Energy-saving automatic control system for air conditioning variable flow water system and control method thereof

Abstract

The application relates to the technical field of energy saving, and discloses an energy-saving automatic control system of an air conditioner variable flow water system and a control method thereof, which aims to solve the problem of poor accuracy and reliability of the energy-saving automatic control mode of an existing air conditioner system, and mainly comprises a data acquisition module, a dynamic modeling module, an energy efficiency identification module, an energy distribution module, a maintenance optimization module, a control strategy optimization module and a real-time monitoring and adjustment module, the data acquisition module is used for generating post-purification data, the dynamic modeling module is used for generating an anti-interference optimization model, the energy efficiency identification module is used for generating energy efficiency identification records, the energy distribution module is used for generating an energy optimization scheme, the maintenance optimization module is used for generating a system maintenance and efficiency proposal, the control strategy optimization module is used for generating a control strategy improvement scheme, and the real-time monitoring and adjustment module is used for generating a performance tuning instruction set. The application improves the accuracy and reliability of the energy-saving automatic control of the air conditioner variable flow water system.

Description

Technical Field

[0001] This application relates to the field of energy-saving technology, specifically to an energy-saving automatic control system and control method for an air conditioning variable flow water system. Background Technology

[0002] Energy conservation technology is a field focused on developing and applying various methods, equipment, and systems to reduce energy consumption and improve energy efficiency. In building environmental control systems, energy conservation goals are achieved by optimizing system operation modes, improving energy efficiency, and reducing unnecessary energy waste. This is crucial for reducing environmental pollution and promoting sustainable development.

[0003] The variable flow water system for air conditioning is an automated system that uses control strategies to regulate the water flow in an air conditioning system. Its main purpose is to adapt to different load demands by controlling the water flow, thereby reducing energy consumption and improving the overall system's energy efficiency. It aims to achieve optimized energy efficiency, reduce operating costs, and minimize environmental impact. By adjusting the water flow, the system can provide appropriate cooling or heating according to actual needs, avoiding excessive energy consumption and thus achieving energy savings.

[0004] Traditional systems lack sufficient flexibility and accuracy in handling external disturbances, data noise, and operational anomalies. The lack of effective data processing methods results in poor data quality, impacting the accuracy of decision-making. Traditional systems fail to adequately consider nonlinear characteristics and dynamic changes, making them poorly adaptable to external variations and prone to performance instability. In energy efficiency management, the lack of accurate state identification and energy demand forecasting prevents optimal energy allocation, leading to energy waste. Maintenance strategies rely heavily on experience and lack data support, affecting the long-term operational efficiency and reliability of the system. Summary of the Invention

[0005] This application aims to address the problem of poor accuracy and reliability in existing energy-saving automatic control methods for air-conditioning variable flow water systems, and proposes an energy-saving automatic control system and its control method for air-conditioning variable flow water systems.

[0006] The technical solution adopted by this application to solve the above-mentioned technical problems is:

[0007] In a first aspect, this application provides an energy-saving automatic control system for a variable flow water system for air conditioning, the system comprising a data acquisition module, a dynamic modeling module, an energy efficiency identification module, an energy distribution module, a maintenance optimization module, a control strategy optimization module, and a real-time monitoring and adjustment module;

[0008] The data acquisition module is used to collect environmental and system operation data, use an outlier detection algorithm to identify and remove outliers in the operation data, and perform data smoothing processing to generate cleaned data.

[0009] The dynamic modeling module is used to construct a mathematical model based on the purified data using a nonlinear dynamic modeling method, and to combine a sliding mode control strategy to optimize the model's resistance to external disturbances, thereby generating an anti-disturbance optimization model.

[0010] The energy efficiency identification module is used to analyze system performance data based on the anti-disturbance optimization model and the support vector machine algorithm, identify operating efficiency modes and potential abnormal states, perform mode classification and state identification, and generate energy efficiency identification records.

[0011] The energy allocation module is used to analyze energy demand based on energy efficiency identification records, apply linear programming, adjust allocation strategies, match real-time demand, optimize energy waste, and generate energy optimization schemes.

[0012] The maintenance optimization module is used to group system operation data based on energy optimization schemes using cluster analysis, extract key performance indicators through principal component analysis, formulate maintenance plans and energy efficiency improvement measures, and generate system maintenance and efficiency proposals.

[0013] The control strategy optimization module is used to learn control strategies based on system maintenance and performance proposals, using a deep Q network and a reward mechanism, to adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement schemes.

[0014] The real-time monitoring and adjustment module is used to improve the control strategy, utilize model predictive control technology, and adjust system parameters based on real-time monitoring data and changes in the external environment to optimize system performance and energy efficiency, and generate a performance tuning instruction set.

[0015] Furthermore, the purified data includes indoor temperature sequence, humidity level time sequence, and air conditioning system flow data sequence;

[0016] The disturbance rejection optimization model includes the model's stability parameters, response coefficient to external temperature changes, and sensitivity index to flow regulation.

[0017] The energy optimization scheme includes an energy demand prediction map, a priority allocation map, and an optimal energy consumption adjustment strategy.

[0018] The system maintenance and performance proposal includes a predetermined maintenance schedule, a monitoring plan for key performance indicators, and performance improvement measures for the target issues;

[0019] The improved control strategy includes temperature control adjustment indicators, humidity control adjustment strategies, and an optimized parameter set for flow rate adjustment.

[0020] The performance tuning instruction set includes temperature adjustment commands, humidity level adjustment commands, and air conditioning system flow control commands.

[0021] Furthermore, the data acquisition module includes a sensor submodule, an outlier processing submodule, and a data smoothing submodule;

[0022] The sensor submodule is used to collect environmental and system operation data, including temperature, humidity and flow data, and to determine the real-time performance and accuracy of the operation data to generate a raw dataset.

[0023] The outlier processing submodule is used to identify and remove statistical outliers based on the original dataset using the Z-score method, verify the quality and reliability of the dataset, and generate an outlier-free dataset.

[0024] The data smoothing submodule is used to reduce random fluctuations in time series data by using a moving average method based on outlier removal datasets, thereby optimizing the smoothness and usability of the data and generating cleaned data.

[0025] Furthermore, the dynamic modeling module includes a nonlinear modeling submodule, a feedback linearization submodule, and a sliding mode design submodule;

[0026] The nonlinear modeling submodule is used to construct a dynamic model of the air conditioning system based on the purified data using a radial basis function network, and to capture the nonlinear characteristics of the system to generate a preliminary mathematical model.

[0027] The feedback linearization submodule is used to simplify the nonlinear model into a linear model by applying feedback linearization based on the preliminary mathematical model and through mathematical transformation, and to adjust the establishment and implementation of the control strategy to generate a linearized model.

[0028] The sliding mode design submodule is used to define the sliding surface and design the control law based on the linearized model and the sliding mode control strategy, optimize the robustness of the model to external disturbances and parameter changes, and generate a disturbance-resistant optimization model.

[0029] Furthermore, the energy efficiency identification module includes a feature analysis submodule, a data classification submodule, and a state identification submodule;

[0030] The feature analysis submodule is used to extract the periodic and trend features of temperature, humidity and energy consumption based on the anti-disturbance optimization model, using time series analysis and frequency domain analysis methods, calculate the statistical mean and variance index, prepare input data for machine learning, and generate feature vector sets.

