A method and device for regulating the hydraulic balance of a heating network

Abstract

The application provides a kind of adjustment method and equipment for heat supply pipe network hydraulic balance, applied to data processing technical field.The application is for heat supply pipe network hydraulic balance adjustment, first collects heat load, heat dissipation / throttling loss, medium physical property and other full-scene operation data, generates the adjustment basic data of multi-source heterogeneous fusion;Relying on thermal-hydraulic coupling model and feedforward prediction and other control modes, calculate the hydraulic parameters and plan to generate the initial scheme of adjustment. Through comparison with historical operating condition data, the scheme is converted into valve pump operating parameters, combined with pipe network operation deviation and user room temperature multi-dimensional feedback to calibrate and optimize the scheme. Introduce the self-learning prediction model of thermal insulation performance attenuation, combined with the dynamic optimization of medium multi-parameter coupling change adjustment strategy, finally through automatic closed-loop adjustment system real-time adjustment control instruction, realize the intelligent and accurate adjustment of pipe network hydraulic balance.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and equipment for adjusting the hydraulic balance of heating pipe networks. Background Technology

[0002] Currently, urban heating pipe networks generally suffer from technical problems such as hydraulic imbalance, uneven flow distribution, and insufficient control precision. Traditional regulation methods have obvious shortcomings, as detailed below: Traditional methods only collect basic hydraulic parameters such as flow rate and pressure, failing to cover comprehensive operational data across all scenarios, including branch heat load, network heat loss, valve throttling losses, circulating water medium properties, pipeline insulation performance, terminal equipment power, and user heating patterns. Furthermore, the heterogeneous communication protocols and data formats of different devices, lacking a unified fusion processing mechanism, cannot provide complete and accurate data support for hydraulic balance regulation. Relying on a single hydraulic calculation model fails to achieve coupled modeling of thermal and hydraulic parameters; passive feedback regulation is used without considering user heating periods and heat load fluctuations for feedforward planning, resulting in strong regulation lag, difficulty in matching dynamically changing heat demands, and inability to generate scientific initial regulation schemes in advance. The electric balancing valve and variable frequency circulating pump lack intelligent linkage, preventing the initial scheme from being accurately converted into executable equipment operating parameters; calibration relies solely on network operating parameters, failing to incorporate multi-dimensional feedback such as user room temperature, terminal heating capacity, and changes in medium properties, leading to low adaptability of the regulation scheme to actual operating conditions and a tendency for near-end overheating and far-end underheating.

[0003] The system lacks a predictive mechanism for the degradation of pipeline insulation performance, leading to a continuous increase in heat loss over long-term operation, and the adjustment strategy cannot be dynamically updated. It also fails to consider the combined effects of changes in circulating water temperature, density, and viscosity on hydraulic parameters, resulting in a fixed flow distribution ratio that is difficult to adapt to long-term aging of the pipeline network and fluctuations in media conditions. Furthermore, it only supports a single mode of global centralized control or end-point distributed control, unable to seamlessly switch according to operating conditions. The system lacks an automatic closed-loop system with real-time sensing, dynamic command generation, and continuous optimization, relying on manual experience for adjustment, resulting in low levels of intelligence and precision, and a difficulty in balancing energy consumption and comfort.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0006] According to one aspect of this application, a method for regulating the hydraulic balance of a heating network is provided, comprising: collecting full-scenario operation data of the heating network, real-time monitoring of branch heat load demand, network friction loss, valve throttling loss, circulating water density, pipeline insulation performance, power of terminal heat dissipation equipment, and user heating time patterns, and generating multi-source heterogeneous fusion hydraulic balance regulation basic data; based on a thermal-hydraulic coupling model, combined with feedforward prediction, feedback calibration, and dynamic adaptation control modes, calculating the optimized branch pressure difference value and flow distribution coefficient by inputting the regulation basic data, and generating an initial regulation scheme by performing feedforward planning based on user heat load demand and heating time patterns; comparing and analyzing the regulation basic data with historical balance operating condition data, and transforming the initial regulation scheme into valve opening parameters and pump parameters through intelligent linkage electric balancing valves and variable frequency circulating pumps. The system operates at high frequency, and then uses feedback parameters based on the deviation of heat loss along the pipeline, the actual heat supply of terminal equipment, and changes in circulating water density, combined with multi-dimensional data such as real-time room temperature feedback from the user side, to perform feedback calibration and optimize the adjustment scheme. It continuously records the thermal parameters of the pipeline and equipment operating data, and optimizes the adjustment strategy through dynamic adaptive control based on user heat load demand and pipeline heat loss, addressing heat load fluctuations and insulation performance degradation. It introduces a self-learning prediction model for pipeline insulation performance degradation, and adjusts the flow distribution ratio by combining circulating water density data with coupled changes in medium viscosity and temperature. Based on the adjustment foundation data, thermal monitoring data, and equipment operating parameters, the system uses an automatic closed-loop adjustment system to dynamically adjust control commands based on real-time heat load and heat loss parameters, outputting precise balance adjustment results to complete the intelligent hydraulic balance adjustment of the heating network.

[0007] Another aspect of this application is a regulating device for hydraulic balance of a heating network, used to execute executable instructions to perform the above-described regulating method for hydraulic balance of a heating network.

[0008] According to another aspect of this application, an electronic device includes: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described method for regulating the hydraulic balance of a heating network by executing the executable instructions.

[0009] According to another aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a second processor, implements the above-described method for adjusting the hydraulic balance of a heating network.

[0010] The beneficial technical effects of this application are as follows: First, data on branch heat load, heat dissipation loss, valve throttling loss, circulating water medium properties, pipeline insulation performance, terminal equipment power, and user heat consumption patterns are collected across the entire scenario. This data is then fused from multiple sources to form basic data for regulation. Next, based on a thermo-hydraulic coupling model, an initial regulation scheme is generated by combining feedforward prediction, feedback calibration, and dynamic adaptive control. By comparing with historical operating conditions, the scheme is transformed into valve and pump operating parameters, and multi-dimensional calibration is completed by combining pipeline network operating deviations and user room temperature. A self-learning prediction model for insulation performance degradation is introduced to dynamically optimize flow distribution by coupling changes in medium temperature, density, and viscosity. Finally, an automatic closed-loop system is used to achieve seamless switching between global centralized and terminal distributed control, real-time adjustment of control commands, and intelligent and precise regulation of hydraulic balance.

[0011] This application achieves comprehensive data acquisition and fusion, solving the problems of traditional single and heterogeneous data that are difficult to use. It provides complete data support for precise regulation, and adopts thermal-hydraulic coupled modeling and feedforward prediction to eliminate regulation lag and achieve on-demand heating and scientific flow allocation. Intelligent valve and pump linkage and multi-dimensional feedback calibration significantly improve regulation accuracy and alleviate the problems of near-end overheating and far-end underheating. It introduces self-learning prediction of insulation decay and media coupling correction to adapt to pipeline aging and operating condition fluctuations, extend the stable operation cycle of the system, support seamless switching between global and terminal modes and automatic closed-loop control, reduce manual dependence, and balance heating quality and operating energy consumption.

[0012] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0013] Figure 1 A flowchart of a method for adjusting the hydraulic balance of a heating network according to an embodiment of this application is shown; Figure 2 A schematic diagram of a hydraulic balance adjustment device for a heating network provided in an embodiment of this application is shown. Detailed Implementation

[0014] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0015] In one implementation, Figure 1 A schematic flowchart of a method for adjusting the hydraulic balance of a heating network according to an embodiment of this application is shown.

[0016] S101 collects full-scenario operation data of the heating network, monitors in real time branch heat load demand, heat loss along the pipeline, valve throttling loss, circulating water density, pipeline insulation performance, power of terminal heat dissipation equipment, and user heating time patterns, and generates multi-source heterogeneous fusion hydraulic balance regulation basic data.

[0017] In one implementation, seven core data collection dimensions are defined based on the hydraulic and thermal characteristics of the heating network and users' heating needs. The specific monitoring objects corresponding to each dimension are clearly defined, achieving full-scenario data coverage of the network from source to end-user heating, from medium properties to equipment operation, and from network losses to user patterns. Examples of each dimension and monitoring object are as follows: For the branch heat load demand dimension, the monitoring object is the real-time heat load value of each water supply branch and return branch in the network, including different types of branches such as residential heating branches and commercial building heating branches. For example, the real-time heat load of water supply branch No. 1 in the residential area is monitored as 800kW, and the real-time heat load of water supply branch No. 2 in the commercial street is monitored as 1200kW.

[0018] Regarding the dimension of heat loss along the pipeline network, the monitoring targets are the heat loss along the pipeline network of different diameters and laying methods, including underground direct-buried pipelines, overhead pipelines, and pipelines laid in trenches. For example, the heat loss along the pipeline network of a DN200 underground direct-buried water supply pipeline is 15kW per kilometer, and the heat loss along the pipeline network of a DN150 overhead return water pipeline is 20kW per kilometer. Regarding the dimension of valve throttling loss, the monitoring targets are the throttling loss values ​​of various regulating valves, balancing valves, and check valves in the pipeline network, including electric balancing valves in each branch, butterfly valves in the main pipeline, and regulating valves at the building entrance. For example, the throttling loss of the electric balancing valve at the building entrance is 0.05MPa, and the throttling loss of the butterfly valve in the main pipeline is 0.02MPa.

