A valve opening control method for user-side hydraulic balance

Abstract

The present application relates to the field of valve opening control technology, and discloses a valve opening control method for user-end hydraulic balance. The method includes: collecting parameters of the underground pipe cold storage system to obtain dynamic system parameter data; performing cooling load fluctuation spectrum analysis and multi-scale decomposition based on the dynamic system parameter data to obtain a user demand prediction model and an ideal flow distribution ratio for hydraulic balance; generating a comprehensive system state evaluation index based on the user demand prediction model and the ideal flow distribution ratio for hydraulic balance; constructing a global optimization objective function based on the comprehensive system state evaluation index, and solving it to obtain a time-divided valve adjustment strategy; constructing a valve flow characteristic model based on the time-divided valve adjustment strategy and the valve historical response data, and calculating the target opening of each regulating valve, and outputting a valve control execution instruction. The present application can dynamically adjust the hydraulic balance strategy according to different working conditions, and realize flexible coordination between hydraulic balance and cooling demand.

Description

Technical Field

[0001] The present application relates to the technical field of valve opening control, and in particular to a valve opening control method for user-end hydraulic balance. Background Art

[0002] As a highly efficient and energy-saving temperature control technology, underground pipe thermal storage systems are widely used in green buildings and district energy systems. This system utilizes the relatively stable underground temperature to store and release thermal energy by circulating fluid through underground pipes. This effectively addresses energy peaks and valleys and improves energy efficiency. However, in practice, due to factors such as uneven distribution of cooling user loads, complex and variable pipe networks, and external environmental interference, underground pipe thermal storage systems often experience hydraulic imbalances on the end-user side. This results in insufficient cooling for some users and excessive cooling for others, reducing the overall system energy efficiency and impacting user comfort and satisfaction.

[0003] Traditional user-side hydraulic balancing methods rely primarily on manual experience to adjust static balancing valves, which cannot adapt to the dynamic changes in cooling loads. While some improved solutions have introduced dynamic automatic adjustment technology, they often only consider a single parameter (such as pressure differential or flow rate) for local optimization, lacking a comprehensive analysis of the system's overall operating status and multi-objective collaborative optimization. This is especially true in complex underground pipe cold storage systems, where the control accuracy and response speed of traditional methods cannot meet actual requirements due to the numerous heat exchange links and high system inertia. Furthermore, most existing technologies use fixed parameter control strategies, failing to fully utilize system operating data and advanced data analysis techniques for adaptive optimization, resulting in low long-term system efficiency. Summary of the Invention

[0004] The present application provides a valve opening control method for household hydraulic balance, which can dynamically adjust the hydraulic balance strategy according to different working conditions to achieve flexible coordination between hydraulic balance and cooling demand.

[0005] In a first aspect, the present application provides a valve opening control method for household hydraulic balancing, the valve opening control method for household hydraulic balancing comprising:

[0006] Collect parameters of the underground pipe cold storage system to obtain dynamic system parameter data;

[0007] Based on the dynamic system parameter data, cooling load fluctuation spectrum analysis and multi-scale decomposition are performed to obtain a user demand prediction model and an ideal flow distribution ratio for hydraulic balance;

[0008] Generate a comprehensive system status evaluation index based on the user demand prediction model and the ideal flow distribution ratio of hydraulic balance;

[0009] Constructing a global optimization objective function based on the comprehensive system state evaluation index and solving it to obtain a time-divided valve adjustment strategy;

[0010] A valve flow characteristic model is constructed according to the time-division valve regulation strategy and valve historical response data, and the target opening of each regulating valve is calculated, and a valve control execution instruction is output.

[0011] In the technical solution provided in this application, a complete dynamic database of system parameters is constructed by setting up multiple groups of sensors at the target nodes, providing a comprehensive and accurate data basis for hydraulic balance control; the cooling load fluctuation spectrum analysis and multi-scale decomposition technology are used to accurately grasp the load characteristics, and users are classified according to the load stability index, and a differentiated prediction model is used to improve the load prediction accuracy; a comprehensive system status evaluation index is constructed to realize the quantitative characterization of the degree of hydraulic imbalance, so that the control is transformed from qualitative judgment to quantitative control; a global optimization function is constructed based on the multi-objective optimization idea, and the optimal adjustment strategy that takes into account system efficiency and user comfort is obtained through genetic algorithm solution; an adaptively corrected flow characteristic model is established to overcome the actual characteristic deviation of the valve, and a smoothing algorithm is used to avoid equipment wear caused by frequent adjustment; a reinforcement learning algorithm is introduced to form an adaptive control mechanism with closed-loop feedback to continuously optimize the adjustment strategy; the system reliability is improved through valve health status evaluation, and the hydraulic balance strategy is dynamically adjusted according to different working conditions to achieve flexible coordination between hydraulic balance and cooling demand to meet diversified operation needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 This is a schematic diagram of an embodiment of a valve opening control method for user-end hydraulic balance in an embodiment of the present application. DETAILED DESCRIPTION

[0014] An embodiment of the present application provides a valve opening control method for user-end hydraulic balancing. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or apparatus.

[0015] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the present application, a method for controlling the valve opening for user-side hydraulic balance includes:

[0016] Step S101: Collect parameters of the underground pipe cold storage system to obtain dynamic system parameter data;

[0017] It is understandable that the execution subject of this application can be a valve opening control system for user-side hydraulic balance, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0018] Specifically, temperature sensors and pressure sensors are installed on the branch pipes on the cold user side of the underground pipe thermal storage system to collect chilled water supply and return temperatures, supply and return pressures, and reflect the heat load and hydraulic conditions on the cold user side. Electromagnetic flowmeters are installed on each user branch to collect flow data, understanding individual cooling needs and ensuring precise adjustment of system flow distribution to meet these needs. This data collection on the cold user side generates a cold user parameter dataset containing key parameters such as chilled water flow, temperature, and pressure. Furthermore, temperature sensors are installed at the inlet and outlet of the underground pipe-in-tube heat exchanger to collect inlet and outlet temperature data, providing real-time insights into the heat exchange system's operating status and heat transfer performance. Temperature sensors are installed at key locations on pre-cooling evaporative cooling chillers to collect air temperatures before and after the direct evaporative cooling packing, as well as the water temperature in the cold water tank. This temperature data helps monitor the efficiency of the heat exchange system, understand the heat exchange process, and ensure efficient system operation. By collecting temperature data from the heat exchange system, cooling load distribution can be optimized and the system's energy efficiency improved. Pressure sensors and temperature sensors are installed inside the cold storage modular unit to collect the pressure and temperature data of the refrigerant before and after the four-way reversing valve, as well as the pressure and temperature at the outlet of the fluorine pump. These data are key indicators for understanding the performance of the refrigeration system. They can reflect the flow state of the refrigerant, the efficiency of the compression process, and the changes in system pressure, thereby evaluating the operating status of the refrigeration module. An electric regulating valve is installed on the water supply main of each cold user to monitor the valve opening and flow changes in real time. By collecting the operating parameters of these core components of the system, a parameter data set for the operation of the core components of the system is obtained. The complete parameter data set on the cold user side, the heat exchange system operation parameter data set, and the system core component operation parameter data set are standardized to convert data from different sources and types into comparable dynamic system parameter data.