[0031] The data classification submodule is used to classify the system's operating state based on the feature vector set, using the support vector machine algorithm, mapping the features to a high-dimensional space through the kernel trick, constructing a classification hyperplane using the maximum margin principle, and generating a classification and recognition model.

[0032] The state recognition submodule is used to perform state recognition on real-time data streams based on classification and recognition models, including using models to make real-time predictions, identifying the current operating state of the system, and generating energy efficiency recognition records.

[0033] Furthermore, the energy allocation module includes a demand analysis submodule, a linear programming submodule, and a strategy generation submodule;

[0034] The demand analysis submodule is used to analyze energy demand curves under different conditions based on energy efficiency identification records, using mathematical modeling methods, and using long short-term memory networks to assess future energy demand and generate energy demand forecast data.

[0035] The linear programming submodule is used to define an optimization model based on energy demand forecast data and using a linear programming algorithm. This includes setting constraints on energy supply and demand, solving for the optimal energy allocation measures under the operating mode, and generating an optimized allocation scheme.

[0036] The strategy generation submodule is used to establish an energy allocation execution plan based on the optimized allocation scheme, including adjusting the operating parameters of the air conditioning system, setting energy usage priorities and allocation time, verifying the feasibility and execution efficiency of the strategy, and generating an energy optimization scheme.

[0037] Furthermore, the maintenance optimization module includes a cluster analysis submodule, a performance index extraction submodule, and a maintenance strategy design submodule;

[0038] The clustering analysis submodule is used to group system operation data based on energy optimization schemes and K-means clustering algorithm, identify different operating states, provide a classification basis for performance analysis, and generate data classification groups.

[0039] The performance index extraction submodule is used to extract key performance indicators from data classification groups using principal component analysis to reflect system operating efficiency and potential problems, and generate a performance index summary.

[0040] The maintenance strategy design submodule is used to design maintenance plans and energy efficiency improvement measures based on performance index summaries, including preventive maintenance, parameter optimization and equipment upgrades, to optimize system reliability and efficiency, and generate system maintenance and efficiency proposals.

[0041] Furthermore, the control strategy optimization module includes a reinforcement learning training submodule, a strategy evaluation submodule, and a strategy implementation submodule;

[0042] The reinforcement learning training submodule is used to learn the optimal control strategy based on system maintenance and performance proposals, using a deep Q-network and a reward feedback mechanism, to optimize energy consumption and maintain environmental comfort, and generate a strategy learning model.

[0043] The strategy evaluation submodule is used to evaluate the effectiveness of a strategy based on a strategy learning model. By simulating system operation under different environmental conditions, it verifies the effectiveness and adaptability of the control strategy and generates strategy evaluation records.

[0044] The strategy implementation submodule is used to plan the application steps of the control strategy based on the strategy evaluation records and using the particle swarm optimization algorithm, including parameter adjustment, personnel training and configuration updates, and to generate control strategy improvement schemes.

[0045] Furthermore, the real-time monitoring and adjustment module includes a real-time data monitoring submodule, an adjustment decision submodule, and a parameter adjustment submodule;

[0046] The real-time data monitoring submodule is used to monitor the collected operating parameters of the air conditioning system in real time, including key indicators such as temperature, humidity and energy consumption, based on the control strategy improvement scheme and using real-time data monitoring technology, and generate monitoring status information.

[0047] The adjustment decision submodule is used to analyze the deviation between real-time data and expected performance based on monitoring status information and DSS technology, determine whether system parameters need to be adjusted, and generate adjustment decision schemes in response to changes in the external environment or deviations in internal performance.

[0048] The parameter adjustment submodule is used to implement parameter adjustment based on the adjustment decision scheme and using model predictive control technology. The parameter adjustment includes adjusting temperature, humidity and wind speed, and verifying the matching of system operating parameters with performance and energy efficiency requirements, and generating a performance tuning instruction set.

[0049] Secondly, this application provides a control method for an energy-saving automatic control system of an air conditioning variable flow water system, applied to the energy-saving automatic control system of the air conditioning variable flow water system as described in the first aspect, the method comprising the following steps:

[0050] Collect environmental and system operation data, use outlier detection algorithms to identify and remove outliers in the operation data, and perform data smoothing processing to generate cleaned data;

[0051] Based on the purified data, a mathematical model is constructed using a nonlinear dynamic modeling method, and the resistance of the model to external disturbances is optimized by combining a sliding mode control strategy to generate an anti-disturbance optimization model.

[0052] Based on the disturbance rejection optimization model, the support vector machine algorithm is used to analyze system performance data, identify operating efficiency modes and potential abnormal states, perform mode classification and state identification, and generate energy efficiency identification records.

[0053] Based on energy efficiency identification records, linear programming is applied to analyze energy demand, adjust allocation strategies, match real-time demand, and optimize energy waste to generate energy optimization solutions.

[0054] Based on the energy optimization scheme, the system operation data is grouped using cluster analysis, and key performance indicators are extracted through principal component analysis. Maintenance plans and energy efficiency improvement measures are then formulated to generate system maintenance and efficiency proposals.

[0055] Based on system maintenance and performance proposals, a deep Q network is used to learn control strategies through a reward mechanism, adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement plans.

[0056] Based on the control strategy improvement scheme, model predictive control technology is used to adjust system parameters according to real-time monitoring data and changes in the external environment, thereby optimizing system performance and energy efficiency and generating a performance tuning instruction set.

[0057] The beneficial effects of this application are as follows: The energy-saving automatic control system and control method for the variable flow water system in air conditioning provided in this application adopt outlier detection algorithms and data smoothing methods, which can effectively reduce data noise and eliminate outliers, ensuring data quality and providing an accurate basis for subsequent analysis. The use of nonlinear dynamic modeling and sliding mode control strategies optimizes resistance to external disturbances and enhances system stability. The application of support vector machine algorithms improves the accuracy of identifying operating efficiency patterns and potential abnormal states, making energy efficiency management more accurate. The combined use of linear programming and cluster analysis not only optimizes energy allocation but also develops targeted maintenance plans and energy efficiency improvement measures, enhancing the system's energy efficiency and reliability. The control strategy learning introduced by deep Q-networks and the implementation of model predictive control technology further optimize system performance and energy efficiency, achieving effective energy consumption control and maintaining indoor environmental quality. Attached Figure Description

[0058] Figure 1 A schematic diagram of the structure of an energy-saving automatic control system for an air conditioning variable flow water system provided in this application embodiment;

[0059] Figure 2 A schematic diagram of another energy-saving automatic control system for a variable flow water system for air conditioning provided in this application embodiment;

[0060] Figure 3 A flowchart illustrating the implementation of a data acquisition module provided in an embodiment of this application;

[0061] Figure 4 A flowchart illustrating the implementation of a dynamic modeling module provided in an embodiment of this application;

[0062] Figure 5 A flowchart illustrating the implementation of an energy efficiency identification module provided in this application embodiment;

[0063] Figure 6 A flowchart illustrating the implementation of an energy distribution module provided in an embodiment of this application;

[0064] Figure 7 A flowchart illustrating the implementation of a maintenance and optimization module provided in an embodiment of this application;

[0065] Figure 8 A flowchart illustrating the implementation of a control strategy optimization module provided in an embodiment of this application;

[0066] Figure 9 A flowchart illustrating the implementation of a real-time monitoring and adjustment module provided in this application embodiment;

[0067] Figure 10 This is a flowchart illustrating a control method for an energy-saving automatic control system of an air conditioning variable flow water system, as provided in an embodiment of this application. Detailed Implementation

[0068] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0069] The technical solution of this application embodiment is applicable to the application scenario of energy saving in air conditioning system, that is, by controlling the water flow of the air conditioning system to adapt to different load demands, thereby reducing energy consumption and improving the energy efficiency of the entire system.