[0019] Regarding the circulating water density dimension, the monitoring targets are the real-time density of the circulating water in the pipe network under different temperatures and pressures. Monitoring points are set at the circulating water pump outlet, key nodes of the main pipe, and the manifolds of each branch. For example, when the circulating water temperature is 60℃ and the pressure is 0.3MPa, the real-time density is 983.2kg / m³; when the temperature is 45℃ and the pressure is 0.25MPa, the real-time density is 990.1kg / m³. Regarding the pipe insulation performance dimension, the monitoring targets are the thermal conductivity, insulation layer thickness, and insulation layer damage rate of the pipe network insulation layer, covering pipes with different service lives. For example, the thermal conductivity of a polyurethane insulation layer after 3 years of use is 0.028W / (m³). The rock wool insulation layer has a thickness of 50mm and a breakage rate of 1%; after 8 years of use, its thermal conductivity is 0.04W / (m²). K), thickness 60mm, breakage rate 8%.

[0020] Regarding the power dimension of terminal heat dissipation equipment, the monitoring targets are the rated power and actual operating power of heat dissipation equipment such as radiators, underfloor heating coils, and fan coil units at the user end. For example, the rated power of underfloor heating coils in residential homes is 10kW, and the actual operating power is 7.5kW; the rated power of fan coil units in commercial buildings is 15kW, and the actual operating power is 12kW. Regarding the user's heating time pattern dimension, the monitoring targets are the distribution of heating time periods for different types of users, as well as the peak and trough values ​​of heat load within each period, including residential users, commercial users, and office users. For example, the peak heating time for residential users is 7:00-9:00 AM and 6:00-10:00 PM, with a peak heat load of 900kW, and the trough is 0:00-5:00 AM, with a trough value of 300kW; the peak heating time for office users is 9:00-6:00 PM, with a peak value of 1000kW, and the heat load is 0 during non-working hours.

[0021] For the seven dimensions of monitoring targets mentioned above, matching IoT monitoring and sensing devices are deployed throughout the entire pipeline network. A unified data acquisition frequency (which can be set to 5-15 minutes / time depending on the scale of the pipeline network, and 1-5 minutes / time for core nodes) and data acquisition accuracy (pressure accuracy 0.01MPa, temperature accuracy 0.1℃, heat load accuracy 1kW) are set to achieve real-time and accurate acquisition of data in each dimension. An example of equipment deployment and data acquisition is shown below.

[0022] Heat meters are installed on each branch line to collect flow rate and supply / return water temperature difference data, automatically calculating real-time heat load. Data is uploaded to a data acquisition terminal via the RS-485 protocol. Temperature sensors and ambient temperature sensors are installed on the outer wall of the pipeline to collect data on the pipeline medium temperature, pipeline outer wall temperature, and ambient temperature. Combined with pipeline insulation performance parameters, heat loss along the pipeline is calculated. Data is uploaded via LoRa wireless communication. Pressure transmitters are installed before and after valves to collect real-time pressure values. Throttling losses are calculated based on the pressure difference. The pressure transmitters are linked in real-time with the pipeline network monitoring system.

[0023] Density meters are installed at key monitoring points, in conjunction with temperature and pressure sensors, to achieve real-time acquisition of density changes with temperature and pressure. Density meter data is directly connected to a data aggregation platform. Handheld testing devices are used to periodically measure the thermal conductivity and breakage rate of the insulation layer. Combined with the insulation layer thickness in the pipeline design documents, the data is entered into the system and updated in real time. Simultaneously, temperature sensor monitoring results are used to infer the insulation performance degradation. Power meters are installed at user-end heat dissipation equipment to collect actual operating power. Combined with the rated power recorded on the equipment nameplate, a power dataset for end-point equipment is formed and uploaded through the community's smart heating terminal. Based on historical data collected from branch heat meters and user-end power meters, combined with user heating survey information, data statistics are used to generate data on heating period patterns for different users, and the data on heat load changes within each period is updated in real time.

[0024] Due to the heterogeneity of communication protocols, data formats, and data types among various monitoring devices (such as 485 protocol, LoRa protocol, Modbus protocol; numerical data, text data, time-series data; real-time dynamic data, periodically entered static data), a unified multi-source heterogeneous data aggregation channel is established to achieve unified reception, transmission, and aggregation of data with different protocols, formats, and types. A specific implementation example is as follows.

[0025] Protocol conversion gateways are set up in each pipeline monitoring area to convert non-standard protocols such as LoRa and RS-485 into standard TCP / IP protocols, achieving communication compatibility between different devices. For example, LoRa wireless data from underground pipeline temperature sensors is converted to TCP / IP protocol before being transmitted to the city's heating pipeline data center. Preprocessing terminals are set up before data aggregation to standardize the format of the collected raw data, converting all data into standardized numerical time-series data. For example, text-based test reports on pipeline insulation performance are converted into numerical data on thermal conductivity and breakage rate, and static data on rated power of equipment is integrated with dynamic data on actual operating power into a time-series dataset. A full-scenario operation data aggregation center for the heating pipeline network is built based on a cloud computing platform. Monitoring data from each region and node is uniformly aggregated to the cloud via a dedicated network, achieving centralized storage and unified management of the data. For example, pipeline data from the four heating areas in the east, west, south, and north of the city are aggregated to the city's heating cloud data platform via a municipal dedicated communication network, forming a comprehensive database of the entire pipeline network's operation data.

[0026] The raw operational data from all scenarios gathered in the cloud undergoes multi-dimensional fusion processing, including data cleaning, data association, data standardization, and data fusion verification. This process eliminates data redundancy, errors, and missing data, establishes relationships between data from various dimensions, and ultimately generates multi-source heterogeneous fusion basic data for hydraulic balance regulation. This data is a time-seriesd, standardized, and correlated dataset that can be directly used as input data for subsequent thermal-hydraulic coupling models. Examples of each processing step are as follows.

[0027] The raw data is processed through outlier removal, missing value completion, and duplicate value deletion. For example, outlier data with a pressure value of 0 due to sensor malfunction is removed; linear interpolation is used to complete the missing 5-minute circulating water density data due to network interruption; and duplicate heat load data of the same branch collected by different devices is deleted. Based on the pipeline network topology, a one-to-one correspondence is established between the data of each dimension and the pipeline network nodes, branches, buildings, and users, realizing the association between data and the physical location of the pipeline network. At the same time, logical associations are established between the data of each dimension. For example, the heat load demand, valve throttling loss, and circulating water density data of the No. 1 water supply branch of the community are associated with the physical nodes of that branch, and the power data of the terminal heat dissipation equipment is associated with the heat consumption period data of the corresponding residential users and commercial users.

[0028] Standardize the units, ranges, and formats for each dimension of data to achieve data normalization. For example, pressure data is standardized to MPa, temperature to °C, heat load to kW, density to kg / m³, and thermal conductivity to W / (m³). K); The heat load values ​​of each branch are normalized according to the maximum heat load of the pipeline network, and the value range is mapped to between 0 and 1 to facilitate subsequent model calculations. The fused data is verified by historical operating data of the pipeline network and hydraulic-thermal simulation data to check the rationality and accuracy of the data. For example, the heat loss data along the DN200 pipeline is compared with the theoretical data calculated by the pipeline network simulation software. If the deviation is controlled within 5%, the verification is considered successful. If the deviation exceeds the range, the data acquisition equipment and data processing process are re-checked.

[0029] The cleaned, correlated, standardized, and verified multi-source heterogeneous data are integrated into a unified hydraulic balance regulation basic dataset, which is classified and stored according to the pipeline branch level and heat load type. This provides standardized input data for the generation of subsequent regulation initial schemes. For example, it generates urban heating pipeline hydraulic balance regulation basic data divided into "main pipe-secondary pipe-branch-building" levels, containing seven core dimensions and time sequence, and stores it in a dedicated dataset module on the cloud data platform, supporting real-time access to the thermal-hydraulic coupling model.

[0030] S102, based on the thermal-hydraulic coupling model, combines feedforward prediction, feedback calibration, and dynamic adaptation control modes. It inputs basic regulation data to calculate the optimized branch pressure difference and flow distribution coefficient, and generates the initial regulation scheme by relying on user heat load demand and heat use period patterns for feedforward planning.

[0031] In one implementation, the basic data for hydraulic balance regulation is processed and layered according to the pipeline branch level and heat load type. Each layer of data includes node parameters, branch parameters, and user-side parameters. Dynamic features such as heat load fluctuations, changes in medium properties, and changes in heat consumption patterns are aggregated through a dynamic monitoring window. Data on pipeline resistance, medium transport, and heat demand supply at each level are extracted to generate corresponding subsets. The basic data is double-layered according to the pipeline branch level and heat load type. The pipeline branch level is divided into main pipeline level, secondary pipeline level, community branch level, and building entrance level according to the actual topology of the heating pipeline network. The heat load type is divided into residential heating load, commercial building heating load, and office building heating load according to user attributes. Each layer of data includes three core data categories: node parameters, branch parameters, and user-side parameters. The main pipeline level residential heating load data layer includes node parameters such as main pipeline supply / return node pressure, temperature, and circulating water density; branch parameters include main pipeline diameter, friction loss, and throttling loss; and user-side parameters include the total heat load demand and heating period patterns of residents in the area covered by the main pipeline. The community branch level commercial building heating load data layer includes node parameters of community branch manifolds / collectors; branch parameters include the pipe insulation performance and heat loss of the community commercial branch; and user-side parameters include the real-time heat load and terminal heat dissipation equipment power of the community commercial building.

[0032] Dynamic monitoring windows are configured for each layer of data. The window duration is set according to the load fluctuation characteristics of the pipeline network level. The window duration for the main / secondary pipeline level is set to 30 minutes, and the window duration for the community branch / building entrance level is set to 15 minutes. Through the windows, three types of dynamic characteristics within the layered data are aggregated in real time: heat load fluctuation, medium property changes, and heat consumption pattern changes, capturing the temporal change patterns of the data. A 15-minute dynamic monitoring window is opened for the residential heating load data layer at the community branch level, aggregating the fluctuation characteristics of the residential heat load from 500kW to 800kW, the density change characteristics caused by the circulating water temperature dropping from 55℃ to 52℃, and the heat load increase pattern during the morning peak heating period. A 30-minute dynamic monitoring window is opened for the office building load data layer at the main pipeline level, aggregating the pattern of stable heat load during office hours and sudden drop in heat load during non-office hours, and the physical property characteristics of circulating water medium viscosity changing with temperature and pressure.