[0019] Step S102: performing cooling load fluctuation spectrum analysis and multi-scale decomposition based on dynamic system parameter data to obtain a user demand prediction model and an ideal flow distribution ratio for hydraulic balance;

[0020] Specifically, the real-time cooling load of each customer branch is extracted from dynamic system parameter data. The cooling load calculation relies on the supply / return water temperature difference and flow rate data for each customer branch. These data are collected by installed temperature sensors and flow meters to generate the real-time cooling load. All real-time cooling load data is then transformed into a three-dimensional matrix based on customer identification and time dimensions, forming a cooling load distribution matrix that effectively describes the cooling load variations across different time periods and customer branches. Based on the cooling load distribution matrix, a spectral analysis of the cooling load time series is performed to reveal the periodic characteristics of load fluctuations. A Fourier transform is used to identify the periodic patterns in load fluctuations, and a wavelet transform is used to perform a multi-scale decomposition of the periodic characteristics. Load fluctuations are then broken down into three different types: base load, cyclical load, and random load, enabling a more accurate understanding of the changing trends in customer demand. Base load represents constant customer demand, cyclical load reflects load fluctuations caused by daily life or seasonal changes, and random load reflects unpredictable demand fluctuations. Based on the decomposed load types, the historical mean and standard deviation of the load for each customer branch are calculated to derive a load stability index. The load stability index is a key parameter for measuring the stability of user demand fluctuations. By calculating the ratio of the standard deviation to the mean for each branch, users are categorized into three categories: high stability, medium stability, and low stability. High-stability users experience smaller load fluctuations, while low-stability users experience larger fluctuations. Demand forecasting uses different models based on the stability of each user. For high-stability users, a linear regression model is used to predict their load, as their load fluctuations are relatively regular. For medium-stability users, a seasonal ARIMA model is used to effectively capture seasonal fluctuations. For low-stability users, a long short-term memory network (LSTM) model is used to capture the highly nonlinear characteristics of load fluctuations. Based on each user's supply and return water temperature difference, flow data, and pipeline characteristics, a series of hydraulic models are used to calculate the ideal flow distribution ratio under hydraulic balance. This ratio serves as a reference for subsequent hydraulic balance adjustments.

[0021] Step S103: Generate a comprehensive system status evaluation index based on the user demand prediction model and the ideal flow distribution ratio of hydraulic balance;

[0022] Specifically, the deviation rate between the actual flow rate of each customer branch and the ideal flow rate calculated based on the ideal flow distribution ratio for hydraulic balance is calculated to produce a global hydraulic imbalance index. This index reflects the difference between the actual flow rate of each branch and the ideal flow rate, providing a basis for valve opening adjustment. A large deviation from this index indicates a significant deviation from the hydraulic balance and requires adjustment. Based on customer-side pressure sensor data, the pressure distribution of the cold water supply and return network is analyzed to determine the network pressure distribution weighting coefficient. This coefficient helps determine which areas of pressure have a greater impact on the hydraulic balance during system adjustment, allowing for more careful consideration when adjusting valves and ensuring overall system stability during adjustment. Based on the supply and return water pressure differential and flow rate of each customer branch, the flow resistance coefficient of each branch is calculated and compared with the average flow resistance coefficient of the entire network to obtain the flow resistance anomaly coefficient. This coefficient reflects the deviation of the flow resistance of certain customer branches from the overall system, helping the system identify branches with hydraulic imbalance or flow problems as key areas for subsequent adjustment. To improve cooling capacity allocation accuracy, the cooling capacity allocation deviation index is corrected and calculated based on load forecast data output by the user demand prediction model and the actual flow rate of each branch. This correction process incorporates the flow resistance anomaly coefficient to improve cooling capacity allocation accuracy, ensuring that each user receives appropriate cooling capacity based on their actual needs and avoiding uneven cooling and heating in the system or over-adjustment. While correcting cooling capacity allocation deviation, the performance of the underground pipe-in-pipe heat exchanger is also calibrated. The efficiency of the heat exchange process is evaluated using collected underground pipe heat exchanger inlet and outlet temperature data, combined with the pipe network pressure distribution weight coefficient, to determine the underground pipe heat transfer efficiency. A weighted combination of the global hydraulic imbalance index, cooling capacity allocation deviation index, and underground pipe heat transfer efficiency is used to generate a comprehensive system status assessment indicator, reflecting the overall system operating status. This comprehensive assessment of these indicators determines whether the current operating status is optimal, whether valve opening adjustment is necessary, and the priority and strategy for adjustment. This ensures precise hydraulic balance control and ensures efficient and stable operation of the underground pipe cold storage system under various load conditions.

[0023] Step S104: constructing a global optimization objective function based on the comprehensive system state evaluation index, and solving it to obtain a time-divided valve adjustment strategy;

[0024] Specifically, the current level of regulation urgency is graded using comprehensive system status assessment indicators. This grading process determines the system's current operating state based on the values ​​of indicators such as the global hydraulic imbalance index, cooling capacity distribution deviation index, and underground pipe heat exchange efficiency. A high global hydraulic imbalance index indicates uneven hydraulic distribution and a more urgent system regulation state. A high cooling capacity distribution deviation index indicates significant cooling capacity distribution errors and requires priority regulation. This assessment prioritizes different regulation states. Based on the regulation priority assessment results, a global optimization objective function is constructed. This objective function incorporates three key factors: user load demand weights, pressure differential balance, and temperature uniformity. User load demand weights reflect the differences in cooling load demand among different users, helping the system prioritize users with higher loads. Pressure differential balance reflects the pressure distribution within the chilled water network, ensuring that system regulation avoids excessively high or low pressure in certain areas, thus maintaining overall system stability. Temperature uniformity measures the temperature distribution across users within the system, ensuring even cooling load distribution and preventing discomfort for some users due to large temperature differences. By setting these weights, the objective function comprehensively considers the needs of various aspects and achieves a globally optimal hydraulic balance. The global optimization objective function is then solved by configuring the solution parameters, adjusting the weights of different variables and the solution parameters to obtain a set of solution parameters appropriate for the current system state. The load forecast data output by the user demand prediction model is divided into time periods and, combined with the magnitude of load fluctuations, a load period characteristic classification result is generated. By analyzing the periodicity and volatility of load fluctuations, load demand in different time periods is classified, providing time-specific load information for formulating valve regulation strategies. This load period characteristic classification helps the system identify time periods with significant load fluctuations and those with more stable loads. This allows the system to determine which periods have greater load fluctuations and which periods are more stable, thereby providing greater regulation margins during periods of greater load fluctuations and finer regulation during periods of lesser load fluctuations. Based on the load period characteristic classification results, constraints for valve regulation are set. These constraints include the valve's adjustment range, adjustment speed, and adjustment frequency, ensuring that over-regulation of the valve during regulation does not cause system instability. The boundary constraints are combined with the objective function to form an optimization problem. After inputting the global optimization objective function, the set of solution parameters, and the boundary constraints for valve regulation into the solver, calculations are performed to determine the time-based valve regulation strategy. This strategy guides the system on how to adjust valve openings during different time periods to ensure hydraulic balance and maximize overall system efficiency and energy savings.

