumption efficiency. It calculates the comprehensive energy efficiency index and triggers the recalculation of the multi-objective collaborative optimization module and the optimization algorithm solution module when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

[0075] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0076] Therefore, the present invention adopts the above-mentioned method and system for optimizing and energy-saving control of heating system operating parameters, realizing coordinated optimization and adaptive energy-saving control of the entire chain from heat source to pipeline network to heat exchange station to user, effectively improving the operating energy efficiency of heating system.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing and energy-saving control of operating parameters of a heating system, characterized in that, Includes the following steps: Step S1: Establish a digital twin model of the heating system; Step S2: Construct a multi-level heat load prediction model and generate a comprehensive heat load prediction curve based on the collected operational and environmental data; Step S3: Perform dynamic evaluation of the thermal inertia of the heating network, calculate the thermal inertia time constant of each pipe section, and construct a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. Step S4: Based on the digital twin model, the comprehensive heat load prediction curve, and the spatiotemporal distribution matrix of the pipeline thermal inertia, construct a multi-objective collaborative operation parameter optimization model; Step S5: Use an optimization algorithm to solve the multi-objective collaborative operation parameter optimization model. The search strategy of the optimization algorithm is dynamically adjusted according to the spatiotemporal distribution matrix of pipeline thermal inertia, and the solution set of the optimal decision variables obtained is converted into collaborative control commands and sent to the actuator. Step S6: Construct a multi-dimensional energy efficiency dynamic tracking mechanism, calculate the comprehensive energy efficiency index, and trigger the recalculation of steps S4 and S5 when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

2. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 1, characterized in that, Step S1 specifically includes: Step S11: Construct a digital twin model using a combination of mechanism modeling and data-driven approach. The digital twin model includes a heat source module, a pipe network module, a heat exchange station module, a heat user module, and a hydraulic calculation module. Step S12: Identify and dynamically optimize the roughness of the pipeline network by region: Divide the heating pipeline network into several regions. Based on the pressure and flow data collected by sensors in each region, use a genetic algorithm to identify the equivalent roughness of the pipelines in each region online, and update the identification results to the hydraulic calculation module of the digital twin model in real time.

3. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 1, characterized in that, Step S2 specifically includes: Step S21: Construct a heat source-level prediction layer and use a model combining long short-term memory network and attention mechanism to predict hourly heat load on the heat source side; Step S22: Construct a pipeline-level prediction layer and use a graph neural network to predict the heat load distribution of each node in the pipeline network; Step S23: Construct a user-level prediction layer and use a temporal convolutional network to predict the hourly heat demand of each user; Step S24: The prediction results of the heat source-level prediction layer, the pipeline-level prediction layer and the user-level prediction layer are fused through a confidence-based dynamic weight allocation mechanism to generate a comprehensive heat load prediction curve.

4. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 1, characterized in that, Step S3 specifically includes: Step S31, calculate the first... Thermal inertia time constant of pipe segment ,in, Density of the heat transfer medium; Specific heat capacity of the heat transfer medium; For the first The volume of the pipe section; For the first The overall heat transfer coefficient of the pipe section; For the first The heat dissipation area of ​​the pipe section; Step S32: Based on the thermal inertia time constant of each pipe segment Construct a spatiotemporal distribution matrix of the network thermal inertia covering the entire network. This matrix is ​​denoted as... , its first Line number Column elements Indicates the first The spatial location of each pipe section The thermal inertia time constant at that location.

5. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 4, characterized in that, Step S4 specifically includes: Step S41: Set the optimization objectives of the multi-objective collaborative operation parameter optimization model as total system operating energy consumption, heating cost and user thermal comfort satisfaction, and the decision variables as heat source water supply temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user-end valve opening. Step S42: The objective function of the multi-objective cooperative operation parameter optimization model is: ； in, The objective function vector; For decision variable vectors; This is the total operating energy consumption function of the system; This is a function representing the operating cost of the heating system. For user thermal comfort deviation function; This represents the vector transpose operation; ； in, It is a time variable; For heat source at all times Instantaneous power; For the first The internal circulation pump of each heat exchange station is at a constant time Instantaneous power; For the first The regulating valves in each heat exchange station are at a constant time Instantaneous power; ； in, Unit fuel cost; For a moment Fuel consumption rate; The unit price of electricity; For a moment Power consumption; ； in, For the first individual users at any time The measured indoor temperature; For the first The target temperature for each user; Decision variable vector ; in: The temperature of the water supplied to the heat source; The heat source circulation flow rate; For the first Primary valve opening of each heat exchange station; For the first Frequency of the secondary circulation pump in each heat exchange station; For the first Individual user-side valve opening; Step S43: Set constraints, including heat supply and demand balance constraints, equipment operation safety constraints, hydraulic operating condition stability constraints, and grid time-of-use electricity price constraints. The grid time-of-use electricity price constraint is: during peak electricity price periods, priority should be given to using the thermal inertia of the pipeline network for heat storage regulation, and during off-peak electricity price periods, the output of the heat source should be increased.

6. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 5, characterized in that, Step S5 specifically includes: Step S51: The improved multi-objective gray wolf optimization algorithm is adopted as the optimization algorithm; Step S52: Based on the spatiotemporal distribution matrix of the pipeline thermal inertia obtained in step S32 The population size and search step size of the improved multi-objective gray wolf optimization algorithm are dynamically adjusted as follows: Define the average thermal inertia time constant of the entire network. ,in The average thermal inertia time constant of the entire network; This represents the total number of pipeline segments. Setting a high thermal inertia threshold and low thermal inertia threshold ,when At that time, population size Search step size ;when At that time, population size Search step size ; in, This refers to the dynamically adjusted population size. This is the dynamically adjusted search step size; Used as the baseline population size; Used as the baseline search step size; , , , This is the preset adjustment coefficient; Step S53: Introduce an elite retention mechanism with Pareto optimal front into the improved multi-objective gray wolf optimization algorithm: After each generation of evolution, store the non-dominated solutions in the current population into the elite solution set. Set the maximum capacity of the elite solution set to be ,when The number of solutions exceeds At that time, calculate the congestion distance for each non-dominated solution. Delete the solution with the minimum crowding distance until ; Step S54: Introduce an adaptive weight adjustment mechanism into the improved multi-objective gray wolf optimization algorithm: Define the first... The adaptive weight coefficients for the three objective functions are: ； ； ； in, , , The first The adaptive weight coefficients for the three objective functions; is the base of the natural logarithm; , , This is the preset sensitivity coefficient; , , These are the normalized distances from the current solution to the Pareto front in the three objective dimensions; Step S55: Convert the optimal decision variable solution set obtained from the solution into a coordinated control command and send it to the execution agency.

7. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 6, characterized in that, Step S6 specifically includes: Step S61: Calculate the heat source efficiency ,in, This refers to the actual heat output of the heat source; To input heat into the fuel; Calculate pipeline transportation efficiency ,in, Provide heat to the heat source; For heat loss in the pipeline network; Calculate the terminal thermal efficiency ,in, Provide users with effective heat utilization; To output heat to the secondary side of the heat exchange station; Step S62: Calculate the overall energy efficiency index of the system. ,in, , , The adaptive weighting coefficients are dynamically adjusted according to the following formula: ； ； ； in, This represents the historical best value for heat source efficiency. This represents the historical best value for pipeline transportation efficiency. This represents the historical best value for end-use thermal efficiency; Step S63: Set the comprehensive energy efficiency index threshold. ,when or When the control decreases for three consecutive control cycles, the recalculation of steps S4 and S5 is triggered.

8. The method for optimizing and energy-saving control of operating parameters of a heating system according to claim 3, characterized in that, The confidence-based dynamic weight allocation mechanism in step S24 specifically includes: Step S241: Obtain the past prediction results of the heat source-level prediction layer, the pipeline-level prediction layer, and the user-level prediction layer. The average prediction error for each control cycle is denoted as . , , ; Step S242: Dynamically adjust the weight coefficients of the prediction results for each layer according to the following formula: ； ； ； in, The weighting coefficients for the heat source-level prediction results; The weighting coefficients for the pipeline-level prediction results; These are the weighting coefficients for user-level prediction results.

9. A heating system operating parameter optimization and energy-saving control system, characterized in that, A method for optimizing and energy-saving control of operating parameters of a heating system as described in any one of claims 1-8, comprising: The digital twin model building module is used to build a digital twin model of the heating system. It adopts a combination of mechanism modeling and data-driven approach to build a digital twin model that includes a heat source module, a pipe network module, a heat exchange station module, a heat user module, and a hydraulic calculation module. It also performs zone identification and dynamic optimization of pipe network roughness. The multi-level heat load prediction module is used to construct a multi-level heat load prediction model. It generates a comprehensive heat load prediction curve based on the collected operational and environmental data, including a heat source-level prediction layer, a pipeline-level prediction layer, and a user-level prediction layer. It also integrates the prediction results of each layer through a dynamic weight allocation mechanism based on confidence level. The thermal inertia dynamic evaluation module is used to perform thermal inertia dynamic evaluation on the heating network, calculate the thermal inertia time constant of each pipe section, and construct a network thermal inertia spatiotemporal distribution matrix covering the entire network. The multi-objective collaborative optimization module is used to construct a multi-objective collaborative operation parameter optimization model based on the digital twin model, the comprehensive heat load prediction curve and the spatiotemporal distribution matrix of the pipeline thermal inertia. The optimization objectives are the total system operating energy consumption, heating cost and user thermal comfort satisfaction. The decision variables are the heat source supply water temperature, heat source circulation flow rate, heat exchange station valve opening, heat exchange station circulation pump frequency and user-end valve opening. The module also sets constraints including heat supply and demand balance constraints, equipment operation safety constraints, hydraulic condition stability constraints and grid time-of-use electricity price constraints. The optimization algorithm solution module is used to solve the multi-objective collaborative operation parameter optimization model using an improved multi-objective gray wolf optimization algorithm. It dynamically adjusts the population size and search step size according to the spatiotemporal distribution matrix of pipeline thermal inertia, introduces the elite retention mechanism and adaptive weight adjustment mechanism of Pareto optimal frontier, and transforms the solution set of optimal decision variables obtained by the solution into collaborative control instructions and sends them to the execution mechanism. The multi-dimensional energy efficiency dynamic tracking module is used to build a multi-dimensional energy efficiency dynamic tracking mechanism that includes heat source efficiency, pipeline transmission efficiency and end-point heat consumption efficiency. It calculates the comprehensive energy efficiency index and triggers the recalculation of the multi-objective collaborative optimization module and the optimization algorithm solution module when the comprehensive energy efficiency index is lower than the preset threshold or continuously decreases.

Images