[0070] Please see Figure 1 The energy-saving automatic control system for air conditioning variable flow water system provided in this application includes a data acquisition module, a dynamic modeling module, an energy efficiency identification module, an energy distribution module, a maintenance optimization module, a control strategy optimization module, and a real-time monitoring and adjustment module.

[0071] The data acquisition module is used to collect environmental and system operation data, use an outlier detection algorithm to identify and remove outliers in the operation data, and perform data smoothing processing to generate cleaned data.

[0072] The dynamic modeling module is used to construct a mathematical model based on the purified data using a nonlinear dynamic modeling method, and to combine a sliding mode control strategy to optimize the model's resistance to external disturbances, thereby generating an anti-disturbance optimization model.

[0073] The energy efficiency identification module is used to analyze system performance data based on the anti-disturbance optimization model and the support vector machine algorithm, identify operating efficiency modes and potential abnormal states, perform mode classification and state identification, and generate energy efficiency identification records.

[0074] The energy allocation module is used to analyze energy demand based on energy efficiency identification records, apply linear programming, adjust allocation strategies, match real-time demand, optimize energy waste, and generate energy optimization schemes.

[0075] The maintenance optimization module is used to group system operation data based on energy optimization schemes using cluster analysis, extract key performance indicators through principal component analysis, formulate maintenance plans and energy efficiency improvement measures, and generate system maintenance and efficiency proposals.

[0076] The control strategy optimization module is used to learn control strategies based on system maintenance and performance proposals, using a deep Q network and a reward mechanism, to adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement schemes.

[0077] The real-time monitoring and adjustment module is used to improve the control strategy, utilize model predictive control technology, and adjust system parameters based on real-time monitoring data and changes in the external environment to optimize system performance and energy efficiency, and generate a performance tuning instruction set.

[0078] In this embodiment, the purified data includes indoor temperature series, humidity level time series, and air conditioning system flow data series; the disturbance rejection optimization model includes model stability parameters, response coefficients to external temperature changes, and sensitivity indicators to flow regulation; the energy optimization scheme includes energy demand prediction charts, priority allocation charts, and optimal energy consumption regulation strategies; the system maintenance and performance proposal includes a predetermined maintenance schedule, a monitoring plan for key performance indicators, and performance improvement measures for the target problem; the control strategy improvement scheme includes temperature control regulation indicators, humidity control regulation strategies, and an optimized parameter set for flow regulation; and the performance tuning instruction set includes temperature regulation commands, humidity level adjustment commands, and air conditioning system flow control commands.

[0079] In the data acquisition module, a sensor network-based approach is used to collect operational data from the air conditioning system, including information such as temperature, humidity, and flow rate. This data is formatted as time-series data. This module uses an outlier detection algorithm, specifically the statistical Z-score method, to standardize the data and calculate the Z-score for each data point, which is the standard deviation multiple between the data point and the mean. Data points with Z-scores exceeding a predetermined threshold are considered outliers and are removed from the dataset. Next, data smoothing techniques, including the weighted moving average method, are applied. By assigning higher weights to recent data points, the average is calculated to reduce the impact of random fluctuations. This series of refinement operations not only improves the accuracy and reliability of the data but also provides high-quality input data for subsequent modules, effectively reducing noise and outlier interference and ensuring the quality of the cleaned data.

[0080] In the dynamic modeling module, a nonlinear dynamic modeling method is used to construct a mathematical model based on the cleaned data. This includes applying differential equation-based modeling techniques, considering the dynamic interaction of internal and external factors of the system, and combining a sliding mode control strategy. By defining a set of sliding mode equations and arrival conditions, the model's disturbance resistance is optimized. The implementation of sliding mode control includes selecting a suitable sliding surface and adjusting the arrival conditions to ensure that the system output can quickly and accurately track the given reference trajectory. Even under external disturbances and parameter changes, this method not only improves the model's adaptability and robustness but also enables the model to more accurately predict the system's performance under various operating conditions.

[0081] In the energy efficiency identification module, the Support Vector Machine (SVM) algorithm is used to analyze the output of the anti-disturbance optimization model, identify the system's operating efficiency patterns and potential abnormal states. By selecting an appropriate kernel function (such as the radial basis function) and adjusting the penalty parameter C, the SVM algorithm finds the optimal decision boundary in the high-dimensional space to minimize classification errors. In addition, cross-validation is used to optimize the model parameters, ensuring the model's generalization ability and accuracy. Through refined operation, the energy efficiency identification module can not only accurately classify different energy efficiency states, but also identify abnormal or inefficient operating modes in the system, providing a scientific basis for energy optimization.

[0082] In the energy allocation module, linear programming is used to analyze the results of the energy efficiency identification module and optimize the energy allocation strategy. Specifically, a linear programming model with the goal of minimizing energy consumption is established, constraints on energy supply and demand are defined, and the simplex algorithm is applied to solve this optimization problem. The energy allocation is adjusted to match real-time demand and reduce energy waste. By accurately calculating the optimal match between energy demand and supply, the energy efficiency of the entire system is effectively improved, ensuring the rational allocation and use of energy and reducing energy consumption.

[0083] In the maintenance optimization module, based on the energy optimization scheme, cluster analysis and principal component analysis (PCA) are used to analyze the system operation data. First, the K-means clustering algorithm is applied to group the data and identify different operating states or modes. Then, PCA is used to reduce the dimensionality of the grouped data and extract key performance indicators. The operation allows the module to accurately identify the system's key maintenance needs and efficiency improvement opportunities, providing data support for the formulation of maintenance plans and energy efficiency improvement measures. Through this refined analysis method, the maintenance optimization module can effectively improve the system's operating efficiency and reliability.

[0084] In the control strategy optimization module, a deep Q-network (DQN) learning algorithm is used to optimize the control strategy. By constructing a reinforcement learning model that interacts with the environment, the DQN algorithm can find the optimal balance between exploration (discovering new knowledge) and utilization (applying known knowledge). The algorithm updates the value function by collecting reward information of state-action pairs, thereby iteratively learning the optimal strategy. This method enables the control strategy optimization module to automatically adjust parameters to cope with different environments and system states, and optimize energy consumption and indoor environmental quality.

[0085] In the real-time monitoring and adjustment module, model predictive control (MPC) technology is used to dynamically adjust system parameters based on real-time data and changes in the external environment. MPC builds a predictive model to predict the system's behavior over a future period of time, and then optimizes the control input based on the prediction to meet performance and energy efficiency targets. Through repeated optimization processes, the real-time monitoring and adjustment module can ensure that the system can operate in the optimal state under various operating conditions, achieving continuous optimization of energy efficiency and performance.