[0033] From the hierarchical data after aggregating dynamic features, core control data for each level of the pipeline network operation are extracted, namely, data related to pipeline resistance, medium transport, and heat demand supply. After removing redundant data, specific data subsets corresponding to each level and load type are generated. These data subsets are standardized time-series data that can directly support subsequent model calculations. From the secondary trunk line level commercial building load data layer, pipeline resistance data (friction coefficient of secondary trunk line, valve throttling resistance), medium transport data (circulating water flow, transport pressure, medium density / viscosity), and heat demand supply data (total heat load of commercial area, supply-demand difference) are extracted to generate a secondary trunk line commercial load data subset. From the building entrance level residential load data layer, local resistance of building entrance pipes, entrance water flow, and heat load demand data of individual buildings are extracted to generate a building entrance residential load data subset.

[0034] Using pipeline branches as core nodes, hydraulic parameters as control nodes, and heat load demand as target nodes, node attribute information is generated by combining pipeline topology mapping and heat load type identification. Edge association information is generated according to parameter adaptability and heat demand response time sequence. The system clearly defines pipeline branches as core nodes, hydraulic parameters as control nodes, and heat load demand as target nodes. Core nodes are the physical carriers of the pipeline network, control nodes are the adjustable parameter carriers for achieving hydraulic balance, and target nodes are the demand carriers for pipeline heating services. These three types of nodes cover the entire link of the heating pipeline network from physical topology to control execution and demand targets. The DN300 secondary water supply trunk line in the eastern part of the city is designated as the core node; hydraulic parameters such as differential pressure, flow rate, valve opening, and pump frequency of this branch are designated as control nodes; and the heat load demand of residential communities and commercial complexes in the eastern part of the city covered by this branch is designated as the target nodes.

[0035] By combining pipeline topology mapping and heat load type identification, unique attribute information is generated for three types of nodes. Pipeline topology mapping realizes a one-to-one correspondence between nodes and the actual physical location of the pipeline network. Heat load type identification clarifies the load attributes of the target nodes. The node attribute information includes core contents such as physical location, parameter type, load attributes, data value range, and real-time monitoring values. The attribute information of the core node (DN300 secondary water supply pipe in the eastern area) is "physical location: XX Road in the eastern area, pipe type: secondary water supply pipe, pipe diameter: DN300, real-time monitoring pressure: 0.35MPa"; the attribute information of the control node (pressure difference of this branch) is "parameter type: hydraulic control parameter, value range: 0.2-0.4MPa, real-time monitoring value: 0.3MPa, control method: electric balancing valve adjustment"; the attribute information of the target node (heat load of residential area in the eastern area) is "load type: residential heating load, number of households covered: 500 households, real-time heat load: 1200kW, peak heating period: 18:00-22:00".

[0036] Based on parameter fit and heat demand response time sequence, three types of edge association information between nodes are constructed. Parameter fit characterizes the degree of adaptation of the hydraulic parameters of the control node to the operation of the core node pipeline network. Heat demand response time sequence characterizes the response time of the core node's medium delivery to the heat load demand of the target node. Edge association information includes associated node pairs, fit values, response time sequence, and association weights. The higher the fit and response time sequence, the greater the edge association weight. Edge association information is constructed between the core node (DN300 secondary water supply pipe in the eastern area) and the control node (pressure difference of this branch), with a fit value of 0.95 (high fit between pressure difference and secondary pipe operation) and an association weight of 0.8. Edge association information is also constructed between the core node (DN300 secondary water supply pipe in the eastern area) and the target node (heat load of the residential area in the eastern area), with a response time sequence of 5 minutes (timeliness of medium delivery to the community) and an association weight of 0.9. The edge association information between each node is labeled with associated node pairs, core feature values, and association weights, forming a complete node association topology.

[0037] A directed graph is constructed, encompassing all branch nodes of the heating network, various types of hydraulic control nodes, and nodes with differentiated heat load demands. Divided into three views based on hydraulic constraints, thermal constraints, and heat consumption constraints, the cross-view edge correlation strength is optimized through a constraint-based collaborative attention mechanism. Based on a thermal-hydraulic coupling model combined with feedforward prediction, feedback calibration, and dynamic adaptive control modes, the optimized branch pressure difference and flow distribution coefficient are calculated using corresponding subset data. Feedforward planning is performed based on user heat load demand and heat consumption patterns to generate an initial hydraulic balance regulation scheme for the heating network. All core nodes of the heating network branches, various types of hydraulic control nodes, and target nodes for differentiated heat load demands are included in the modeling scope. Based on the generated edge correlation information, a directed graph model of the entire network is constructed. The edge directions of the directed graph are set according to the regulation logic of "control node - core node - target node," intuitively reflecting the supply and regulation relationship of hydraulic control parameters to heat load demand through the network branches. Ten main trunk lines, 25 secondary trunk lines, and 80 community branch roads of the urban heating network are designated as core nodes. Twelve types of hydraulic parameters, such as differential pressure, flow rate, valve opening, and pump frequency, are designated as control nodes. Sixty heat load demand areas in three categories, namely residential, commercial, and office, are designated as target nodes. Based on edge association information, a directed graph of the urban heating network with 175 nodes and 320 edges is constructed. The edge direction is from the hydraulic control node to the core node of the network, and then from the core node of the network to the heat load target node.

[0038] Based on the core constraint types of the heating network operation, the directed graph is divided into three independent views: hydraulic constraint view, thermal constraint view, and heat consumption constraint view. Each view focuses on a single constraint type, extracting the corresponding node and edge association information to ensure the targeted modeling of each constraint. The hydraulic constraint view only includes the core nodes and control nodes and the edge association information between them, focusing on hydraulic parameter constraints such as network pressure difference, flow rate, and resistance. The thermal constraint view includes core nodes, control nodes, and target nodes, focusing on thermal parameter constraints such as network heat loss, heat medium transport efficiency, and heat load supply and demand balance. The heat consumption constraint view only includes the core nodes and target nodes and the edge association information between them, focusing on heat demand constraints such as user heating periods, peak / valley heat load, and room temperature comfort.

[0039] A constraint-based collaborative attention mechanism is introduced to optimize the edge association strength between the three views. This mechanism calculates the influence weight of each constraint view on the hydraulic balance of the pipeline network and adjusts the edge association strength across views accordingly. The basic weights for hydraulic constraints, thermal constraints, and heat consumption constraints are set to 0.4, 0.35, and 0.25, respectively, and then dynamically adjusted according to the real-time operating status of the pipeline network. This ensures that the edge association strength of the directed graph accurately matches the actual constraint requirements of the pipeline network, avoiding control deviations caused by single-constraint modeling. When the pipeline network is in peak residential heating season, the influence weight of the heat consumption constraint dynamically increases from 0.25 to 0.35, the weight of the hydraulic constraint is adjusted to 0.35, and the weight of the thermal constraint remains unchanged at 0.3. Through this weight change, the mechanism strengthens the edge association strength between the heat consumption constraint view and the other two views, making the directed graph more closely reflect the pipeline network constraint characteristics during peak heating season. When the pipeline network experiences abnormal hydraulic resistance, the weight of the hydraulic constraint increases to 0.5, and the weights of the thermal and heat consumption constraints are adjusted to 0.3 and 0.2, respectively, strengthening the cross-view association strength of the hydraulic constraint view.

[0040] The core calculation model is a thermal-hydraulic coupling model that integrates the thermal characteristics (heat dissipation, heat load, and heat medium temperature) and hydraulic characteristics (pressure difference, flow rate, and resistance) of the heating network. It can realize the coupled calculation of thermal-hydraulic parameters and embed a three-mode collaborative control logic of feedforward prediction, feedback calibration, and dynamic adaptation. Feedforward prediction is the core, which plans the control strategy in advance based on real-time data. Feedback calibration and dynamic adaptation are auxiliary, reserving adjustment space for subsequent scheme optimization.

[0041] Data subsets from each level and load type are input into the thermal-hydraulic coupling model. Based on the network topology, constraints, and dynamic characteristics, the model uses a fluid mechanics and thermal engineering coupling algorithm to calculate the optimized differential pressure and flow distribution coefficient for each network branch. The calculation results are then validated against actual network operating conditions; if the deviation exceeds 5%, data is re-entered for iterative calculation. For example, a subset of residential load data from a community branch is input into the model. The model, considering the branch's network resistance, residential heat load demand, and circulating water properties, calculates an optimized differential pressure of 0.25 MPa and a flow distribution coefficient of 0.8 (i.e., 80% of the total flow is distributed to this branch). Similarly, a subset of commercial load data from a secondary trunk line is input into the model, resulting in an optimized differential pressure of 0.32 MPa and a flow distribution coefficient of 0.9.

[0042] Based on the optimized differential pressure value and flow distribution coefficient calculated by the model, feedforward planning is carried out according to the user's heat load demand and heating period patterns. The hydraulic control parameters of each branch are pre-allocated and set according to the load size, load type, and peak / valley periods of different users, ensuring that the control strategy matches the actual heating demand of users and achieving "on-demand heating" feedforward control. For residential users' peak heating periods of 7:00-9:00 AM and 6:00-10:00 PM, the feedforward planning increases the optimized differential pressure value of residential heating branches by 10% and the flow distribution coefficient by 15% 30 minutes before the peak. For office users' peak heating period of 9:00-6:00 PM, the feedforward planning adjusts the flow distribution coefficient of office heating branches to 0.95 15 minutes before 9:00 AM, and reduces the flow distribution coefficient to 0.4 during non-working hours. For commercial users' continuous heating throughout the day, the feedforward planning keeps the differential pressure and flow distribution coefficient of commercial heating branches stable, making only minor adjustments based on real-time heat load.