[0025] Step S105: construct a valve flow characteristic model based on the time-division valve adjustment strategy and the valve historical response data, calculate the target opening of each regulating valve, and output a valve control execution instruction.

[0026] Specifically, each control valve is classified by flow characteristic. Each valve has different flow characteristics, including equal percentage valves, linear valves, and quick-opening valves. Flow characteristic models are developed based on valve type. These models reflect the relationship between valve opening and flow rate, with each valve type exhibiting its own specific flow response pattern. Regression analysis is performed on historical valve response data. By analyzing the deviation between each valve's actual flow response and its theoretical flow response, a valve characteristic correction factor is calculated. Regression analysis reveals the relationship between valve and flow rate response during actual control and calculates a correction factor. This correction factor compensates for the discrepancy between the valve characteristic model and actual operation, making the flow characteristic model more consistent with actual conditions. Based on the target flow distribution determined by the time-based valve control strategy and current differential pressure data, a reverse calculation is performed on the valve's flow characteristic model to determine the target opening for each control valve. The target flow distribution is derived from user demand forecasts and system load analysis. Through reverse calculation, the target valve opening is determined to ensure that the flow rate reaches the target value. The target opening is then compared with the current valve opening, and the difference between them is calculated. A threshold judgment is performed on the difference to ensure that the valve opening does not fluctuate too drastically. If the difference between the target opening and the current opening exceeds the set threshold, appropriate adjustments are made to ensure a smooth valve adjustment process and avoid system instability caused by frequent adjustments. This judgment determines the target valve opening value and avoids over-adjustment. Based on the determined valve target opening value and the adjustment time window in the time-based valve adjustment strategy, valve control execution instructions are generated, specifying the opening value, adjustment rate, and adjustment time period for each valve to be adjusted. This ensures that the valve can be adjusted according to the precise adjustment strategy to achieve global hydraulic balance optimization.

[0027] Valve control execution instructions are prioritized based on the urgency of hydraulic balance. The urgency of hydraulic balance is determined by calculating the global hydraulic imbalance index and the cooling capacity distribution deviation index. Large indices indicate that the system's hydraulic balance is unsatisfactory and require immediate adjustment. Based on this, the execution order of adjustment instructions is determined based on the adjustment priority of each valve. Valve control execution instructions are transmitted to the field controller via Modbus or BACnet protocols. The field controller drives the electric actuator to adjust the valve opening according to the instructions, achieving the desired hydraulic balance. The system status after each valve adjustment is compared and analyzed, and the rate of change of the hydraulic imbalance index, cooling capacity distribution deviation index, and comprehensive system status assessment index before and after adjustment is calculated. By monitoring the changes in these indicators, the adjustment effect is quantified to obtain the adjustment effect index. This index reflects the effectiveness of the adjustment action. A large rate of change indicates significant adjustment results and effective improvement in system hydraulic balance. A small rate of change indicates insignificant adjustment results and requires further optimization of the control strategy. Data is collected on the adjustment response characteristics of each valve. The dynamic response characteristics of the valve are reflected by recording the time delay and response gain from issuing a control command to flow stabilization. This data is modeled using a first-order hysteresis model to account for the hysteresis present in the valve adjustment process. This model helps the system better predict the time characteristics of the valve response, ensuring that subsequent adjustments account for this delay, thereby improving system adjustment accuracy and response speed. Based on the adjustment effect index and the first-order hysteresis model, a set of valve adaptive correction parameters is calculated to adjust the valve flow characteristics to compensate for errors or hysteresis introduced during the adjustment process. Through adaptive correction, the control strategy is dynamically adjusted to ensure that valve adjustment remains optimal under different operating conditions, improving hydraulic balancing accuracy and system stability. The valve adaptive correction parameter set, the system state during the hydraulic balancing control process, the valve adjustment action, and environmental feedback are constructed into a state-action-reward sequence and input into a reinforcement learning algorithm for optimization analysis. Through continuous learning and adjustment, the reinforcement learning algorithm optimizes the valve adjustment strategy under different environmental conditions and generates a valve opening control execution plan.

[0028] In an embodiment of the present application, a complete dynamic database of system parameters is constructed by setting up multiple groups of sensors at the target nodes, providing a comprehensive and accurate data basis for hydraulic balance control; the cooling load fluctuation spectrum analysis and multi-scale decomposition technology are used to accurately grasp the load characteristics, and users are classified according to the load stability index, and a differentiated prediction model is used to improve the load prediction accuracy; a comprehensive system status evaluation index is constructed to realize the quantitative characterization of the degree of hydraulic imbalance, so that the control is transformed from qualitative judgment to quantitative control; a global optimization function is constructed based on the multi-objective optimization idea, and the optimal adjustment strategy that takes into account system efficiency and user comfort is obtained through genetic algorithm solution; an adaptively corrected flow characteristic model is established to overcome the actual characteristic deviation of the valve, and a smoothing algorithm is used to avoid equipment wear caused by frequent adjustment; a reinforcement learning algorithm is introduced to form an adaptive control mechanism with closed-loop feedback to continuously optimize the adjustment strategy; the system reliability is improved through valve health status evaluation, and the hydraulic balance strategy is dynamically adjusted according to different working conditions to achieve flexible coordination between hydraulic balance and cooling demand to meet diversified operation needs.

[0029] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0030] Temperature sensors and pressure sensors are installed on the cold user-side branch pipes of the buried pipe cold storage system to collect the cold water supply and return water temperatures, supply and return water pressures. Electromagnetic flowmeters are installed on each user branch to collect the user branch flow, thus obtaining a complete parameter data set for the cold user side.

[0031] Temperature sensors were installed at the inlet and outlet of the buried pipe-in-tube heat exchanger to collect the inlet and outlet temperatures. Temperature sensors were also installed at the target locations of the pre-cooling evaporative cooling chiller to collect the air temperature before and after the direct evaporative cooling packing and the water temperature in the cold water tank. This provided a data set of heat exchange system operating parameters.

[0032] Pressure and temperature sensors were installed in the cold storage modular unit to collect the refrigerant pressure and temperature before and after the four-way reversing valve, as well as the pressure and temperature at the fluorine pump outlet. Electric regulating valves were installed on the water supply mains of each cold user to obtain a data set of operating parameters for the system's core components.