[0086] Please see Figure 2 and Figure 3 In this embodiment of the application, the data acquisition module includes a sensor submodule, an outlier processing submodule, and a data smoothing submodule;

[0087] The sensor submodule is used to collect environmental and system operation data, including temperature, humidity and flow data, and to determine the real-time performance and accuracy of the operation data to generate a raw dataset.

[0088] The outlier processing submodule is used to identify and remove statistical outliers based on the original dataset using the Z-score method, verify the quality and reliability of the dataset, and generate an outlier-free dataset.

[0089] The data smoothing submodule is used to reduce random fluctuations in time series data by using a moving average method based on outlier removal datasets, thereby optimizing the smoothness and usability of the data and generating cleaned data.

[0090] In the sensor submodule, through a sensor network-based design, the system meticulously collects data related to the environment and the operation of the air conditioning system. This data includes temperature, humidity, and flow rate data, presented in a time-series format with each data point bearing a precise timestamp, ensuring real-time performance and accuracy. The dense layout of the sensor network enables the collection of comprehensive data from different parts of the system and the environment, increasing both data coverage and monitoring accuracy. The data is transmitted to the data processing center via a pre-defined communication protocol to generate a raw dataset. This process ensures data integrity and timeliness, providing a foundation for subsequent data processing and analysis.

[0091] In the outlier handling submodule, the Z-score method is applied to identify and process outliers in the original dataset, systematically removing statistical outliers. The Z-score method identifies outliers by calculating the standard deviation of each data point from the overall data mean. In practice, the mean and standard deviation of the entire dataset are first calculated, and then the Z-score of each data point is calculated, which is the difference between the data point and the mean divided by the standard deviation. If the Z-score of a data point exceeds a pre-set threshold (e.g., ±3), it is identified as an outlier and removed from the dataset. The application of this method not only improves the quality and reliability of the dataset but also provides a guarantee to avoid errors caused by outliers in subsequent analysis. The generated outlier-free dataset more accurately reflects the true operating state of the system.

[0092] In the data smoothing submodule, the moving average method is used to process the outlier removal dataset, effectively reducing random fluctuations in time series data and optimizing data smoothness and usability. The moving average method smooths the data series by calculating the average value of data points within a certain time window. In specific operations, an appropriate time window size is selected, and the average value of data points including that point and the time windows before and after it is calculated for each time point. This average value is used as the smoothing value of the data at that time point. This method effectively reduces random fluctuations in time series data, making the data series more stable and facilitating subsequent analysis and modeling. The generated cleaned data not only retains the main trends and characteristics of the data but also improves the quality of data analysis and the accuracy of modeling, providing reliable data support for the system's energy efficiency identification and performance optimization.

[0093] Please see Figure 2 and Figure 4 In this embodiment, the dynamic modeling module includes a nonlinear modeling submodule, a feedback linearization submodule, and a sliding mode design submodule;

[0094] The nonlinear modeling submodule is used to construct a dynamic model of the air conditioning system based on the purified data using a radial basis function network, and to capture the nonlinear characteristics of the system to generate a preliminary mathematical model.

[0095] The feedback linearization submodule is used to simplify the nonlinear model into a linear model by applying feedback linearization based on the preliminary mathematical model and through mathematical transformation, and to adjust the establishment and implementation of the control strategy to generate a linearized model.

[0096] The sliding mode design submodule is used to define the sliding surface and design the control law based on the linearized model and the sliding mode control strategy, optimize the robustness of the model to external disturbances and parameter changes, and generate a disturbance-resistant optimization model.

[0097] In the nonlinear modeling submodule, a dynamic model of the air conditioning system is constructed based on the purified data using the Radial Basis Function Network (RBFN) algorithm. This process begins with a thorough analysis of the purified data, which is preserved in time series format, containing sequences of key parameters such as temperature, humidity, and flow rate. This ensures that the model can capture the real-time characteristics of the system operation. The application of the RBFN aims to address the challenges of nonlinear system modeling. By selecting appropriate radial basis functions as activation functions for the hidden layers, the RBFN can map the input data to a higher-dimensional space, thereby capturing the nonlinear relationships in the data. During training, the network weights and the center of the basis functions are adjusted to make the network output approximate the actual output of the system. The core of this method lies in its ability to adaptively learn the nonlinear features in the data. The generated preliminary mathematical model not only reflects the dynamic behavior of the air conditioning system in detail but also accurately captures its nonlinear characteristics.

[0098] In the feedback linearization submodule, the complex nonlinear model is transformed into a more easily analyzed and controlled linear model based on the preliminary mathematical model by using feedback linearization technology. The feedback linearization process involves in-depth mathematical analysis of the preliminary mathematical model, accurately identifying the nonlinear parts in the model, and applying mathematical transformations, such as variable substitution and feedback control strategies, to eliminate nonlinear effects. The execution of this transformation not only depends on the accurate identification of the nonlinear characteristics of the model, but also requires careful design of the control law to ensure that the control performance of the model is not compromised while eliminating the nonlinear effects. The generated linearized model simplifies the system control problem, making the design and implementation of control strategies more direct and effective.

[0099] In the sliding mode design submodule, the sliding mode control strategy is used to further optimize the system's control performance based on the linearized model. The core of sliding mode control lies in defining one or more sliding surfaces and designing corresponding control laws so that the system state can reach the desired equilibrium point along the sliding surface. In specific implementation, the key state variables of the system are first identified based on the linearized model, and then the sliding surface is designed. This usually involves selecting appropriate sliding conditions and arrival rates to ensure that the system state can converge quickly and accurately to the sliding surface from any initial position. Subsequently, by designing control laws, the system input is adjusted to ensure that once the system state reaches the sliding surface, it can slide along the surface to reach the desired equilibrium state, while maintaining high robustness to external disturbances and parameter changes. The implementation of this control strategy significantly improves the system's resistance to various uncertainties. The generated disturbance rejection optimization model not only ensures the stability and reliability of the system, but also ensures its robustness in the face of external disturbances and internal parameter changes.

[0100] Please see Figure 2 and Figure 5 In this embodiment of the application, the energy efficiency identification module includes a feature analysis submodule, a data classification submodule, and a state identification submodule;

[0101] The feature analysis submodule is used to extract the periodic and trend features of temperature, humidity and energy consumption based on the anti-disturbance optimization model, using time series analysis and frequency domain analysis methods, calculate the statistical mean and variance index, prepare input data for machine learning, and generate feature vector sets.

[0102] The data classification submodule is used to classify the system's operating state based on the feature vector set, using the support vector machine algorithm, mapping the features to a high-dimensional space through the kernel trick, constructing a classification hyperplane using the maximum margin principle, and generating a classification and recognition model.

[0103] The state recognition submodule is used to perform state recognition on real-time data streams based on classification and recognition models, including using models to make real-time predictions, identifying the current operating state of the system, and generating energy efficiency recognition records.