[0043] By integrating the optimized differential pressure values, flow distribution coefficients, and feedforward planning control strategies of each branch, an initial hydraulic balance control scheme for the heating network is generated. The scheme clearly defines the hydraulic control target parameters, control periods, and control methods for each network level and load type branch, providing a core basis for subsequent conversion of the scheme into actual valve and pump operating parameters and scheme optimization and calibration. The generated initial hydraulic balance control scheme for the urban heating network specifies an optimized differential pressure value of 0.38 MPa and a flow distribution coefficient of 0.9 for the main pipeline residential heating branches, with the control period covering the residential double peak hours and the control method being frequency adjustment by a variable frequency circulating pump; and an optimized differential pressure value of 0.28 MPa and a flow distribution coefficient of 0.85 for the community commercial heating branches, with the control period being all day and the control method being adjustment of the opening by an electric balancing valve. The control parameters, periods, and methods for all branches are clearly marked in the scheme, forming a standardized and executable initial control scheme.

[0044] S103 compares and analyzes the basic adjustment data with historical balance operating data. Through intelligent linkage electric balance valve and variable frequency circulating pump, the initial adjustment scheme is transformed into valve opening parameters and pump operating frequency. Then, based on the feedback parameters of heat loss deviation along the pipeline, actual heat supply of terminal equipment, and circulating water density change, combined with multi-dimensional data of real-time room temperature feedback from the user side, feedback calibration is performed to optimize the adjustment scheme.

[0045] In one implementation, integrated data comparison technology is used to process multi-source heterogeneous fusion of regulation base data and historical balance condition data to generate data deviation analysis results, quantified data of operating condition matching, a set of pipeline network operation characteristics, and prediction results of hydraulic parameter adaptation trends. First, the regulation base data (real-time time-series data) and historical balance condition data (historical optimal operating condition dataset) are standardized and aligned, unifying data units, time dimensions, and pipeline network hierarchical classification standards. Historical data is matched one-to-one with real-time base data according to pipeline branches and heat load types to ensure the accuracy of the comparison. The real-time collected residential heating load base data (units: kW, MPa, ℃) of the community branch is unified with the historical balance condition data of the branch's winter heating over the past three years, aligned according to the "24-hour daily" time dimension, and matched according to the "building-branch-secondary trunk" hierarchy to eliminate data format and dimensional differences.

[0046] Based on the aligned data, an integrated comparison and calculation is performed from three dimensions: pipeline hydraulic parameters, thermal parameters, and load matching degree. The data deviation is calculated by the difference method, the operating condition matching degree is calculated by the similarity algorithm, the pipeline operation characteristics are sorted out by the feature extraction algorithm, and the hydraulic parameter adaptation trend is predicted by the trend fitting algorithm. The calculations show that the deviation between the real-time water supply pressure of the community branch (0.28 MPa) and the historical equilibrium pressure (0.30 MPa) is -0.02 MPa, and the deviation between the real-time heat load (850 kW) and the historical equilibrium heat load (800 kW) is +50 kW, generating data deviation analysis results. The similarity algorithm calculates that the matching degree between the current operating condition and the historical equilibrium condition of this branch is 89%, and the matching degree between the current operating condition and the historical equilibrium condition of the commercial branch is 92%, generating quantitative data on operating condition matching. Key characteristics of pipeline operation, such as "low differential pressure, high heat load, and slightly increased heat loss," are extracted for each branch and integrated into a set of pipeline operation characteristics. Trend fitting predicts that "as the heat load continues to rise, the circulating water density decreases, and the branch flow adaptation coefficient needs to be increased by 5%," generating a hydraulic parameter adaptation trend prediction result.

[0047] The four types of analysis results are integrated and stored according to the hierarchical classification of main pipelines, secondary pipelines, and community branch roads. The rationality is verified in combination with the actual operating status of the pipeline network. If the deviation data exceeds the normal fluctuation range (such as the absolute value of pressure deviation > 0.05MPa), the data comparison and calculation are re-performed to ensure the accuracy of the analysis results.

[0048] This process integrates and processes the feature set with hydraulic regulation association standards, parameter correction ranking rules, and operating condition adaptation threshold information. It incorporates user-side room temperature comfort threshold requirements to establish a precise matching relationship between features and regulation parameter correction needs, generating a standardized set of regulation parameters with multi-dimensional constraints. The process collects hydraulic regulation association standards within the heating network industry (such as the "Design Code for Urban Heating Networks"), parameter correction ranking rules established for this network (such as "priority of differential pressure correction > priority of flow rate correction > priority of heat loss correction"), and operating condition adaptation threshold information (such as heat load matching degree ≥85%, pressure deviation ≤±0.05MPa). It also incorporates user-side room temperature comfort threshold requirements (such as residential room temperature 20±2℃, commercial room temperature 22±2℃) to form a multi-dimensional constraint information system. The integrated constraint information includes "allowable deviation of branch differential pressure ±0.03MPa, heat load matching degree ≥90%, and residential room temperature not lower than 18℃," while clarifying the ranking rule that "room temperature deviation correction priority is higher than network hydraulic parameter deviation correction priority."

[0049] Based on the integrated constraint information system, a one-to-one correspondence is established between pipeline network operation characteristics and adjustment parameter correction requirements. This means that according to different operation characteristics, corresponding adjustment parameter correction directions, ranges, and requirements are matched to ensure that correction requirements accurately correspond to actual pipeline network problems. For the operation characteristic of "branch pressure differential 0.02 MPa low and residential room temperature 17℃ (below the threshold)," the adjustment parameter correction requirement is matched to "increase branch water supply pressure differential and increase circulating water flow, with a pressure differential correction range of 0.02-0.03 MPa and a flow correction range of 5%-8%." ​​For the operation characteristic of "slightly increased heat loss and commercial room temperature 23℃ (above the threshold)," the correction requirement is matched to "slightly reduce the water supply temperature, with a correction range of 2-3℃, controlling heat loss while ensuring room temperature remains within the threshold."

[0050] The matched operational characteristics and correction requirements are standardized, unifying the expression, numerical range, and priority ranking of parameter corrections. Classified by pipeline branch level and heat load type, a standardized set of multi-dimensional constraints is generated. This parameter set includes core information such as correction parameter type, correction direction, correction range, constraint threshold, and priority ranking. All parameters are standardized values, directly supporting subsequent information extraction. The generated standardized adjustment parameter set for residential heating in the community branch includes: "Correction parameter: Water supply pressure difference, correction direction: increase, correction range: 0.28-0.31MPa, constraint threshold: Residential room temperature ≥18℃, priority: 1; Correction parameter: Circulating water flow rate, correction direction: increase, correction range: 80-86m³ / h, constraint threshold: Flow rate matching degree ≥90%, priority: 2."

[0051] Based on a standardized set of multi-dimensional constraints, core control information is extracted. The initial control scheme is used as the input dimension, hydraulic deviation characteristics as the core parameters, and parameter correction degree evaluation rules as the judgment criteria. Dynamic optimization judgment indicators based on reinforcement learning are added to achieve accurate reading of branch differential pressure target values, flow distribution coefficients, valve adjustment ranges, and pump frequency ranges. A four-tiered judgment system of "input dimension + core parameters + judgment criteria + dynamic optimization indicators" is constructed. The hydraulic control target in the initial control scheme is used as the basic input, the hydraulic deviation characteristics of the pipeline network (such as differential pressure deviation and flow deviation) are used as the core judgment parameters, and the parameter correction degree evaluation rules (such as "the larger the deviation, the higher the correction degree, and the greater the parameter reading weight") are used as the basic judgment criteria. Simultaneously, dynamic optimization judgment indicators based on reinforcement learning (such as "optimal energy consumption, optimal room temperature comfort, and optimal operating condition adaptability") are introduced to make parameter reading more aligned with multi-objective optimization needs. In the established judgment system, the input dimension is the initial plan for the heating regulation of residents in the community branch road (pressure difference target 0.25MPa, flow distribution coefficient 0.8), the core parameters are "pressure difference deviation -0.02MPa, room temperature deviation -1℃", the judgment basis is "the room temperature deviation correction degree is higher than the pressure difference deviation correction degree", and the dynamic optimization index is "the optimal room temperature comfort of residents + the optimal energy consumption of equipment operation".

[0052] Based on the established judgment system, information from the standardized adjustment parameter set is screened, calculated, and extracted to accurately read the branch pressure difference target value, flow distribution coefficient, valve adjustment range, and pump frequency range of each pipeline branch. The reading results must simultaneously meet the requirements of constraint thresholds, correction priorities, and dynamic optimization indicators. If the reading results conflict with any requirement, recalculation and adjustment are performed. For the residential heating conditions of the community branch, the judgment system accurately reads the branch pressure difference target value of 0.30MPa (meeting the constraint of room temperature ≥18℃), flow distribution coefficient of 0.85 (adapting to the demand for increased heat load), valve adjustment range of 20%-30% (electric balancing valve opening, corresponding to the demand for increased pressure difference), and pump frequency range of 35-40Hz (variable frequency circulating pump frequency, corresponding to the demand for increased flow). This reading result simultaneously meets the reinforcement learning dynamic optimization indicators of "optimal room temperature comfort" and "optimal energy consumption," and there is no conflict, so no adjustment is required. The core control information read is verified against the actual operating capacity of the pipeline equipment to confirm that the valve adjustment range and pump frequency range are within the rated operating range of the equipment (such as the rated opening of the electric balance valve 0-100% and the rated frequency of the variable frequency pump 20-50Hz). If they are outside the range, the reading results are corrected to ensure the feasibility of the parameters.