[0033] The complete parameter data set of the cold user side, the heat exchange system operation parameter data set and the system core component operation parameter data set are normalized to obtain dynamic system parameter data.

[0034] Specifically, temperature sensors and pressure sensors are installed on the branch pipes on the cold user side to collect data on the cold water supply temperature, return water temperature, supply water pressure, and return water pressure. The temperature sensors monitor the supply and return water temperatures in real time, reflecting the changing trends of the cooling load and helping the system understand the cooling load requirements of individual cold users. The pressure sensors collect pressure data on the supply and return pipes, helping to analyze the operational status of the pipe network and identify any hydraulic imbalances. Electromagnetic flowmeters are installed on each cold user branch line to collect real-time flow data. The flow data measured by the flowmeters is used to determine the cooling capacity used by each user, ensuring proper cooling capacity allocation. The data collected by these sensors and flowmeters constitutes the cold user-side parameter dataset. To monitor the operational status of the heat exchange system, temperature sensors are installed at the inlet and outlet of the buried pipe-in-pipe heat exchanger to collect the inlet and outlet temperatures. This temperature data reflects the heat exchanger's operating efficiency and performance. A large temperature difference indicates high heat exchange efficiency, while a small temperature difference indicates suboptimal heat exchange, requiring further optimization. Temperature sensors are installed at key locations in pre-cooling evaporative cooling chillers to collect air temperatures before and after the direct evaporative cooling packing and water temperatures within the cold water tank. By monitoring these temperature data, the cooling effect and heat exchange efficiency are monitored, ensuring the chiller maintains efficient operation while preventing adverse system effects from excessively high or low temperatures. All data collected from the heat exchange system forms a heat exchange system operating parameter dataset. Pressure and temperature sensors are installed within the cold storage modular units to collect refrigerant pressure and temperature data before and after the four-way reversing valve. Analysis of this data provides insights into the refrigerant flow and operating conditions within the cold storage module and assesses the overall operating efficiency of the refrigeration system. The outlet pressure and temperature of the fluorine pump are also collected. The fluorine pump is a critical component of the system, and its performance directly impacts the system's cooling effect and energy efficiency. Electric regulating valves are installed on the water supply mains to each cold user to precisely adjust the chilled water flow. Monitoring these core components provides visibility into the operating status of the cold storage system and timely detection of any faults or performance degradation. Monitoring data from the cold storage modules and core system components forms a core component operating parameter dataset. The complete parameter dataset for the cold user side, the heat exchange system operating parameter dataset, and the system's core component operating parameter dataset are standardized. This transforms data from different sources and formats into a unified data set with unified units and formats, ensuring that all data can be analyzed and calculated on the same platform. This standardized data forms a dynamic system parameter dataset, including temperature, pressure, and flow data for the cold user side, temperature data for the heat exchange system, and pressure and temperature data for the cold storage module and core system components.

[0035] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0036] The supply and return water temperature difference and flow rate of each user branch in the dynamic system parameter data are calculated to obtain the real-time cooling load. A three-dimensional matrix of the real-time cooling load is constructed based on the user identification and time dimension to obtain the cooling load distribution matrix.

[0037] The cooling load time series data in the cooling load distribution matrix is ​​Fourier transformed to obtain the periodic characteristics of load fluctuations. The periodic characteristics of load fluctuations are then decomposed into multiple scales using wavelet transform to obtain basic load types, periodic load types, and random load types.

[0038] According to the basic load type, periodic load type and random load type, the historical load mean and standard deviation of each user branch are calculated to obtain the load stability index. Users are then divided into high stability, medium stability and low stability according to the load stability index.

[0039] For users with high stability, a linear regression model is used; for users with medium stability, a seasonal ARIMA model is used; and for users with low stability, a long short-term memory network model is used. The user demand forecasting model is obtained by taking the current load data at time t and the historical data of the previous n time windows as input.

[0040] Based on the supply and return water temperature difference, flow data and pipeline characteristics of each user, the ideal flow distribution ratio under the hydraulic balance state is calculated to obtain the ideal hydraulic balance flow distribution ratio.

[0041] Specifically, the supply and return water temperature difference and flow rate for each customer branch in the system are calculated in real time to obtain the real-time cooling load for that branch. This reflects the cooling demand of each branch and provides data support for subsequent load distribution and hydraulic balance adjustment. A three-dimensional matrix is ​​constructed from the cooling load data based on the customer ID and time dimension, forming a cooling load distribution matrix. The three-dimensional matrix, composed of customer ID, time dimension, and cooling load value, displays the cooling load variations on different customer branches over different time periods. This matrix effectively analyzes cooling load fluctuations on each customer branch and enables more precise optimization of the system's cooling load distribution. After the cooling load distribution matrix is ​​constructed, the cooling load time series data is Fourier transformed to extract the periodic characteristics of load fluctuations. The Fourier transform converts the cooling load time series from the time domain to the frequency domain, revealing the frequency components of the cooling load fluctuations. This analysis helps identify regular characteristics of load fluctuations, such as the periodicity of cooling load peaks and troughs, and provides a reference for system adjustment. The periodic characteristics obtained through Fourier transforms help the system identify time periods with large and small cooling load fluctuations, thereby adjusting valve openings and flow distribution strategies based on actual load fluctuations. Wavelet transforms are used to perform multi-scale decomposition of the periodic characteristics of load fluctuations. Wavelet transforms are a method that can analyze signals at different scales, extracting information at different levels from load data. By applying a wavelet transform to cooling load fluctuations, the system decomposes the load fluctuations into three components: base load, cyclical load, and random load. Base load represents the system's stable load demand over long-term operation, cyclical load represents load fluctuations due to seasonal variations or other cyclical factors, and random load represents load fluctuations due to unpredictable factors such as equipment failures or sudden demand. Based on the base load, cyclical load, and random load types obtained through multi-scale decomposition, the historical load mean and standard deviation are calculated for each customer branch to generate a load stability index. The load mean reflects the average level of cooling load demand among customers, while the standard deviation reflects the magnitude of cooling load fluctuations. By calculating the load stability index for each user branch, the user's load stability is determined, and users are then divided into three categories: high-stability users, medium-stability users, and low-stability users. High-stability users have smaller load fluctuations and relatively stable load demand, while low-stability users have larger load fluctuations and unstable demand. Different forecasting models are adopted based on the user's load stability to more accurately predict the user's cooling load demand. Different load forecasting models are selected for users with different stability levels. For high-stability users, a linear regression model is used for forecasting. The linear regression model is suitable for users with relatively stable load changes and clear trends. This model effectively captures the linear change trend of load demand over time. For medium-stability users, a seasonal ARIMA (autoregressive integrated moving average) model is used.The seasonal ARIMA model can process time series data with seasonal fluctuations and is suitable for users whose load demand fluctuates regularly within certain periods. For users with low stability, the long short-term memory (LSTM) model is used for forecasting. The LSTM model can capture complex patterns in nonlinear, long time series and is suitable for users with complex and unpredictable load fluctuations. These different forecasting models provide more accurate demand forecasts based on the load stability characteristics of each user. Based on each user's supply and return water temperature difference, flow rate data, and pipeline characteristics, the ideal flow distribution ratio is calculated under hydraulic balance. Hydraulic balance refers to ensuring that the flow rate of each branch within the system can meet the load demand of each user while maintaining stable operation of the pipeline network. By calculating the supply and return water temperature difference, flow rate, and pipeline characteristics of each branch, the ideal flow distribution ratio, that is, the flow share that each user should receive, is determined.