[0104] In the feature analysis submodule, the output data of the disturbance rejection optimization model is refined by comprehensively applying time series analysis and frequency domain analysis methods. This mainly involves key indicators such as temperature, humidity, and energy consumption. The data retains its time series format, enabling the analysis to fully utilize the information about data changes over time. Time series analysis reveals the long-term trend and periodic fluctuations of the data by calculating statistics such as moving average and moving standard deviation. Frequency domain analysis transforms the data into frequency space and applies Fourier transform to identify the main frequency components of the data, thereby revealing its periodic characteristics. In this process, the statistical mean and variance of the data are also calculated. The statistical indicators provide important information about the data distribution and variability for subsequent machine learning models. Through refinement operations, a vector set containing key features is generated. The feature vector set not only contains important information from the original data but also enhances the expressive power of the data through statistical and frequency domain analysis.

[0105] In the data classification submodule, the Support Vector Machine (SVM) algorithm is used to process the feature vector set and construct a classification model of the system's operating state. The SVM algorithm maps the feature space to a high-dimensional space through kernel tricks, and finds the maximum margin hyperplane in this high-dimensional space to distinguish different operating states. In the specific implementation, a suitable kernel function (such as the radial basis function kernel) is first selected to handle the classification problem of nonlinear features. Then, the SVM parameters, including the penalty parameter C and the kernel function parameters, are adjusted through training data to optimize the performance of the classification model. Through fine-tuning, the SVM model can effectively distinguish different operating states of the system. The generated classification and recognition model not only improves the accuracy of state recognition, but also enhances the model's generalization ability to unseen data, providing support for real-time monitoring and control.

[0106] In the state recognition submodule, based on the classification and recognition model, detailed state recognition is performed on the real-time data stream. This submodule receives the system's operating data in real time, including indicators such as temperature, humidity, and energy consumption, and converts the data into feature vectors. These vectors interact with the trained SVM model, which uses the learned classification hyperplane to classify the input feature vectors and predict the current operating state of the system. During this process, the model continuously receives new data and updates the system state recognition results in real time. Through this real-time data processing and state recognition, an energy efficiency recognition record is generated. This record not only provides an immediate reference for the system's energy efficiency management but also provides an important basis for subsequent energy efficiency optimization and control strategy adjustments.

[0107] Please see Figure 2 and Figure 6 In this embodiment of the application, the energy allocation module includes a demand analysis submodule, a linear programming submodule, and a strategy generation submodule;

[0108] The demand analysis submodule is used to analyze energy demand curves under different conditions based on energy efficiency identification records, using mathematical modeling methods, and using long short-term memory networks to assess future energy demand and generate energy demand forecast data.

[0109] The linear programming submodule is used to define an optimization model based on energy demand forecast data and using a linear programming algorithm. This includes setting constraints on energy supply and demand, solving for the optimal energy allocation measures under the operating mode, and generating an optimized allocation scheme.

[0110] The strategy generation submodule is used to establish an energy allocation execution plan based on the optimized allocation scheme, including adjusting the operating parameters of the air conditioning system, setting energy usage priorities and allocation time, verifying the feasibility and execution efficiency of the strategy, and generating an energy optimization scheme.

[0111] In the demand analysis submodule, mathematical modeling methods and the Long Short-Term Memory (LSTM) network algorithm are used to analyze energy demand curves under different conditions based on energy efficiency identification records and to assess future energy demand. The data format involved in this process is mainly time series data, including historical temperature, humidity, and energy consumption information. First, mathematical modeling methods are used to perform preliminary analysis on historical energy demand data to identify the periodicity and trend characteristics of energy demand. Then, the LSTM network is used to learn and predict the features in depth. The key advantage of LSTM is that it can handle and remember long-term dependencies, which is particularly important for accurately predicting energy demand. By configuring parameters such as the number of hidden layer nodes, learning rate, and training period of the LSTM network, the network can learn complex patterns in historical data and predict future energy demand. The generated energy demand prediction data describes in detail the energy demand at each time point within the prediction period.

[0112] In the linear programming submodule, an optimization model is defined based on energy demand forecast data using a linear programming algorithm. This includes setting constraints on energy supply and demand. The data format used in this process is the energy demand forecast data generated by the demand analysis submodule, which includes the energy demand at each point in time during the forecast period. In the construction of the linear programming model, the objective function is first defined, which is to minimize total energy consumption or cost. At the same time, the limitations of energy supply and the requirements of the demand side are taken into account as constraints. By applying linear programming solution techniques such as the simplex method or the interior point method, the optimal energy allocation scheme that satisfies all constraints is found. The generated optimized allocation scheme details how to allocate energy resources under different operating modes to achieve energy minimization and efficiency maximization.

[0113] In the strategy generation submodule, an energy allocation execution plan is established based on the optimized allocation scheme. This includes adjusting the operating parameters of the air conditioning system, setting energy usage priorities and allocation times. Through detailed analysis of the optimized allocation scheme, the optimal operating parameters of the air conditioning system in different time periods are determined, such as temperature setpoints, humidity control ranges, and system runtime. In addition, based on the priority of energy demand and supply conditions, the priority order and time schedule of energy use are formulated to ensure sufficient energy supply during peak demand periods and to implement energy-saving measures during low demand periods. By simulating and verifying the implementation effect of the strategy, the feasibility and execution efficiency of the strategy are evaluated, and further optimization and adjustments are made. The generated energy optimization scheme provides a detailed energy management and scheduling guide for the air conditioning system.

[0114] Please see Figure 2 and Figure 7 In this embodiment, the maintenance optimization module includes a cluster analysis submodule, a performance index extraction submodule, and a maintenance strategy design submodule;

[0115] The clustering analysis submodule is used to group system operation data based on energy optimization schemes and K-means clustering algorithm, identify different operating states, provide a classification basis for performance analysis, and generate data classification groups.

[0116] The performance index extraction submodule is used to extract key performance indicators from data classification groups using principal component analysis to reflect system operating efficiency and potential problems, and generate a performance index summary.

[0117] The maintenance strategy design submodule is used to design maintenance plans and energy efficiency improvement measures based on performance index summaries, including preventive maintenance, parameter optimization and equipment upgrades, to optimize system reliability and efficiency, and generate system maintenance and efficiency proposals.

[0118] In the clustering analysis submodule, the K-means clustering algorithm is used to group system operation data based on an energy optimization scheme to identify differential operating states. During this process, the data format is primarily a multidimensional dataset containing various system operation parameters such as temperature, humidity, and energy consumption. The K-means algorithm first randomly selects K data points as initial cluster centers. Then, it assigns the data points to the nearest cluster centers according to their Euclidean distance, forming preliminary cluster groups. Next, it recalculates the center point of each group, i.e., the mean of all points in the group, and then reassigns data points based on the new centers. This process iterates until the cluster centers no longer change or the change is below a set threshold. Through this method, system operation data is effectively divided into several groups, each representing a specific operating state. The generated data classification groups provide a classification basis for subsequent performance analysis and optimization. This operation not only reveals the inherent differences in system operating states but also provides a basis for identifying potential operational efficiency problems and designing improvement measures.