[0053] The core control parameters are quantified and corrected, with values ​​ranging from 0 to 1. Higher correction values ​​indicate a greater impact of the parameter on hydraulic balance. Simultaneously, considering control priority requirements (e.g., room temperature correction > differential pressure correction > flow rate correction) and the weighting of user-side room temperature feedback (e.g., resident room temperature feedback weight 0.6, network hydraulic parameter feedback weight 0.4), correlation weights are assigned to each parameter. The quantified correction values ​​and correlation weights serve as the basis for parameter ranking. Core control parameters for community branch lines are quantified: differential pressure target value correction degree 0.95, correlation weight 0.6 (room temperature feedback is dominant); flow rate distribution coefficient correction degree 0.90, correlation weight 0.5; valve adjustment range correction degree 0.85, correlation weight 0.4; pump frequency range correction degree 0.85, correlation weight 0.4. Energy consumption constraints are also set for equipment operation: variable frequency pump operating power ≤ 15kW, electric balancing valve adjustment energy consumption ≤ 0.5kW.

[0054] Based on the comprehensive score of the corrected quantified value and the correlation weight, the core control parameters are sorted in descending order to generate a list of control parameters arranged in order of correction degree and correlation. At the same time, the equipment adaptation attributes of each parameter are clearly specified in the list (such as which valve / pump the parameter corresponds to) and the adjustment range (such as the specific numerical range that the parameter needs to be increased / decreased) to ensure that the parameters and the executing equipment are accurately matched. The control parameter list of the community branch is generated by sorting by comprehensive score, and the order is: "Differential pressure target value 0.30MPa (correction degree 0.95, weight 0.6) -- Flow distribution coefficient 0.85 (correction degree 0.90, weight 0.5) -- Valve adjustment range 20%-30% (adapted to the No.1 electric balancing valve in the East Area, adjustment range +10%) -- Pump frequency range 35-40Hz (adapted to the No.1 variable frequency circulating pump in the East Area, adjustment range +5Hz)". The list also marks the equipment operating energy consumption constraint thresholds "Pump power ≤15kW, valve energy consumption ≤0.5kW".

[0055] Based on the list of control parameters, and combined with the equipment characteristics of the electric balancing valve and the variable frequency circulating pump (such as the correspondence between valve opening and differential pressure, and the correspondence between pump frequency and flow rate), the abstract control parameters are transformed into directly executable equipment operating parameters, namely, specific values ​​for valve opening and specific pump operating frequencies. The transformation results must meet the equipment's adjustment accuracy requirements (such as valve opening accuracy ±1%, pump frequency accuracy ±0.5Hz). Based on the characteristic of the No. 1 electric balancing valve that "for every 5% increase in opening, the branch differential pressure increases by 0.01MPa," the "valve adjustment range 20%-30%" is transformed into a valve opening parameter of 25%. Based on the characteristic of the No. 1 variable frequency circulating pump that "for every 5Hz increase in frequency, the circulating water flow rate increases by 8m³ / h," the "pump frequency range 35-40Hz" is transformed into a pump operating frequency of 38Hz, forming a detailed list of valve and pump operating parameters.

[0056] Based on IoT communication technology, valve and pump operating parameters (valve opening degree, pump frequency) are transmitted to each electric balancing valve and variable frequency circulating pump through the control system, realizing intelligent linkage and precise control of the equipment. During the control execution process, the actual operating parameters of the equipment are collected in real time to confirm that the equipment is working normally according to the instructions. If equipment failure occurs (such as valve jamming or pump frequency loss), an alarm is immediately issued and control is suspended. The parameters "1# electric balancing valve opening degree 25%, 1# variable frequency circulating pump frequency 38Hz" are transmitted to the pipeline equipment in the eastern area. The control system collects the equipment operating data in real time, confirming that the actual valve opening degree is 25% and the actual pump frequency is 38Hz, and that the equipment is executing the control instructions normally.

[0057] After the control measures were implemented, two types of feedback data were collected in real time: first, pipeline operation feedback parameters, including the deviation of heat loss along the pipeline (the difference between real-time heat loss and theoretical heat loss), the actual heat output of terminal equipment (the actual heating power of radiators / underfloor heating), and the change in circulating water density (the difference in circulating water density before and after the control measures); second, user-side room temperature feedback data, including real-time indoor temperature for residents / commercial / office users and room temperature satisfaction feedback (such as whether it is too cold / too hot), forming a multi-dimensional feedback data system. After the control measures were implemented, the deviation of heat loss along the branch pipeline in the community was +0.5kW (slightly increased), the actual heat output of the terminal underfloor heating was 9kW (an increase of 1.5kW), and the change in circulating water density was -0.5kg / m³ (a slight decrease); the user-side room temperature feedback was that the average room temperature for residents was 19℃ (an increase of 2℃), with no feedback indicating that it was too cold / too hot, and the average room temperature for commercial users was 22℃ (meeting the threshold).

[0058] Multi-dimensional feedback data is compared with constraint thresholds (e.g., room temperature ≥ 18℃, heat loss deviation ≤ ±1kW). If the feedback data meets all constraint thresholds, the current adjustment scheme is suitable for the pipeline network conditions and no major adjustments are needed. If the feedback data exceeds the constraint thresholds, the control parameters are slightly modified according to the magnitude of the deviation to complete the scheme calibration. Finally, the calibrated control parameters, valve and pump operating parameters, and feedback data are integrated to generate an optimized hydraulic balance control scheme for the heating pipeline network. The collected multi-dimensional feedback data all meet the constraint thresholds; with residential room temperature reaching 19℃ and heat loss deviation ≤ 0.5kW, only a slight modification to the pump frequency (from 38Hz to 37Hz) is needed to avoid excessive energy consumption. The calibrated parameters and feedback data are integrated to generate an optimized hydraulic balance control scheme for the eastern heating pipeline network. The scheme clearly defines the final operating parameters of each device, the pipeline network control target, and the feedback data verification results, providing a basis for subsequent dynamic adaptation and control.

[0059] S104 continuously records the thermal parameters of the pipeline network and equipment operation data. Based on user heat load demand and pipeline heat loss, it optimizes the adjustment strategy through dynamic adaptation control to address heat load fluctuations and insulation performance degradation. It introduces a self-learning prediction model for pipeline insulation performance degradation and adjusts the flow distribution ratio by combining circulating water density data and coupled change data of medium viscosity and temperature.

[0060] In one implementation, a multi-dimensional data storage algorithm is used to perform time-series archiving and categorized labeling of pipeline thermal parameters and equipment operation data, defining the boundaries between parameter monitoring units and data associations. Data is archived time-series based on a dual dimension of "timestamp + pipeline hierarchy," with timestamp precision set to 1 minute (core monitoring nodes) and 5 minutes (ordinary monitoring nodes). The pipeline hierarchy is divided into "main pipe - secondary pipe - branch - building," ensuring data can be traced chronologically and retrieved hierarchically. A distributed storage architecture is employed to guarantee data storage security and retrieval efficiency. All archived data is categorized and labeled according to data type, into four main categories: thermal parameters (temperature, heat load, heat loss), hydraulic parameters (pressure, flow rate, density, viscosity), equipment operation (valve opening, pump frequency, equipment energy consumption), and user-side data (room temperature, heating period). Each category of data is labeled with corresponding monitoring point location, equipment number, data unit, and other attribute information.

[0061] Based on the pipeline network topology and data classification and labeling results, independent parameter monitoring units are divided. Each unit corresponds to a section of the pipeline or a group of devices. The data correlation within each unit and the boundary range between units are clearly defined to avoid data cross-interference. The DN300 secondary water supply main pipe in the eastern area and its covered 3 community branch roads and 12 residential buildings are divided into one monitoring unit. The pressure / flow data of the secondary main pipe, the heat loss data of the branch road, and the room temperature data of the building are correlated within the unit. The unit boundary is set as the connection node between the secondary main pipe and the main pipe, and the water distributor node between the branch road and the secondary main pipe. Data outside the boundary is not included in the correlation range of this unit.

[0062] By combining real-time pipeline network operation data streams, a load fluctuation analysis model is used to extract core indicators including the rate of change of heat load, peak demand, and attenuation trend, and to add abrupt change warning feature indicators for heat load fluctuations. The load fluctuation analysis model adopts a dual-module architecture of time-series data smoothing processing and feature extraction. The input data is real-time heat load time-series data stream (sampling frequency 5 minutes / time), and the output is the core indicators and warning features of heat load fluctuations. The key parameters of the model are set as follows: smoothing window duration of 15 minutes, and abrupt change judgment threshold of ±20% (instantaneous change rate of heat load).

[0063] The model calculates the rate of change of heat load through a sliding window, identifies peak heat load demand within a time period through a peak detection algorithm, and analyzes the decay trend of heat load through a linear fitting algorithm. These three types of indicators comprehensively reflect the dynamic characteristics of heat load changes. Real-time heat load data streams from branch roads in the commercial area are analyzed to extract core indicators such as "heat load increases from 1200kW to 1440kW between 10:00 and 10:15, with a rate of change of 16kW / minute," "peak heat load demand of 1500kW (occurring at 14:00)," and "heat load decays at a rate of 8kW / minute after 18:00," providing a basis for adjusting regulation strategies in the commercial area.