[0042] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0043] The deviation rate between the actual flow of each user branch and the ideal flow calculated according to the ideal flow distribution ratio of hydraulic balance is calculated to obtain the global hydraulic imbalance index;

[0044] Based on the user-side pressure sensor data, the pressure distribution of the cold water supply and return pipe network is analyzed to obtain the pipe network pressure distribution weight coefficient;

[0045] Based on the supply and return water pressure difference and flow rate of each user branch, the flow resistance coefficient of each user branch is calculated and compared with the average flow resistance coefficient of the pipe network to obtain the flow resistance abnormality coefficient;

[0046] Based on the load forecast data output by the user demand forecast model and the actual flow of each branch, combined with the flow resistance abnormality coefficient, a correction calculation is performed to obtain the cooling capacity distribution deviation index;

[0047] Based on the collected data of the inlet and outlet temperatures of the buried pipe-in-pipe heat exchanger, the heat exchange process is corrected in combination with the pipe network pressure distribution weight coefficient to obtain the buried pipe heat exchange efficiency;

[0048] The global hydraulic imbalance index, cooling capacity distribution deviation index and buried pipe heat exchange efficiency are weightedly combined to obtain the comprehensive system status evaluation index.

[0049] Specifically, the actual operating conditions of each user branch are analyzed. During system operation, cold water circulates continuously in the user-side pipe network, and the actual flow of each branch deviates due to changes in cooling demand or differences in pipe network structure. The actual flow of each user branch is compared with the ideal flow calculated according to the hydraulic balance principle, and the hydraulic balance of the current system is measured by the degree of deviation between the two. After the deviation data of all branches are aggregated, an indicator representing the overall hydraulic imbalance of the system is formed. The larger the indicator, the more the water flow distribution of the system deviates from the ideal state and the more serious the hydraulic balance problem; conversely, the system operation is closer to the design conditions. Using the data from the pressure sensors deployed on the user side, the pressure distribution of the water supply pipe and the return pipe is analyzed. Through a series of spatial distribution calculations, a pressure gradient map is constructed to reveal which areas have uneven pressure, abnormal flow rate changes, or sudden changes in local resistance. The pressure distribution analysis results are quantified into a weighting coefficient for subsequent comprehensive assessments. This coefficient represents the pressure sensitivity of different user locations. For example, if a branch is located in an area with a steep pressure gradient, it will experience more severe hydraulic disturbances, requiring priority consideration during regulation. Furthermore, by analyzing the pressure differential between the supply and return water of each branch and the corresponding flow rate, the flow resistance of each branch is calculated. This reflects the degree to which inherent pipe characteristics, such as length, diameter, and local elbow losses, inhibit water flow. These branch resistance values ​​are then compared with the average system-wide level to identify branches with abnormally high resistance. These abnormal branches may exhibit problems such as blockage, scaling, or improper pipe design, becoming key factors contributing to hydraulic imbalance across the entire system. Cooling capacity matching analysis is performed based on user load forecasts and the actual flow rate of each branch. Each branch flow rate is corrected to more closely align with the water volume corresponding to actual load demand. After this correction process, an overall cooling capacity allocation rationality index is calculated, reflecting whether the system suffers from excessive cooling concentration or insufficient cooling supply. At the same time, the heat exchange efficiency is evaluated. By long-term monitoring of the inlet and outlet water temperatures of the buried pipe-in-tube heat exchanger, its heat exchange performance is analyzed to determine whether it is within the design level. Since pressure also affects heat exchange efficiency, especially when the heat exchanger flow fluctuates violently, the heat exchanger performance index is corrected in combination with the pressure distribution weight coefficient obtained previously to obtain a more accurate heat exchange efficiency evaluation result. If the heat exchange efficiency is low, it means that the system has problems such as energy waste or dirt deposition on the heat exchanger surface. A weighted combination of the global hydraulic imbalance index, the cooling capacity distribution deviation index, and the buried pipe heat exchange efficiency is performed. By setting weights for each indicator, they are combined into a comprehensive evaluation indicator that reflects the overall operating status. Based on the value of the comprehensive evaluation indicator, it is determined whether the current operating status is stable, slightly unbalanced, or severely unbalanced, and different levels of valve adjustment strategies are selected accordingly.

[0050] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0051] The adjustment urgency is graded according to the comprehensive system status evaluation indicators to obtain the adjustment priority level assessment result;

[0052] Based on the regulation priority evaluation results, a global optimization objective function including user load demand weight, pressure difference balance and temperature uniformity is constructed;

[0053] Configure the solution parameters of the global optimization objective function to obtain a solution parameter set;

[0054] The load forecast data output by the user demand forecast model is divided into time periods and combined with the load variation amplitude to generate load period characteristic classification results;

[0055] The valve regulation constraint conditions are set based on the load period characteristic classification results to obtain the boundary constraint conditions of the valve regulation;

[0056] The global optimization objective function, the solution parameter set and the boundary constraints of valve regulation are input into the solver for calculation to obtain the time-divided valve regulation strategy.