[0119] In the performance index extraction submodule, Principal Component Analysis (PCA) is used to extract key performance indicators based on data classification groups. PCA is a statistical method that transforms the original data into a set of linearly independent variables, i.e., principal components, through orthogonal transformation. The principal components are arranged in descending order of variance, and the first few principal components usually retain most of the data information. In practice, the covariance matrix of the data is first calculated, and then the eigenvalues ​​and eigenvectors of this covariance matrix are found. The eigenvectors correspond to the principal component directions of the data, and the magnitude of the eigenvalues ​​represents the variance in the corresponding principal component direction, i.e., the amount of information. By selecting the eigenvectors with the largest eigenvalues, key performance indicators reflecting system operating efficiency and potential problems can be extracted. The generated performance index summary not only simplifies the complexity of the data but also highlights the most important performance information.

[0120] In the maintenance strategy design submodule, maintenance plans and energy efficiency improvement measures are designed based on the performance index summary. These measures include preventative maintenance, parameter optimization, and equipment upgrades, aiming to optimize system reliability and efficiency. During the design process, key information extracted from the performance index summary is comprehensively considered, and in-depth analysis of existing system problems is conducted to identify key factors affecting system efficiency and reliability. Then, based on the analysis results, specific maintenance plans are formulated, such as regularly inspecting and cleaning key components, adjusting or optimizing operating parameters, and upgrading outdated or inefficient equipment. In addition, the actual operating conditions of the system and resource availability are taken into account to ensure that the designed maintenance strategy is both practical and efficient. The generated system maintenance and efficiency proposal details the implementation plan and expected effects of each measure, providing clear guidance for improving overall system performance and reducing operating costs.

[0121] Please see Figure 2 and Figure 8 In this embodiment, the control strategy optimization module includes a reinforcement learning training submodule, a strategy evaluation submodule, and a strategy implementation submodule.

[0122] The reinforcement learning training submodule is used to learn the optimal control strategy based on system maintenance and performance proposals, using a deep Q-network and a reward feedback mechanism, to optimize energy consumption and maintain environmental comfort, and generate a strategy learning model.

[0123] The strategy evaluation submodule is used to evaluate the effectiveness of a strategy based on a strategy learning model. By simulating system operation under different environmental conditions, it verifies the effectiveness and adaptability of the control strategy and generates strategy evaluation records.

[0124] The strategy implementation submodule is used to plan the application steps of the control strategy based on the strategy evaluation records and using the particle swarm optimization algorithm, including parameter adjustment, personnel training and configuration updates, and to generate control strategy improvement schemes.

[0125] In the reinforcement learning training submodule, a Deep Q-Network (DQN) is used to learn the optimal control policy based on system maintenance and efficiency proposals to optimize energy consumption and maintain environmental comfort. The DQN combines the advantages of deep learning and reinforcement learning, enabling it to handle high-dimensional state spaces and allowing the algorithm to learn policies in complex decision-making environments. During training, the DQN obtains sequences of states, actions, and rewards through interaction with the environment, updating the Q-value function (the expected reward for taking an action in a given state) using this information. The training algorithm employs an experience replay mechanism and a fixed Q-objective technique to improve learning stability and efficiency. Experience replay allows the network to learn from past experiences, avoiding interference from the temporal correlation of data in the learning process, while the fixed Q-objective reduces the moving target problem during training by stabilizing the target value. Through iterative optimization, the DQN can find the optimal control policy that minimizes energy consumption while maintaining environmental comfort. The generated policy learning model details the optimal actions to be taken under different environmental states, providing strong decision support for automated control systems and ensuring the efficiency and comfort of system operation.

[0126] In the strategy evaluation submodule, the effectiveness of the strategy is evaluated based on the strategy learning model. By simulating system operation under different environmental conditions, this submodule verifies the effectiveness and adaptability of the control strategy. The strategy evaluation process involves constructing a simulation environment that reflects various operating conditions in the real world, and then applying the strategy learning model in this simulation environment to observe its impact on system operation. Evaluation indicators include the energy consumption reduction rate, the level of environmental comfort maintenance, and the ability to respond to emergencies. Through this method, we can comprehensively understand the performance of the strategy in practical applications, identify advantages and potential areas for improvement, and generate a detailed report on the performance of the strategy under various conditions, providing important feedback information for strategy optimization and implementation.

[0127] In the strategy implementation submodule, based on the strategy evaluation records, the application steps of the control strategy are planned using the particle swarm optimization algorithm. Particle swarm optimization (PSO) is an optimization algorithm based on swarm intelligence that finds the optimal solution by simulating the social behavior of a flock of birds. In the context of strategy implementation, PSO is used to optimize the specific implementation steps of the control strategy, such as the order and magnitude of parameter adjustments, personnel training arrangements, and configuration update plans. Each particle represents an implementation plan. Through information sharing and iterative search among particles, PSO can find the optimal steps for implementing the control strategy. This process not only considers implementation efficiency but also ensures a smooth transition of strategy changes and minimizes interference with system operation. The generated control strategy improvement scheme details the implementation plan, including the required resources, timeline, and expected results.

[0128] Please see Figure 2 and Figure 9 In this embodiment, the real-time monitoring and adjustment module includes a real-time data monitoring submodule, an adjustment decision submodule, and a parameter adjustment submodule;

[0129] The real-time data monitoring submodule is used to monitor the collected operating parameters of the air conditioning system in real time, including key indicators such as temperature, humidity and energy consumption, based on the control strategy improvement scheme and using real-time data monitoring technology, and generate monitoring status information.

[0130] The adjustment decision submodule is used to analyze the deviation between real-time data and expected performance based on monitoring status information and DSS technology, determine whether system parameters need to be adjusted, and generate adjustment decision schemes in response to changes in the external environment or deviations in internal performance.

[0131] The parameter adjustment submodule is used to implement parameter adjustment based on the adjustment decision scheme and using model predictive control technology. The parameter adjustment includes adjusting temperature, humidity and wind speed, and verifying the matching of system operating parameters with performance and energy efficiency requirements, and generating a performance tuning instruction set.

[0132] In the real-time data monitoring submodule, real-time data monitoring technology is used to monitor the operating parameters of the air conditioning system in real time. The data format involved is mainly continuous time series data, including key indicators such as temperature, humidity, and energy consumption. The real-time monitoring technology collects data in real time through sensors deployed at key locations in the system and transmits it to the monitoring center via the network. The monitoring system analyzes the data in real time and detects any deviations from normal operating parameters. This monitoring not only includes real-time display of values ​​but also analysis of data trends and the setting of early warning mechanisms. When the monitored parameters exceed the preset normal operating range, the system will automatically trigger an alarm. The generated monitoring status information records the real-time operating status of the system in detail, providing a basis for further adjustment decisions and system optimization.

[0133] In the adjustment decision submodule, based on monitoring status information, Decision Support System (DSS) technology is applied to analyze the deviation between real-time data and expected performance to determine whether system parameters need to be adjusted to cope with changes in the external environment or internal performance deviations. DSS technology combines data analysis, simulation prediction, and optimization algorithms to provide data-driven decision support. In practice, this submodule first defines the evaluation criteria for performance deviation, and then uses data analysis tools to conduct in-depth analysis of real-time monitoring data to identify deviations from expected performance targets. Based on the severity and nature of the deviation, DSS proposes adjustment suggestions, including adjusting temperature setpoints, humidity control levels, or wind speeds. The generated adjustment decision scheme details the required adjustment measures, providing a basis for ensuring the system operates in optimal condition.