[0064] The model calculates the instantaneous rate of change of heat load by comparing the current heat load with the heat load of the previous cycle. When the rate of change exceeds a set threshold (±20%), a sudden change warning is triggered, generating warning characteristic indicators for "sudden increase in heat load" or "sudden decrease in heat load," providing signal support for emergency regulation. In a residential area, the heat load of a branch road in the previous cycle was 900kW, and the heat load in the current cycle suddenly increased to 1100kW, with an instantaneous rate of change of approximately 22.2%, exceeding the ±20% threshold. The model generated a warning characteristic indicator for "sudden increase in heat load," prompting the control system to adjust the flow distribution in a timely manner to avoid excessively high room temperature.

[0065] Using a heat loss quantification model, the real-time heat loss of the pipeline network is calculated based on pipeline insulation performance, friction distance, and ambient temperature difference. A self-learning prediction model for pipeline insulation performance degradation is introduced to predict future heat loss. A medium property correction algorithm is used to convert medium parameters such as circulating water density and viscosity into hydraulic adjustment correction coefficients, increasing the coupling correction relationship between medium temperature and density / viscosity, and differentiating the adjustment adaptability under different medium conditions. The heat loss quantification model uses pipeline insulation performance (thermal conductivity, breakage rate), friction distance, and ambient temperature difference as input parameters, and calculates the real-time heat loss of the pipeline network using heat transfer formulas. The model's calculation accuracy is controlled within ±3%.

[0066] The model employs a BP neural network architecture. The input layer consists of the pipe's service life, historical insulation performance data, environmental temperature and humidity data, and pipe laying method. There are three hidden layers (each with 32, 64, and 32 neurons respectively). The output layer shows the insulation performance degradation rate (increase in thermal conductivity and increase in breakage rate) for the next 1-3 months. The training steps are as follows: ① Collect historical data of different types of pipes over 5 years as the training set (sample size ≥ 1000 groups); ② Optimize the network weights using gradient descent, setting the learning rate to 0.001 and the number of iterations to 500; ③ Validate the model using a validation set (sample size ≥ 200 groups) to ensure the prediction error is ≤ 5%.

[0067] Using a media property correction algorithm, based on the real-time density and viscosity data of circulating water and combined with the media property parameter table in the industry standard, a hydraulic regulation correction coefficient (within the range of 0.8-1.2) is calculated. The larger the correction coefficient, the greater the adjustment range required under the media condition. At the same time, the coupling correction relationship between media temperature and density and viscosity is added, that is, the correction coefficient is dynamically adjusted with temperature changes to distinguish the adjustment adaptability under different media conditions.

[0068] Set a control priority matrix to clarify the control priority of each parameter: room temperature comfort (weight 0.3) > heat load matching degree (weight 0.25) > pressure / flow balance (weight 0.2) > equipment energy consumption (weight 0.15) > heat dissipation loss (weight 0.1); the dynamic correction coefficient is dynamically adjusted according to the real-time operating conditions, with a value range of 0.9-1.1. When a parameter deviates from the constraint threshold, the corresponding correction coefficient increases (e.g., when the room temperature is lower than the threshold, the room temperature correction coefficient is adjusted to 1.08).

[0069] Based on the control priority matrix, a weighted average method was used to calibrate multi-source data. The measured values ​​of each parameter were multiplied by the dynamic correction coefficient to obtain standardized calibrated data, eliminating the influence of data bias and operating condition fluctuations. The multi-source data of the monitoring unit in the eastern area were calibrated. The measured room temperature was 17℃ (below the threshold of 18℃, correction coefficient 1.08), and after calibration, it was 18.36℃; the measured heat load was 850kW (matching degree 89%, correction coefficient 1.02), and after calibration, it was 867kW; the measured pressure was 0.28MPa (deviation -0.02MPa, correction coefficient 1.05), and after calibration, it was 0.294MPa. The calibrated data better matches the control requirements.

[0070] By incorporating equipment operating energy consumption constraints (thresholds: pump power ≤ 15kW, valve energy consumption ≤ 0.5kW) and user room temperature comfort constraints (thresholds: residential 18-22℃, commercial 20-24℃), the calibrated data undergoes a secondary screening, removing data exceeding the constraint thresholds and recalibrating. For example, a pump's operating power, after calibration, was 15.8kW, exceeding the energy consumption constraint threshold of 15kW. Its correction factor was adjusted to 0.95, and after recalibration, the power was 15.01kW, meeting the constraint requirements. Similarly, a building's room temperature, after calibration, was 17.5℃, below the comfort threshold. The correction factor was readjusted to 1.1, and after calibration, the temperature was 19.25℃, meeting the constraint standards.

[0071] Integrating calibrated thermal, hydraulic, equipment, and user-side data, a structured dataset is generated according to the structure of "monitoring unit - parameter type - constraint threshold - calibration value - regulation adaptability". The dataset contains comprehensive information with clear parameter monitoring, clear load status, accurate loss calculation, and reasonable regulation adaptability, which can directly support the optimization of regulation strategies. The generated structured dataset of monitoring unit #1 in the East Area contains complete information such as "thermal parameters - heat load: 867kW - constraint threshold: 800-900kW - calibration value: 867kW - regulation adaptability: 0.92; hydraulic parameters - pressure: 0.294MPa - constraint threshold: 0.28-0.32MPa - calibration value: 0.294MPa - regulation adaptability: 0.95; user-side parameters - room temperature: 19.25℃ - constraint threshold: 18-22℃ - calibration value: 19.25℃ - regulation adaptability: 0.98".

[0072] Dynamic adaptive control, centered on "real-time operating condition matching + future trend prediction," optimizes and adjusts the original regulation strategy based on the heat load status, heat loss value, and constraint satisfaction in the structured data, clarifying the regulation priority, direction, and amplitude of each monitoring unit. For the operating condition of monitoring unit #1 in the East area, which is "heat load near peak (867kW), heat loss 22kW, and room temperature meets standard (19.25℃)," the optimized regulation strategy is to "maintain the current pump frequency of 37Hz, slightly adjust the branch valve opening from 25% to 26%, increase the branch flow rate to 92m³ / h, and maintain a dynamic balance between heat load and heat loss." For the trend of "predicted degradation of insulation performance over the next two months and an increase in heat loss to 25kW," the pre-optimized strategy is to "reserve 5% flow rate regulation redundancy, and gradually increase the pump frequency to 38.5Hz as heat loss increases."

[0073] Based on the optimized adjustment strategy, and combined with the medium property correction coefficient (e.g., currently 1.05) and pipeline heat loss data, the flow distribution ratio adjustment value for each branch is calculated to ensure that the flow distribution is accurately matched with heat load demand, heat loss, and medium status. The East Area No. 1 monitoring unit contains three sub-branches. The original flow distribution ratio was 40% for branch A, 35% for branch B, and 25% for branch C. Using structured datasets, branch A has a heat load share of 42%, a heat loss share of 45%, and a correction coefficient of 1.05; branch B has a heat load share of 33%, a heat loss share of 30%, and a correction coefficient of 1.02; and branch C has a heat load share of 25%, a heat loss share of 25%, and a correction coefficient of 1.00. After adjustment, the flow distribution ratio is changed to 43% for branch A, 34% for branch B, and 23% for branch C. This is achieved through the linkage of an electric balancing valve and a variable frequency circulating pump, ensuring precise matching of flow rate in each branch with actual demand.

[0074] S105, based on adjustment base data, thermal monitoring data and equipment operating parameters, through an automatic closed-loop adjustment system, relies on real-time heat load and heat loss parameters to dynamically adjust control commands, outputs precise balance adjustment results, and completes intelligent adjustment of the hydraulic balance of the heating network.

[0075] In one implementation, based on the requirements for switching control modes and the output targets of control results, the numerical characteristics, temporal changes, and correlation coupling relationships of basic control data, thermal monitoring data, and equipment operating parameters are collected in a structured and standardized manner to generate basic control data containing parameter types, fluctuation amplitudes, and control correlations. The core dimensions of data collection are clearly defined: numerical characteristics (specific parameter values, extreme values, and average values), temporal changes (the rate of change of parameters over time, fluctuation periods, and trend directions), and correlation coupling relationships (the correlation between different parameters, such as the negative correlation between pressure and flow rate, and the positive correlation between temperature and density). Data is collected in a structured manner according to "pipeline network level + parameter type" to ensure data coverage of the entire control chain.

[0076] The collected raw data undergoes standardized processing: unified data units (pressure MPa, flow rate m³ / h, temperature ℃, frequency Hz), standardized data format (retaining two decimal places), and elimination of dimensional influence (normalizing parameter values ​​and mapping them to the 0-1 range). Finally, basic control data containing "parameter type, fluctuation range, and control correlation" is generated, where the control correlation characterizes the degree of influence of the parameter on hydraulic balance (values ​​range from 0 to 1, with greater influence closer to 1).

[0077] Based on the coordinated characteristics of hydraulic regulation in heating networks, configuration rules for control data are designed, clarifying the switching thresholds and dynamic response coefficients for global-terminal and centralized-distributed modes, and generating multi-mode control configuration specifications. The thresholds for switching between two core modes are defined: the global-terminal mode switching threshold (based on the pressure difference between the network terminal and the global network) and the centralized-distributed mode switching threshold (based on the uniformity of regional heat load). These thresholds are set in conjunction with network design standards and actual operating experience to ensure precise switching timing. The threshold for switching between global and terminal modes is set to "pressure difference between terminal and global > 0.05MPa". When the terminal pressure difference of a certain community is 0.28MPa and the global main pipeline pressure difference is 0.35MPa, and the difference is 0.07MPa > the threshold, the switch from global mode to terminal mode is triggered. The threshold for switching between centralized and distributed modes is set to "regional heat load uniformity < 85%" (uniformity = average heat load of each branch / maximum heat load of the branch). When the heat loads of three branches in a certain area are 100kW, 80kW, and 60kW respectively, and the uniformity = 80 / 100 = 80% < the threshold, the switch from centralized mode to distributed mode is triggered.