[0057] Specifically, the hydraulic balance status is quantitatively assessed based on the assessed index value. This index, which incorporates multiple key factors such as the degree of hydraulic imbalance, cooling capacity allocation error, and heat exchange efficiency, fully reflects the urgency of system regulation needs. A series of grading criteria are established based on the index's numerical range. For example, a high index value indicates that the system is significantly deviating from its designed operating state, indicating a regulation emergency; medium or low index values ​​correspond to moderate and mild imbalances, respectively. Based on this grading rule, a classification of regulation urgency is output, and a regulation priority assessment is generated based on this. Based on the regulation priority assessment results, a global optimization objective function is constructed, which includes user load demand weights, pressure differential balance, and temperature uniformity. User load demand weights reflect the importance of cooling loads for different users at the current moment or in future cycles. For example, users experiencing high load periods or significant load fluctuations will receive a higher weight to ensure they are prioritized in the regulation plan. Pressure differential balance, a key dimension in evaluating the rationality of hydraulic transmission paths, aims to achieve uniformity in the supply and return pressure differentials between branches through optimized regulation, thereby improving system stability and responsiveness. The goal of temperature uniformity is to ensure minimal variations in water supply temperature across user terminals in the system, thereby avoiding uneven heating and cooling, improving user comfort, and enhancing energy efficiency. Taking all of these factors into consideration, a global optimization objective function is constructed, encompassing heat distribution rationality, hydraulic path balance, and thermal response coordination. The objective function's solution parameters are configured based on the control requirements at different priorities. This parameter configuration process includes adjusting the adjustment rate coefficient, optimizing the required accuracy, setting the iteration range, and handling constraints. This provides the solver with a set of technical parameters that adapt to the current system state. For example, when the adjustment priority is high, the system tends to accelerate convergence and adopt a larger adjustment range to quickly reduce system deviation. In contrast, when the priority is low, the system prioritizes stable operation and minimizing energy consumption, and the parameter settings favor fine-grained adjustment and high-precision calculations. These solution parameters are combined into a parameter set and input into the solver along with the objective function structure. Furthermore, to enhance the dynamic adaptability of the control strategy, the output of the user demand forecasting model is combined to identify future trends in cooling load. The forecast data is divided into multiple time periods based on the time dimension, and the magnitude and frequency of load fluctuations within each time period are calculated. Based on this analysis, the different time periods are divided into high-load fluctuation periods, medium-load fluctuation periods, and stable load periods, forming a load period characteristic classification result. This classification result serves as the time basis for determining the system's regulation intensity, frequency, and valve response priority. Periods with large load fluctuations are given greater regulation space, allowing valves to have wider adjustment ranges and shorter adjustment intervals. Conversely, periods with more stable loads prioritize maintaining system stability, appropriately compressing the regulation range to reduce system disturbances.After obtaining the time-period characteristic classification results, boundary constraints for valve regulation are set accordingly. These constraints include the maximum allowable valve opening variation for each adjustment, the minimum time interval between adjustments, the minimum water flow rate for each branch, and flow safety thresholds agreed upon based on the network characteristics. These boundary constraints protect field equipment and prevent system overshoot, ensuring the continuity and stability of the regulation process. For example, during low-load stable periods, even if the objective function recommends adjusting certain valves to improve energy efficiency, the system determines whether adjustments are necessary based on boundary conditions, thus avoiding mechanical wear and increased energy consumption caused by frequent fine-tuning. The constructed global optimization objective function, the set of solution parameters, and the corresponding valve regulation boundary constraints for the load characteristic classification results are input into the solver for unified calculation. The solver uses an iterative algorithm (such as a genetic algorithm, particle swarm optimization, or adaptive variable-step gradient descent) to find a solution that minimizes the optimization objective function while satisfying the constraints, generating an optimal set of time-period valve regulation strategies. This strategy specifically defines the target opening, regulation rate, and execution timing of each regulating valve within each time period, and continuously adjusts the optimization direction based on a real-time feedback mechanism to adapt to nonlinear interference and load fluctuations in actual operation, thereby ensuring that the entire system can maintain the optimal hydraulic distribution state under various operating conditions.

[0058] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0059] Classify the flow characteristics of each regulating valve according to the valve type and obtain the valve flow characteristic model;

[0060] Perform regression analysis on the valve historical response data, calculate the deviation between the actual flow response and the theoretical flow response of each valve, and obtain the valve characteristic correction coefficient;

[0061] According to the target flow distribution and current pressure difference data determined by the time-division valve regulation strategy, reverse calculation is performed in combination with the valve flow characteristic model to obtain the target opening of each regulating valve;

[0062] Perform threshold judgment on the difference between the target opening and the current opening to obtain the target valve opening value;

[0063] Generate valve control execution instructions based on the valve target opening value and the adjustment time window in the time-division valve adjustment strategy.

[0064] Specifically, the flow characteristics of each control valve are classified and a valve flow characteristic model is established. Different valve types have different flow characteristics, such as equal percentage valves, linear valves, and quick-opening valves. These valve types exhibit different response characteristics in flow control and are classified based on their structure and operating principle. The valve flow characteristic model is a mathematical model that describes the relationship between valve opening and flow rate, helping the system understand the flow range that the valve can adjust at a given opening. Based on this classification, a corresponding flow control model is designed for each valve type. Regression analysis is performed on historical valve response data. Using actual operating data, the deviation between the actual flow response of each valve and the theoretical flow response is evaluated. Valves are affected by various factors in actual operation, such as valve aging, pipeline pressure fluctuations, and fluid characteristics. These factors can cause errors between the valve flow response and the theoretical model. Regression analysis quantifies this deviation and calculates a characteristic correction factor for each valve. The correction factor is a key parameter used to adjust the theoretical flow model based on actual conditions, ensuring the accuracy of flow control and compensating for the discrepancy between the theoretical model and actual operation. The introduction of correction coefficients allows each valve's control strategy to better align with actual operating conditions, thereby improving the regulation and operational efficiency of the entire system. Based on the target flow distribution determined by the time-based valve regulation strategy and current differential pressure data, a reverse calculation is performed in conjunction with the valve flow characteristic model to determine the target valve opening. The target flow distribution represents the required flow rate for each user branch, determined by the system based on load forecasts and hydraulic balance analysis, while the current pipeline differential pressure reflects the system's hydraulic status at different time periods. Combining these two data, a reverse calculation is performed based on the valve flow characteristic model, inferring the required valve opening from the target flow rate and pipeline differential pressure. This reverse calculation is based on the system's hydraulic balance requirements, ensuring that each valve's opening adjustment meets cooling load requirements and hydraulic balance conditions, thereby achieving global optimal control. A threshold is applied to the difference between the target opening and the current opening. This threshold ensures that when the difference between the target opening and the current opening is small, the valve is either not adjusted or only fine-tuned. This threshold determination prevents unnecessary fine-tuning and over-adjustment. If the difference between the target opening and the current opening exceeds a set threshold, the system issues a control instruction and makes appropriate adjustments to gradually adjust the valve opening to the target value. Based on the calculated target valve opening value and the adjustment time window in the time-segmented valve adjustment strategy, valve control execution instructions are generated. These instructions include the target valve opening value, the specific adjustment time, and the adjustment rate. The setting of the adjustment time window takes into account the user's load demand changes and the system's hydraulic balance, ensuring that the valve adjustment process is not too hasty or delayed, and that the necessary adjustments can be completed within the appropriate time period.By controlling the execution instructions, the field controller accurately drives the electric actuator to adjust the valve opening, ensuring that the cooling load demand of each user is met and the hydraulic balance of the entire system is optimized.