[0134] In the parameter tuning submodule, based on the tuning decision scheme, parameter tuning is implemented using Model Predictive Control (MPC) technology. MPC is an advanced control strategy that can predict the system's behavior over a future period and optimize the current control actions accordingly. In this process, MPC first predicts the future operating state of the system under the current control strategy based on the system's mathematical model. Then, it calculates the control input that optimizes the system's performance indicators through optimization algorithms. The control input is then used to adjust the system's operating parameters, such as temperature, humidity, and wind speed, to match performance and energy efficiency requirements. In this way, the parameter tuning submodule can ensure real-time optimization of system parameters, improve the system's operating efficiency and comfort. The generated performance tuning instruction set details the specific content and execution steps of the tuning measures, providing real-time and precise tuning support for system operation and ensuring that the system can operate efficiently and stably.

[0135] In summary, the energy-saving automatic control system for the variable flow water system in air conditioning provided in this application ensures the accuracy and reliability of decision-making through precise data acquisition and purification, providing high-quality input data for the system. This not only reduces errors in the data processing process but also lays a solid foundation for subsequent model construction and analysis. The nonlinear dynamic model constructed using a radial basis function network combined with a sliding mode variable structure control strategy significantly improves the system's adaptability and robustness to environmental changes and external disturbances. This model accurately captures the system's dynamic behavior and nonlinear characteristics, providing a scientific basis for formulating effective control strategies and ensuring the stability and reliability of system operation. The identification of efficiency patterns and potential abnormal states through the support vector machine algorithm further enhances the system's intelligence level. This step not only enables the system to promptly identify and respond to efficiency changes and potential anomalies but also provides guidance for energy allocation and system maintenance, helping to adjust strategies in a timely manner and optimize operating efficiency. The application of linear programming plays a crucial role in energy allocation. Through refined energy demand analysis and optimized allocation strategies, it significantly improves energy use efficiency, reducing both energy consumption and operating costs, thus contributing to the achievement of energy conservation and emission reduction goals. The combined use of K-means clustering and principal component analysis extracted key performance indicators from system performance data, providing a scientific basis for the design of maintenance plans and energy efficiency improvement measures. This step not only improved the targeting and effectiveness of maintenance work but also laid the foundation for long-term performance optimization and energy efficiency improvement. The application of deep Q-networks and model predictive control technology enabled dynamic learning and optimization of the control strategy, ensuring optimal system operation under real-time data and external environmental changes. This intelligent control method not only enhanced the system's adaptability and flexibility but also ensured that energy consumption was minimized while maintaining environmental comfort.

[0136] Based on the above technical solution, this application also proposes a control method for an energy-saving automatic control system of an air conditioning variable flow water system, applied to the energy-saving automatic control system of the air conditioning variable flow water system described in this application. The method includes the following steps:

[0137] Step 1: Collect environmental and system operating data, use outlier detection algorithms to identify and remove outliers in the operating data, and perform data smoothing to generate cleaned data;

[0138] Step 2: Based on the purified data, construct a mathematical model using a nonlinear dynamic modeling method, and combine it with a sliding mode control strategy to optimize the model's resistance to external disturbances, thereby generating an anti-disturbance optimization model.

[0139] Step 3: Based on the disturbance rejection optimization model, the support vector machine algorithm is used to analyze the system performance data, identify the operating efficiency mode and potential abnormal state, perform mode classification and state identification, and generate energy efficiency identification records.

[0140] Step 4: Based on energy efficiency identification records, apply linear programming to analyze energy demand, adjust allocation strategies to match real-time demand and optimize energy waste, and generate energy optimization solutions;

[0141] Step 5: Based on the energy optimization scheme, the system operation data is grouped using cluster analysis. Through principal component analysis, key performance indicators are extracted, maintenance plans and energy efficiency improvement measures are formulated, and system maintenance and efficiency proposals are generated.

[0142] Step 6: Based on system maintenance and performance proposals, a deep Q-network is used to learn control strategies through a reward mechanism, adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement plans.

[0143] Step 7: Based on the control strategy improvement scheme, utilize model predictive control technology and adjust system parameters according to real-time monitoring data and changes in the external environment to optimize system performance and energy efficiency, and generate a performance tuning instruction set.

[0144] It is understood that since the control method of the energy-saving automatic control system of the air conditioning variable flow water system described in the embodiments of this application is based on the method of the energy-saving automatic control system of the air conditioning variable flow water system described in the embodiments, the method disclosed in the embodiments is relatively simple to describe because it corresponds to the system disclosed in the embodiments. For relevant parts, please refer to the description of the system.

Claims

1. An energy-saving automatic control system for air conditioning variable flow water systems, characterized in that, The system includes a data acquisition module, a dynamic modeling module, an energy efficiency identification module, an energy allocation module, a maintenance optimization module, a control strategy optimization module, and a real-time monitoring and adjustment module. The data acquisition module is used to collect environmental and system operation data, use an outlier detection algorithm to identify and remove outliers in the operation data, and perform data smoothing processing to generate cleaned data. The dynamic modeling module is used to construct a mathematical model based on the purified data using a nonlinear dynamic modeling method, and to combine a sliding mode control strategy to optimize the model's resistance to external disturbances, thereby generating an anti-disturbance optimization model. The energy efficiency identification module is used to analyze system performance data based on the anti-disturbance optimization model and the support vector machine algorithm, identify operating efficiency modes and potential abnormal states, perform mode classification and state identification, and generate energy efficiency identification records. The energy allocation module is used to analyze energy demand based on energy efficiency identification records, apply linear programming, adjust allocation strategies, match real-time demand, optimize energy waste, and generate energy optimization schemes. The maintenance optimization module is used to group system operation data based on energy optimization schemes using cluster analysis, extract key performance indicators through principal component analysis, formulate maintenance plans and energy efficiency improvement measures, and generate system maintenance and efficiency proposals. The control strategy optimization module is used to learn control strategies based on system maintenance and performance proposals, using a deep Q network and a reward mechanism, to adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement schemes. The real-time monitoring and adjustment module is used to improve the scheme based on the control strategy, utilize model predictive control technology and adjust system parameters according to real-time monitoring data and changes in the external environment, optimize system performance and energy efficiency, and generate a performance tuning instruction set. The purified data includes indoor temperature sequence, humidity level time sequence, and air conditioning system flow data sequence. The disturbance rejection optimization model includes the model's stability parameters, response coefficient to external temperature changes, and sensitivity index to flow regulation. The energy optimization scheme includes an energy demand prediction map, a priority allocation map, and an optimal energy consumption adjustment strategy. The system maintenance and performance proposal includes a predetermined maintenance schedule, a monitoring plan for key performance indicators, and performance improvement measures for the target issues; The improved control strategy includes temperature control adjustment indicators, humidity control adjustment strategies, and an optimized parameter set for flow rate adjustment. The performance tuning instruction set includes temperature adjustment commands, humidity level adjustment commands, and air conditioning system flow control commands. The energy allocation module includes a demand analysis submodule, a linear programming submodule, and a strategy generation submodule; The demand analysis submodule is used to analyze energy demand curves under different conditions based on energy efficiency identification records, using mathematical modeling methods, and using long short-term memory networks to assess future energy demand and generate energy demand forecast data. The linear programming submodule is used to define an optimization model based on energy demand forecast data and using a linear programming algorithm. This includes setting constraints on energy supply and demand, solving for the optimal energy allocation measures under the operating mode, and generating an optimized allocation scheme. The strategy generation submodule is used to establish an energy allocation execution plan based on the optimized allocation scheme, including adjusting the operating parameters of the air conditioning system, setting energy usage priorities and allocation time, verifying the feasibility and execution efficiency of the strategy, and generating an energy optimization scheme. The real-time monitoring and adjustment module includes a real-time data monitoring submodule, an adjustment decision submodule, and a parameter adjustment submodule; The real-time data monitoring submodule is used to monitor the collected operating parameters of the air conditioning system in real time, including key indicators such as temperature, humidity and energy consumption, based on the control strategy improvement scheme and using real-time data monitoring technology, and generate monitoring status information. The adjustment decision submodule is used to analyze the deviation between real-time data and expected performance based on monitoring status information and DSS technology, determine whether system parameters need to be adjusted, and generate adjustment decision schemes in response to changes in the external environment or deviations in internal performance. The parameter adjustment submodule is used to implement parameter adjustment based on the adjustment decision scheme and using model predictive control technology. The parameter adjustment includes adjusting temperature, humidity and wind speed, and verifying the matching of system operating parameters with performance and energy efficiency requirements, and generating a performance tuning instruction set.