[0078] The dynamic response coefficient characterizes the sensitivity of a parameter to control commands (value 1-2, with a larger coefficient indicating a more sensitive response). It is dynamically adjusted based on parameter type, network level, and operational complexity to ensure efficient execution of control commands in different scenarios. The dynamic response coefficients are set as follows: main pipeline pressure response coefficient 1.8 (for rapid response in global control), terminal branch flow response coefficient 1.5 (for precise adaptation in terminal adjustments), and pump frequency response coefficient 1.2 (for smooth equipment operation). Under complex conditions (such as extreme cold waves), all parameter response coefficients are increased by 0.2 to enhance control sensitivity.

[0079] The switching threshold and dynamic response coefficient are integrated, and configuration specifications are generated according to the structure of "mode type - judgment condition - response parameter - coefficient value", which clarifies the control logic and parameter requirements under different modes. The generated multi-mode control configuration specification entries are as follows: "Mode type: global mode, judgment condition: pressure difference between the terminal and the global system ≤ 0.05MPa, response parameters: main pipe pressure, total flow rate, response coefficient: 1.8, 1.6; Mode type: terminal mode, judgment condition: pressure difference between the terminal and the global system > 0.05MPa, response parameters: terminal branch flow rate, valve opening, response coefficient: 1.5, 1.7; Mode type: centralized mode, judgment condition: regional heat load uniformity ≥ 85%, response parameters: total pump frequency, main pipe temperature, response coefficient: 1.2, 1.4; Mode type: distributed mode, judgment condition: regional heat load uniformity < 85%, response parameters: pump frequency of each branch, valve opening of each branch, response coefficient: 1.3, 1.6".

[0080] Based on the closed-loop control system architecture and command output requirements, a multi-level control mechanism is established, encompassing real-time parameter sensing, dynamic command generation, and intelligent mode switching. Distributed sensing nodes (pipeline node sensors, equipment IoT modules, and user-side room temperature data acquisition devices) are deployed, with a set sensing frequency (1 minute / time for core parameters, 5 minutes / time for ordinary parameters). An abnormal sensing and early warning mechanism is established. When parameters exceed the normal range (e.g., pressure > 0.4 MPa, room temperature < 16℃), the sensing frequency is automatically increased to 30 seconds / time, and an early warning signal is triggered. Sensing nodes deployed on the terminal branches in the southern area normally collect room temperature data every 5 minutes. When a room temperature of 15.5℃ is collected (below the normal range of 16-22℃), the sensing frequency automatically increases to 30 seconds / time, and a "low room temperature" early warning signal is sent to the control system.

[0081] Based on the basic data of regulation and control and the multi-mode configuration specifications, an instruction generation logic is established: taking the hydraulic balance target (differential pressure balance, flow rate adaptation, room temperature compliance) as the core, and combining the real-time parameter sensing results, the regulation instructions are dynamically output through the process of "deviation calculation - coefficient matching - instruction generation". The instructions include the execution device, adjustment direction, and adjustment range. The room temperature of the terminal branch in the southern area is sensed to be 15.5℃ (deviation -0.5℃). The response parameters (branch flow rate) and response coefficient (1.5) of the terminal mode in the configuration specifications are matched, and it is calculated that "the branch flow rate needs to be increased by 5m³ / h". The regulation instruction is generated as follows: "Execution device: No. 1 electric balance valve, adjustment direction: increase opening, adjustment range: 10% (corresponding to a flow rate increase of 5m³ / h)".

[0082] Based on the judgment conditions of the multi-mode control configuration specification, the trigger threshold is monitored in real time. When the switching conditions are met, the mode switching process is automatically initiated, and the response parameters and response coefficients are adjusted synchronously to ensure a smooth and shock-free switching process. After the switch, the control command adapted to the new mode is generated immediately. Real-time monitoring shows that the heat load uniformity in the southern area has dropped from 86% to 84% (below the centralized mode judgment threshold of 85%). The switching process from centralized mode to distributed mode is automatically initiated. The response parameters are switched from "total pump frequency and main pipe temperature" to "branch pump frequency and branch valve opening". The response coefficients are adjusted synchronously to 1.3 and 1.6, and the command adapted to the distributed mode is generated: "Increase the frequency of branch pump #2 by 3Hz and increase the opening of branch valve #3 by 8%".

[0083] Using basic control data as input, multi-mode control configuration specifications as guidelines, and a multi-level control mechanism as the execution path, the system achieves real-time linkage of data, specifications, and mechanisms through a system bus, ensuring that the generation and execution of control commands meet operating conditions. The South Area Control System integrates relevant elements, inputs basic control data (pressure fluctuation 0.04MPa, flow fluctuation 30m³ / h), and, based on the multi-mode configuration specifications (currently meeting the centralized mode judgment conditions), generates and issues initial control commands for "maintaining the main pump frequency at 38Hz and the main pipeline water supply temperature at 85℃" through a multi-level control mechanism (sensing frequency once every 1 minute).

[0084] The closed-loop control system collects real-time pipeline operation data (such as actual changes in pressure, flow rate, and room temperature) after command execution, compares it with the control targets (such as differential pressure balance ±0.03MPa, room temperature 18-22℃), calculates the deviation value, and immediately optimizes the control strategy and adjusts the control command if the deviation exceeds the allowable range. One hour after the initial command was executed, the system detected that the room temperature of branch line 3 in the south area was 17℃ (lower than the target lower limit of 18℃), with a deviation of -1℃, exceeding the allowable deviation ±0.5℃. Through closed-loop feedback optimization strategy: matching the distribution mode configuration specification, adjusting the response parameter to the valve opening of branch line 3, with a response coefficient of 1.6, and generating the optimization command: "Increase the opening of the electric balance valve of branch line 3 by 15%, and increase the flow rate by 7m³ / h."

[0085] The system continuously monitors the mode switching threshold. When the operating conditions change and meet the switching conditions, it automatically completes the seamless mode switching. During the switching process, a "gradual adjustment" is adopted (such as adjusting the pump frequency by 1Hz every 30 seconds to avoid sudden parameter changes) to ensure the stable operation of the pipeline network. Finally, the system generates a pipeline network hydraulic balance control result that includes "precise adjustment parameters, intelligent switching instructions, and dynamic adaptation schemes" to achieve intelligent regulation of the hydraulic balance of the heating pipeline network. Two hours after the optimization command was executed, the heat load uniformity in the southern area increased from 84% to 87% (higher than the centralized mode threshold of 85%). The system automatically initiated a seamless switch from distributed mode to centralized mode, with the pump frequency gradually changing from 41Hz to 38Hz at a rate of 1Hz every 30 seconds, and the valve opening being finely adjusted synchronously. The final control result was: "Precise adjustment parameters: main pipe pressure 0.34MPa, total flow rate 115m³ / h, average room temperature 19.5℃; intelligent switching command: centralized--distributed--centralized (within 24 hours); dynamic adaptation scheme: distributed mode is activated during the morning peak (7:00-9:00), and centralized mode is activated during other times," completing the intelligent adjustment of the hydraulic balance of the pipe network in this area.

[0086] In one implementation, such as Figure 2 As shown, this application also provides a regulating device for the hydraulic balance of a heating network, comprising: The multi-source operation data acquisition module 201 is used to collect operation data of the heating network in all scenarios, monitor the branch heat load demand, heat loss along the pipeline, valve throttling loss, circulating water density, pipeline insulation performance, power of terminal heat dissipation equipment, and user heat consumption period patterns in real time, and generate multi-source heterogeneous fusion hydraulic balance regulation basic data. The initial regulation scheme generation module 202 is used to generate an initial hydraulic balance regulation scheme based on the thermal-hydraulic coupling model, combined with feedforward prediction, feedback calibration, and dynamic adaptation control mode. It takes the basic regulation data as input to calculate the branch pressure difference optimization value and flow distribution coefficient, and performs feedforward planning based on user heat load demand and heat use period patterns. The regulation scheme optimization and calibration module 203 is used to compare and analyze the basic regulation data with historical balance operating condition data. Through intelligent linkage electric balance valve and variable frequency circulating pump, the initial regulation scheme is converted into valve opening parameters and pump operating frequency. Then, based on the deviation of heat loss along the pipeline, the actual heat supply of the terminal equipment, and the feedback parameters of circulating water density change, combined with multi-dimensional data of real-time room temperature feedback from the user side, feedback calibration is performed to optimize the hydraulic balance regulation scheme. The dynamic adaptation module 204 for adjusting strategies is used to continuously record the thermal parameters of the pipeline network and the operating data of the equipment. In response to the problems of heat load fluctuation and insulation performance degradation, the adjustment strategy is optimized through dynamic adaptation control based on the user's heat load demand and pipeline heat loss. A self-learning prediction model for pipeline insulation performance degradation is introduced, and the flow distribution ratio is adjusted by combining circulating water density data and coupled change data of medium viscosity and temperature. The closed-loop intelligent regulation execution module 205 is used to intelligently regulate the hydraulic balance of the heating network by dynamically adjusting control commands based on the basic regulation data, thermal monitoring data and equipment operating parameters through the automatic closed-loop regulation system and relying on real-time heat load and heat loss parameters, and outputting precise balance regulation results.

[0087] The computer-readable storage medium provided in the above embodiments of this application and the method for adjusting the hydraulic balance of heating pipe networks provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0088] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for evaluating the adjustment method, system, electronic device, and readable storage medium for the hydraulic balance of a heating network are basically similar to the embodiments for the adjustment method for the hydraulic balance of a heating network described above, and are therefore described simply. Relevant parts can be referred to in the descriptions of the embodiments for the adjustment method for the hydraulic balance of a heating network described above.

