Wall-hanging stove constant-temperature energy-saving control system based on enhanced autonomous learning and SAC algorithm

Abstract

The invention relates to the technical field of intelligent control, in particular to a wall-hanging stove constant-temperature energy-saving control system based on enhanced autonomous learning and an SAC algorithm, and the system comprises a gas flow optimization module which collects heating operation data of a wall-hanging stove, classifies the heating operation data according to the change characteristics of a plurality of events, calculates the frequency of occurrence of each event, and outputs the frequency of occurrence of each event; and analyzing the change condition of the operation data according to the event occurrence frequency, and generating a gas flow data set. According to the method, the gas flow data is generated through real-time collection and multi-dimensional classification analysis of the operation data of the wall-hanging stove in combination with the event triggering characteristics, and accurate data support is provided for follow-up heat supply demand prediction and combustion control. And the opening degree of the combustion valve and the gas supply pressure are dynamically adjusted through the enhanced autonomous learning algorithm, dynamic adaptation of the gas flow and the heating requirement is achieved, and the gas utilization efficiency is remarkably improved. And dynamic division and adjustment of a heat exchange area are combined, so that the distribution characteristic of a heat flow path is optimized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a wall-mounted boiler constant temperature energy-saving control system based on enhanced autonomous learning and a SAC algorithm. Background Art

[0002] The field of intelligent control technology includes technical methods and tools widely used in industrial, civil and various equipment control. The core content of this technology is to achieve efficient operation of equipment or systems by sensing information such as equipment operating status and environmental parameters, and using control algorithms and actuators. Intelligent control technology is based on information acquisition, analysis and real-time decision-making, covering technical elements such as sensor application, control logic design and actuator coordination.

[0003] Among them, the constant temperature energy-saving control system for wall-mounted boilers refers to a control system based on real-time data acquisition and processing designed to meet the needs of temperature control and energy consumption optimization during the operation of wall-mounted boiler equipment. The system obtains information such as water temperature and ambient temperature during operation, adopts a control strategy based on reinforced autonomous learning algorithm, adjusts the combustion time and fire intensity, and thus achieves precise control of heat output.

[0004] The existing technology mainly relies on a single data source such as water temperature and ambient temperature to adjust the combustion time and fire intensity, which cannot fully reflect the dynamic change characteristics during the operation of the wall-mounted boiler, resulting in inaccurate adjustment of the gas valve opening and gas pressure. In different operating scenarios, the existing technology is difficult to flexibly respond to the rapid changes in heat demand, resulting in frequent over- or under-heating, reducing the overall gas utilization efficiency. During the heat exchange process, the existing technology failed to optimize the temperature difference and flow rate in different areas, resulting in uneven heat distribution, affecting the heat exchange efficiency and heating effect. In addition, the existing technology lacks a constant temperature control method based on dynamic simulation, and cannot achieve accurate evaluation of temperature control deviations in a variety of operating scenarios, which easily leads to unstable temperature control effects. This limitation is particularly prominent during high-load or low-load operation, further increasing energy consumption and reducing user experience. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and propose a wall-mounted boiler constant temperature energy-saving control system based on enhanced autonomous learning and SAC algorithm.

[0006] In order to achieve the above purpose, the present invention adopts the following technical scheme: A constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm includes: The gas flow optimization module collects the heating operation data of the wall-mounted boiler, classifies the heating operation data according to the change characteristics of multiple events, calculates the occurrence frequency of each event, analyzes the change of the operation data according to the frequency of event occurrence, and generates a gas flow data set; The dynamic combustion control module adopts the SAC algorithm in reinforcement learning, inputs the information of the gas flow data set into the strategy network of the SAC algorithm, and performs strategy network training. The gas energy saving and temperature control response of the strategy network are weighed by the entropy adjustment coefficient of the SAC algorithm to generate a combustion valve and pressure optimization parameter set; The heat exchange optimization and adjustment module optimizes and adjusts the internal heat flow distribution of the heat exchanger based on the combustion valve and pressure optimization parameter set and the operating state of the wall-mounted boiler heat exchanger to generate a heat exchange area optimization parameter set; The constant temperature energy-saving evaluation module predicts the current wall-mounted boiler heating demand based on the wall-mounted boiler heating operation data, combines the predicted heat demand with the temperature difference adjustment and flow rate optimization results in the heat exchange area optimization parameter set, refines the combustion valve opening and heat exchange path optimization parameters, and generates the wall-mounted boiler constant temperature energy-saving control optimization results.