[0065] In a specific embodiment, executing the valve opening control method for user-side hydraulic balance further includes the following steps:

[0066] Prioritize valve control execution instructions based on the urgency of hydraulic balance, and transmit them to the field controller via Modbus or BACnet protocol to drive the electric actuator to adjust the valve opening;

[0067] Compare and analyze the system status after each valve control execution instruction is completed, calculate the hydraulic imbalance index, cooling capacity distribution deviation index and the change rate of the comprehensive system status evaluation index before and after the adjustment, and obtain the adjustment effect index;

[0068] Collect data on the regulating response characteristics of each valve, record the time delay and response gain from issuing a command to flow stabilization, and construct a first-order lag model;

[0069] Calculate the valve adaptive correction parameter set according to the regulation effect index and the first-order hysteresis model;

[0070] The valve adaptive correction parameter set and the system state, valve adjustment action and environmental feedback in the hydraulic balance control process are constructed into a state-action-reward sequence, and input into the reinforcement learning algorithm for optimization analysis to generate a valve opening control execution plan.

[0071] Specifically, valve control execution instructions are prioritized based on the urgency of hydraulic balance. The urgency of hydraulic balance is determined by the global hydraulic imbalance index, the cooling capacity distribution deviation index, and other key performance indicators. When these indicators are high, the system has a significant hydraulic imbalance or cooling capacity distribution deviation, and the system prioritizes adjusting the relevant valves to restore balance. Therefore, the current hydraulic balance status of the zones controlled by each valve is first assessed, and the adjustment priority of each valve is determined based on the assessment results. For example, customer branches located in hydraulically unbalanced areas or with cooling capacity distribution deviations are assigned a higher priority to ensure that these zones are prioritized for regulation. Control instructions for valves with higher adjustment priorities are processed first during regulation execution, ensuring a rapid system response and restoration of hydraulic balance. After determining the adjustment priority, the valve control execution instructions are transmitted to the field controller via Modbus or BACnet protocols. Modbus and BACnet protocols enable efficient and reliable data transmission between different devices. After receiving the instructions, the field controller controls the electric actuator to precisely adjust the valve opening. This process is the core of automated control, ensuring the system can respond in real time to control commands and adjust valve openings based on varying hydraulic balance requirements, thereby affecting flow and pressure distribution and ensuring proper cooling capacity distribution. After the control execution is complete, the system status of each valve after the control execution command is completed is compared and analyzed. The hydraulic imbalance index, cooling capacity distribution deviation index, and comprehensive system status assessment indicators are calculated before and after the control is executed. These change rates reflect the actual impact of valve control on system operation. If the change rates of the hydraulic imbalance index and cooling capacity distribution deviation index are large, it indicates that the valve control has effectively improved the system's hydraulic balance and cooling capacity distribution. If the change rates are small, the control effect is not significant, and further adjustment or optimization of the control strategy is required. The control effect index is a comprehensive indicator that reflects the actual effectiveness of valve control and guides the optimization of subsequent control strategies. Data is collected on the control response characteristics of each valve. The time delay and response gain from issuing the control command to flow stabilization are recorded. Time delay and response gain are important indicators of valve response speed and control accuracy, revealing any hysteresis or uneven response during the valve control process. By collecting this data, a first-order hysteresis model of the valve is constructed. This model describes the dynamic characteristics of the valve during regulation, specifically the changes in time delay and response gain. Establishing a first-order hysteresis model helps the system understand the valve's regulation efficiency and its responsiveness to flow changes, providing a parameter basis for future regulation strategies. Based on the regulation effectiveness index and the first-order hysteresis model, the valve's adaptive correction parameter set is calculated. This adaptive correction parameter set aims to compensate for errors that occur during regulation, particularly in situations where the valve exhibits slow response or significant hysteresis.The adaptive correction parameter set is adjusted based on historical data and real-time feedback to optimize the accuracy and responsiveness of the valve control strategy. By continuously adjusting and optimizing these correction parameters, the valve's response in future adjustments is more precise and adaptable to environmental changes and load fluctuations. The valve adaptive correction parameter set, the current system state, valve control actions, and environmental feedback are constructed into a state-action-reward sequence and input into a reinforcement learning algorithm for optimization analysis. Reinforcement learning is a machine learning method that optimizes decision-making processes through interaction with the environment. In this process, the current system state is considered the "state," each valve control action is considered the "action," and the resulting effect is considered the "reward." Through training with the reinforcement learning algorithm, the system continuously optimizes the valve opening control strategy, ensuring that each control action delivers optimal system performance. Through multiple iterations and feedback adjustments, the reinforcement learning algorithm automatically selects the most appropriate valve control strategy under different hydraulic imbalance conditions, thereby achieving optimal system control.

[0072] Among them, the valve adaptive correction parameter set and the system state, valve adjustment action and environmental feedback in the hydraulic balance control process are constructed into a state-action-reward sequence, and input into the reinforcement learning algorithm for optimization analysis to generate a valve opening control execution plan, including: feature extraction and normalization of key system parameters such as hydraulic imbalance index, cooling capacity distribution deviation index, buried pipe heat exchange efficiency, user load demand and valve response characteristics to obtain a state vector representing the current state of the system; discretization of the adjustable opening range of all valves, dividing the adjustment amplitude of each valve into three basic operations: increase, decrease and maintain, and combining the adjustment step size to construct a multi-dimensional action space to obtain a valve adjustment action set; based on the three goals of hydraulic balance, user comfort and system energy efficiency A multi-objective reward function is designed to perform positive and negative incentive evaluation on the system state changes after valve adjustment, and obtain a reward signal that characterizes the quality of the adjustment effect; the state vector, action set and reward signal in the historical adjustment process are organized in a time series to form a state-action-reward triple sequence, and a training data set for reinforcement learning is obtained; the training data set is input into the deep Q network for model training, and the experience replay mechanism and target network separation technology are used to reduce sample correlation. By gradually reducing the exploration rate, the exploration and utilization are balanced to obtain the optimized valve adjustment strategy network; based on the optimized valve adjustment strategy network, the current system state is analyzed and evaluated, and the optimal valve adjustment action sequence is generated and converted into a standard control instruction format to obtain a valve opening control execution scheme with adaptive learning ability.