Claims

1. A method for adjusting the hydraulic balance of a heating network, characterized in that, include: Collect full-scenario operation data of heating pipe network, monitor in real time branch heat load demand, heat loss along pipe network, valve throttling loss, circulating water density, pipe insulation performance, power of terminal heat dissipation equipment, and user heat consumption period patterns, and generate multi-source heterogeneous fusion hydraulic balance regulation basic data; Based on the thermal-hydraulic coupling model, combined with feedforward prediction, feedback calibration, and dynamic adaptation control mode, the branch pressure difference optimization value and flow distribution coefficient are calculated by inputting basic regulation data. The initial regulation scheme is generated by feedforward planning based on user heat load demand and heat use period patterns. By comparing and analyzing the basic adjustment data with historical balance operating data, the initial adjustment scheme is transformed into valve opening parameters and pump operating frequency through intelligent linkage electric balance valves and variable frequency circulating pumps. Then, based on the feedback parameters of heat loss deviation along the pipeline, actual heat supply of terminal equipment, and change in circulating water density, and combined with multi-dimensional data of real-time room temperature feedback from the user side, feedback calibration is performed to optimize the adjustment scheme. Continuously record the thermal parameters of the pipeline network and equipment operation data. In response to heat load fluctuations and insulation performance degradation, optimize the adjustment strategy through dynamic adaptation control based on user heat load demand and pipeline heat loss. Introduce a self-learning prediction model for pipeline insulation performance degradation, and adjust the flow distribution ratio by combining circulating water density data and coupled change data of medium viscosity and temperature. Based on basic adjustment data, thermal monitoring data, and equipment operating parameters, the automatic closed-loop adjustment system dynamically adjusts control commands according to real-time heat load and heat loss parameters, outputs precise balance adjustment results, and completes intelligent hydraulic balance adjustment of the heating network.

2. The method as described in claim 1, characterized in that, Based on a thermal-hydraulic coupling model, and combining feedforward prediction, feedback calibration, and dynamic adaptive control modes, the optimized branch pressure difference and flow distribution coefficient are calculated using input regulation basic data. Based on user heat load demand and heating period patterns, feedforward planning is performed to generate an initial regulation scheme, including: The basic data of hydraulic balance regulation is processed and layered according to the pipeline branch level and heat load type. The data of each layer includes node parameters, branch parameters and user-side parameters. The dynamic characteristics of heat load fluctuation, medium property change and heat consumption law change are aggregated through the dynamic monitoring window. The pipeline resistance, medium transportation and heat demand supply data of each level are extracted to generate corresponding subsets. Using pipeline branches as core nodes, hydraulic parameters as control nodes, and heat load demand as target nodes, node attribute information is generated by combining pipeline topology mapping and heat load type identification, and edge association information is generated according to parameter fit and heat demand response time sequence. A directed graph containing all branch nodes of the heating network, multiple types of hydraulic control nodes, and nodes with differentiated heat load demands is constructed. It is divided into three views according to hydraulic constraints, thermal constraints, and heat consumption constraints. The cross-view edge correlation strength is optimized through a constraint collaborative attention mechanism. Based on the thermal-hydraulic coupling model, combined with feedforward prediction, feedback calibration, and dynamic adaptation control mode, the optimized value of branch pressure difference and flow distribution coefficient are calculated by inputting the corresponding subset data. Feedforward planning is performed based on user heat load demand and heat consumption period patterns to generate an initial scheme for hydraulic balance regulation of the heating network.

3. The method as described in claim 1, characterized in that, By comparing and analyzing the basic adjustment data with historical balanced operating data, and through intelligent linkage electric balancing valves and variable frequency circulating pumps, the initial adjustment plan is transformed into valve opening parameters and pump operating frequency. Feedback calibration is then performed based on feedback parameters such as deviations in heat loss along the pipeline, actual heat supply from terminal equipment, and changes in circulating water density. Combined with multi-dimensional data from real-time room temperature feedback on the user side, the adjustment plan is optimized, including: Integrated data comparison technology is used to process the multi-source heterogeneous fusion of regulation basic data and historical balance condition data to generate data deviation analysis results, condition matching quantification data, pipeline network operation characteristic set, and hydraulic parameter adaptation trend prediction results. The feature set is integrated with the hydraulic regulation association standard, parameter correction degree ranking rules and operating condition adaptation threshold information, and the threshold requirements of user-side room temperature comfort are incorporated to establish a precise matching relationship between features and regulation parameter correction requirements, and generate a standardized regulation parameter set with multi-dimensional constraints. Based on a standardized set of multi-dimensional constraints, core control information is extracted. The initial control scheme is used as the input dimension, hydraulic deviation characteristics are used as the core parameters, and parameter correction degree evaluation rules are used as the judgment basis. Dynamic optimization judgment indicators of reinforcement learning are added to achieve accurate reading of branch pressure difference target value, flow distribution coefficient, valve adjustment range and pump frequency range. By combining the correction intensity of parameters for hydraulic balance, the priority requirements of regulation, and the weight of user-side room temperature feedback, the core regulation information is quantified and sorted for optimization. The correlation weights, equipment adaptability attributes, and adjustment ranges of various regulation parameters are clarified, and the constraint thresholds for equipment operating energy consumption are added. A list of regulation parameters arranged in order of correction degree and correlation is generated. Through intelligent linkage electric balancing valves and variable frequency circulating pumps, these parameters are converted into valve opening parameters and pump operating frequency. Then, based on the feedback parameters of heat loss deviation along the pipeline, actual heat supply of terminal equipment, and changes in circulating water density, and combined with multi-dimensional data of real-time user-side room temperature feedback, feedback calibration is performed to optimize the hydraulic balance regulation scheme of the heating network.

4. The method as described in claim 1, characterized in that, Continuously record pipeline thermal parameters and equipment operation data. Addressing heat load fluctuations and insulation performance degradation, optimize adjustment strategies through dynamic adaptive control based on user heat load demands and pipeline heat loss. Introduce a self-learning prediction model for pipeline insulation performance degradation, and adjust flow distribution ratios by combining circulating water density data with coupled changes in medium viscosity and temperature. This includes: Based on the full-dimensional data storage algorithm, the thermal parameters of the pipeline network and the equipment operation data are processed by time-series archiving and classification labeling, and the boundary between parameter monitoring units and data association is delineated. By combining real-time operation data streams of the pipeline network, core indicators including the rate of change of heat load, peak demand, and attenuation trend are extracted through a load fluctuation analysis model, and abrupt change warning characteristic indicators of heat load fluctuation are added. Using a heat loss quantification model, the real-time heat loss value of the pipeline network is calculated based on pipeline insulation performance, friction distance, and ambient temperature difference. A self-learning prediction model for pipeline insulation performance decay is introduced to predict future heat loss. A medium property correction algorithm is used to convert medium parameters such as circulating water density and viscosity into hydraulic regulation correction coefficients, increasing the coupling correction relationship between medium temperature and density and viscosity, and distinguishing the regulation adaptability under different medium conditions. The linkage control priority matrix and dynamic correction coefficient complete the calibration of multi-source operation data, and incorporate the dual constraint coefficients of equipment operation energy consumption and user room temperature comfort. Finally, a multi-dimensional control aligned structured operation dataset with clear parameter monitoring, clear load status, accurate loss calculation and reasonable adjustment is generated. Based on user heat load demand and pipeline heat loss, the adjustment strategy is optimized through dynamic adaptation control. A self-learning prediction model for pipeline insulation performance degradation is introduced, and the flow distribution ratio is adjusted by combining circulating water density data and coupled change data of medium viscosity and temperature.

5. The method as described in claim 4, characterized in that, Based on basic adjustment data, thermal monitoring data, and equipment operating parameters, seamless switching between centralized global control and distributed terminal control is achieved. Through an automatic closed-loop adjustment system, dynamic adjustment commands are generated based on real-time heat load and heat loss parameters to output precise balance adjustment results, completing intelligent hydraulic balance adjustment of the heating network, including: Based on the requirements for switching control modes and the output targets of control results, the numerical characteristics, temporal changes, and correlation coupling relationships of basic control data, thermal monitoring data, and equipment operating parameters are collected in a structured and standardized manner to generate basic control data containing parameter types, fluctuation amplitudes, and control correlations. Based on the coordinated characteristics of hydraulic regulation in heating pipe networks, configuration rules for regulation data are designed, and the switching thresholds for global-terminal and centralized-distributed regulation, as well as the dynamic response coefficients of parameters, are clarified to generate multi-mode regulation configuration specifications. Based on the closed-loop control system architecture and command output requirements, a multi-level control mechanism is set up, which includes real-time parameter sensing, dynamic command generation, and intelligent mode switching. The system integrates and executes basic control data, multi-mode control configuration specifications, and multi-level control mechanisms. Through real-time feedback from a closed-loop system, it optimizes control strategies and generates hydraulic balance control results for the heating network, including precise adjustment parameters, intelligent switching instructions, and dynamic adaptation schemes. This enables seamless switching between global centralized control and terminal distributed control. By dynamically adjusting control instructions based on real-time heat load and heat loss parameters, it outputs precise balance control results and completes intelligent hydraulic balance control of the heating network.

6. A regulating device for hydraulic balance in heating pipe networks, characterized in that, The device is used to execute executable instructions to perform the method for adjusting the hydraulic balance of a heating network as described in any one of claims 1 to 5.

7. An electronic device, characterized in that, include: First processor; The processor also includes a memory for storing executable instructions of the first processor, wherein the first processor is configured to execute the method for adjusting the hydraulic balance of a heating network as described in any one of claims 1 to 5 by executing the executable instructions.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the method for adjusting the hydraulic balance of the heating network as described in any one of claims 1 to 5.

Images