2. The energy-saving automatic control system for air conditioning variable flow water system according to claim 1, characterized in that, The data acquisition module includes a sensor submodule, an outlier processing submodule, and a data smoothing submodule; The sensor submodule is used to collect environmental and system operation data, including temperature, humidity and flow data, and to determine the real-time performance and accuracy of the operation data to generate a raw dataset. The outlier processing submodule is used to identify and remove statistical outliers based on the original dataset using the Z-score method, verify the quality and reliability of the dataset, and generate an outlier-free dataset. The data smoothing submodule is used to reduce random fluctuations in time series data by using a moving average method based on outlier removal datasets, thereby optimizing the smoothness and usability of the data and generating cleaned data.

3. The energy-saving automatic control system for air conditioning variable flow water system according to claim 1, characterized in that, The dynamic modeling module includes a nonlinear modeling submodule, a feedback linearization submodule, and a sliding mode design submodule; The nonlinear modeling submodule is used to construct a dynamic model of the air conditioning system based on the purified data using a radial basis function network, and to capture the nonlinear characteristics of the system to generate a preliminary mathematical model. The feedback linearization submodule is used to simplify the nonlinear model into a linear model by applying feedback linearization based on the preliminary mathematical model and through mathematical transformation, and to adjust the establishment and implementation of the control strategy to generate a linearized model. The sliding mode design submodule is used to define the sliding surface and design the control law based on the linearized model and the sliding mode control strategy, optimize the robustness of the model to external disturbances and parameter changes, and generate a disturbance-resistant optimization model.

4. The energy-saving automatic control system for air conditioning variable flow water system according to claim 1, characterized in that, The energy efficiency identification module includes a feature analysis submodule, a data classification submodule, and a state identification submodule; The feature analysis submodule is used to extract the periodic and trend features of temperature, humidity and energy consumption based on the anti-disturbance optimization model, using time series analysis and frequency domain analysis methods, calculate the statistical mean and variance index, prepare input data for machine learning, and generate feature vector sets. The data classification submodule is used to classify the system's operating state based on the feature vector set, using the support vector machine algorithm, mapping the features to a high-dimensional space through the kernel trick, constructing a classification hyperplane using the maximum margin principle, and generating a classification and recognition model. The state recognition submodule is used to perform state recognition on real-time data streams based on classification and recognition models, including using models to make real-time predictions, identifying the current operating state of the system, and generating energy efficiency recognition records.

5. The energy-saving automatic control system for air conditioning variable flow water system according to claim 1, characterized in that, The maintenance optimization module includes a cluster analysis submodule, a performance index extraction submodule, and a maintenance strategy design submodule; The clustering analysis submodule is used to group system operation data based on energy optimization schemes and K-means clustering algorithm, identify different operating states, provide a classification basis for performance analysis, and generate data classification groups. The performance index extraction submodule is used to extract key performance indicators from data classification groups using principal component analysis to reflect system operating efficiency and potential problems, and generate a performance index summary. The maintenance strategy design submodule is used to design maintenance plans and energy efficiency improvement measures based on performance index summaries, including preventive maintenance, parameter optimization and equipment upgrades, to optimize system reliability and efficiency, and generate system maintenance and efficiency proposals.

6. The energy-saving automatic control system for air conditioning variable flow water system according to claim 1, characterized in that, The control strategy optimization module includes a reinforcement learning training submodule, a strategy evaluation submodule, and a strategy implementation submodule. The reinforcement learning training submodule is used to learn the optimal control strategy based on system maintenance and performance proposals, using a deep Q-network and a reward feedback mechanism, to optimize energy consumption and maintain environmental comfort, and generate a strategy learning model. The strategy evaluation submodule is used to evaluate the effectiveness of a strategy based on a strategy learning model. By simulating system operation under different environmental conditions, it verifies the effectiveness and adaptability of the control strategy and generates strategy evaluation records. The strategy implementation submodule is used to plan the application steps of the control strategy based on the strategy evaluation records and using the particle swarm optimization algorithm, including parameter adjustment, personnel training and configuration updates, and to generate control strategy improvement schemes.

7. A control method for an energy-saving automatic control system of an air conditioning variable flow water system, characterized in that, The method, applied to the energy-saving automatic control system of the air conditioning variable flow water system as described in any one of claims 1 to 6, comprises the following steps: Collect environmental and system operation data, use outlier detection algorithms to identify and remove outliers in the operation data, and perform data smoothing processing to generate cleaned data; Based on the purified data, a mathematical model is constructed using a nonlinear dynamic modeling method, and the resistance of the model to external disturbances is optimized by combining a sliding mode control strategy to generate an anti-disturbance optimization model. Based on the disturbance rejection optimization model, the support vector machine algorithm is used to analyze system performance data, identify operating efficiency modes and potential abnormal states, perform mode classification and state identification, and generate energy efficiency identification records. Based on energy efficiency identification records, linear programming is applied to analyze energy demand, adjust allocation strategies, match real-time demand, and optimize energy waste to generate energy optimization solutions. Based on the energy optimization scheme, the system operation data is grouped using cluster analysis, and key performance indicators are extracted through principal component analysis. Maintenance plans and energy efficiency improvement measures are then formulated to generate system maintenance and efficiency proposals. Based on system maintenance and performance proposals, a deep Q network is used to learn control strategies through a reward mechanism, adjust energy consumption and maintain indoor environmental quality, and generate control strategy improvement plans. Based on the control strategy improvement scheme, model predictive control technology is used to adjust system parameters according to real-time monitoring data and changes in the external environment, thereby optimizing system performance and energy efficiency and generating a performance tuning instruction set.

Images