[0073] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0074] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a valve opening control device for user-side hydraulic balancing (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0075] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A valve opening control method for user-side hydraulic balance, characterized in that: include: Parameters of the underground pipe cold storage system are collected to obtain dynamic system parameter data. Temperature sensors and pressure sensors are installed on the cold user-side branch pipes of the underground pipe cold storage system to collect the cold water supply and return water temperatures, supply and return water pressures. Electromagnetic flowmeters are installed on each user branch to collect the user branch flow rate, thus obtaining a complete parameter data set for the cold user side. Based on the dynamic system parameter data, cooling load fluctuation spectrum analysis and multi-scale decomposition are performed to obtain a user demand prediction model and an ideal flow distribution ratio for hydraulic balance, including: The supply and return water temperature difference and flow rate of each user branch in the dynamic system parameter data are calculated to obtain a real-time cooling load, and a three-dimensional matrix is ​​constructed for the real-time cooling load according to the user identifier and time dimension to obtain a cooling load distribution matrix; the cooling load time series data in the cooling load distribution matrix are Fourier transformed to obtain the periodic characteristics of load fluctuations, and the periodic characteristics of load fluctuations are multi-scale decomposed by wavelet transform to obtain basic load type, periodic load type and random load type; based on the basic load type, the periodic load type and the random load type, the historical load mean and standard deviation of each user branch are calculated to obtain a load stability index, and users are classified into high stability, medium stability and low stability according to the load stability index; A linear regression model is used for high-stability users, a seasonal ARIMA model is used for medium-stability users, and a long-short-term memory network model is used for low-stability users. The user demand forecasting model is derived using the current load data at time t and the historical data of the previous n time windows as input. Based on each user's supply and return water temperature difference, flow data, and pipeline characteristics, the ideal flow distribution ratio under hydraulic balance is calculated to obtain the ideal hydraulic balance flow distribution ratio. Generate a comprehensive system status evaluation index based on the user demand prediction model and the ideal flow distribution ratio of hydraulic balance; Constructing a global optimization objective function based on the comprehensive system state evaluation index and solving it to obtain a time-divided valve adjustment strategy; A valve flow characteristic model is constructed according to the time-division valve regulation strategy and valve historical response data, and the target opening of each regulating valve is calculated, and a valve control execution instruction is output.

2. The valve opening control method for user-end hydraulic balance according to claim 1, characterized in that: The parameter collection of the underground pipe cold storage system to obtain dynamic system parameter data includes: Temperature sensors and pressure sensors are installed on the cold user-side branch pipes of the buried pipe cold storage system to collect the cold water supply and return water temperatures, supply and return water pressures. Electromagnetic flowmeters are installed on each user branch to collect the user branch flow, thus obtaining a complete parameter data set for the cold user side. Temperature sensors were installed at the inlet and outlet of the buried pipe-in-tube heat exchanger to collect the inlet and outlet temperatures. Temperature sensors were also installed at the target locations of the pre-cooling evaporative cooling chiller to collect the air temperature before and after the direct evaporative cooling packing and the water temperature in the cold water tank. This provided a data set of heat exchange system operating parameters. Pressure and temperature sensors were installed in the cold storage modular unit to collect the refrigerant pressure and temperature before and after the four-way reversing valve, as well as the pressure and temperature at the fluorine pump outlet. Electric regulating valves were installed on the water supply mains of each cold user to obtain a data set of operating parameters for the system's core components. The complete parameter data set of the cold user side, the operating parameter data set of the heat exchange system and the operating parameter data set of the system core components are subjected to data standardization processing to obtain dynamic system parameter data.

3. The valve opening control method for user-end hydraulic balance according to claim 2, characterized in that: Generating a comprehensive system status evaluation index based on the user demand prediction model and the ideal hydraulic balance flow distribution ratio includes: Calculating the deviation rate between the actual flow of each user branch and the ideal flow calculated according to the ideal hydraulic balance flow distribution ratio to obtain a global hydraulic imbalance index; Based on the user-side pressure sensor data, the pressure distribution of the cold water supply and return pipe network is analyzed to obtain the pipe network pressure distribution weight coefficient; Based on the supply and return water pressure difference and flow rate of each user branch, the flow resistance coefficient of each user branch is calculated and compared with the average flow resistance coefficient of the pipe network to obtain the flow resistance abnormality coefficient; According to the load forecast data output by the user demand forecast model and the actual flow of each branch, combined with the flow resistance abnormality coefficient, a correction calculation is performed to obtain a cooling capacity distribution deviation index; Based on the collected data of the inlet temperature and outlet temperature of the buried pipe-in-pipe heat exchanger, the heat exchange process is corrected in combination with the pipe network pressure distribution weight coefficient to obtain the buried pipe heat exchange efficiency; A comprehensive system status evaluation index is obtained by performing a weighted combination of the global hydraulic imbalance index, the cooling capacity distribution deviation index, and the buried pipe heat exchange efficiency.

4. The valve opening control method for user-end hydraulic balance according to claim 1, characterized in that: The global optimization objective function is constructed based on the comprehensive system state evaluation index, and the time-division valve adjustment strategy is obtained by solving the problem, including: Classifying the adjustment urgency according to the comprehensive system status evaluation index to obtain an adjustment priority evaluation result; Constructing a global optimization objective function including user load demand weight, pressure difference balance and temperature uniformity based on the adjustment priority evaluation result; Performing solution parameter configuration on the global optimization objective function to obtain a solution parameter set; Dividing the load forecast data output by the user demand forecast model into time periods and generating load period characteristic classification results in combination with the load variation amplitude; Setting valve regulation constraint conditions based on the load period characteristic classification result to obtain boundary constraint conditions for valve regulation; The global optimization objective function, the solution parameter set and the boundary constraint conditions of the valve regulation are input into the solver for calculation to obtain the time-division valve regulation strategy.

5. The valve opening control method for user-end hydraulic balance according to claim 1, characterized in that: The valve flow characteristic model is constructed according to the time-division valve adjustment strategy and the valve historical response data, and the target opening of each regulating valve is calculated, and the valve control execution instruction is output, including: Classify the flow characteristics of each regulating valve according to the valve type and obtain the valve flow characteristic model; Perform regression analysis on the valve historical response data, calculate the deviation between the actual flow response and the theoretical flow response of each valve, and obtain the valve characteristic correction coefficient; According to the target flow distribution and current pressure difference data determined by the time-division valve regulation strategy, reverse calculation is performed in combination with the valve flow characteristic model to obtain the target opening of each regulating valve; Performing a threshold judgment on the difference between the target opening and the current opening to obtain a target valve opening value; A valve control execution instruction is generated according to the valve target opening value and the adjustment time window in the time-division valve adjustment strategy.

6. The valve opening control method for user-end hydraulic balance according to claim 1, characterized in that: The valve opening control method for user-end hydraulic balance also includes: Prioritizing the valve control execution instructions according to the urgency of the hydraulic balance, and transmitting the valve control execution instructions to the field controller via the Modbus protocol or the BACnet protocol to drive the electric actuator to adjust the valve opening; Compare and analyze the system status after each valve control execution instruction is completed, calculate the hydraulic imbalance index, cooling capacity distribution deviation index and the change rate of the comprehensive system status evaluation index before and after the adjustment, and obtain the adjustment effect index; Collect data on the regulating response characteristics of each valve, record the time delay and response gain from issuing a command to flow stabilization, and construct a first-order lag model; Calculating a valve adaptive correction parameter set according to the regulation effect index and the first-order hysteresis model; The valve adaptive correction parameter set and the system state, valve adjustment action and environmental feedback in the hydraulic balance control process are constructed into a state-action-reward sequence, and input into the reinforcement learning algorithm for optimization analysis to generate a valve opening control execution plan.

Images