[0007] As a further solution of the present invention, the step of calculating the occurrence frequency of each event is specifically: Collect the heating operation data of the wall-mounted boiler, including gas valve pressure, water temperature changes, and the target water temperature value set by the user. Use sensors to monitor and record the gas valve pressure, water temperature changes, and corresponding operation timestamps. Classify the heating operation data according to the change characteristics of multiple events such as combustion start, combustion maintenance, heating load changes, and combustion stop, and obtain a classified data set. Based on the classified data set, the formula is adopted: ; Calculate the Frequency of occurrence of similar events ; in, The first The number of occurrences of this type of event, The first The total duration of the event, The first The gas valve pressure change value of this type of event, is the reference pressure value of the gas valve, It is Adaptive adjustment coefficient for class events.

[0008] As a further solution of the present invention, the steps of acquiring the gas flow data set are specifically as follows: Based on the occurrence frequency of the events of combustion start, combustion maintenance, heating load change and combustion stop, the gas valve pressure fluctuation, the outlet water temperature change rate and the temperature difference change between the target water temperature and the outlet water temperature are analyzed, and the dynamic change characteristics of the gas flow rate triggered by the event are extracted to obtain the gas flow rate change analysis results; Based on the gas flow rate change analysis results, the gas flow rate ranges for combustion start, combustion maintenance, heating load change and combustion stop events are set, and the flow rate ranges are converted into time series data to generate a gas flow rate data set.

[0009] As a further solution of the present invention, the steps of performing strategy network training are specifically: Extract the gas valve pressure fluctuation, outlet water temperature change rate and temperature difference change between the target water temperature and the current outlet water temperature from the gas flow data set, construct the state and action range of the combustion valve opening adjustment amount and the combustion chamber gas supply pressure change value with the extracted information, and generate the gas flow dynamic characteristic information; Based on the dynamic characteristic information of the gas flow, all state records of gas valve pressure fluctuations, outlet water temperature change rate, and the temperature difference between the target water temperature and the current outlet water temperature, as well as action data records of the combustion valve opening adjustment amount and the gas supply pressure change value are used to adjust the control of the combustion valve opening and the gas supply pressure, and multiple rounds of optimization training are performed to obtain the trained control strategy network.

[0010] As a further solution of the present invention, the steps of obtaining the combustion valve and pressure optimization parameter set are specifically as follows: Based on the trained control strategy network, the formula is adopted: ;

[0011] Calculate the value of the reinforcement learning objective function , according to the reinforcement learning objective function value Evaluate the pros and cons of the current strategy and obtain the strategy evaluation results; in, Representation strategy The weighted average of all actions, Indicates in status Next action The accumulated reward value is Indicates in status Next select action The probability distribution of is the entropy adjustment coefficient, is the logarithm of the action probability used to calculate the entropy term of the policy; Based on the strategy evaluation results, the pressure fluctuation, temperature difference and temperature change rate are monitored under various fluctuation conditions, the combustion valve opening and pressure parameters are dynamically adjusted, and the adjusted parameter range is recorded to generate a combustion valve and pressure optimization parameter set.

[0012] As a further solution of the present invention, the step of obtaining the heat exchange area optimization parameter set is specifically: Based on the combustion valve and pressure optimization parameter set, combined with the operating state of the wall-mounted boiler heat exchanger, according to the influence of the combustion valve opening and the combustion chamber pressure change on the temperature difference distribution and fluid flow rate of the heat exchange area, the heat exchanger is divided into zones, and the temperature difference distribution range and flow rate target value of each zone are dynamically adjusted to generate the heat flow control zone division result; Based on the heat flow control area division results, the formula is adopted: ; Calculate the total heat exchange of the entire heat exchanger within the target time , in W, based on the total amount of heat exchange Dynamically adjust the heat flow path of each temperature zone to obtain the optimized parameter set of the heat exchange area; in, is the total number of regions, It is The temperature difference of the heat flow control area is It is The flow rate in the heat flow control area is It is The thermal resistance of a heat flow control area.

[0013] As a further solution of the present invention, the step of predicting the current heating demand of the wall-mounted boiler is specifically as follows: According to the heating operation data of the wall-mounted boiler, the formula is used: ; Calculate the heat required for heating demand in the future target time period , unit is kWh; in, is the time index, is the total sampling hours, It is Heat loss per hour, It is Output energy efficiency of the equipment per hour, It is Hourly heating area, is the thermal efficiency coefficient of the wall-mounted boiler; Based on the heat required for heating demand in the future target time period, the demand trend data for each time period is sorted out, the dynamic parameters of the combustion valve opening and the gas supply pressure are adjusted in combination with the operating period, and regional integration is performed to obtain the wall-mounted boiler heating demand forecast result.

[0014] As a further solution of the present invention, the steps for obtaining the optimization result of the constant temperature energy-saving control of the wall-mounted boiler are specifically as follows: Based on the prediction result of the heating demand of the wall-mounted boiler, the heat demand is decomposed into multiple areas, and the temperature difference adjustment range and flow rate optimization range of each area in the heat exchange area optimization parameter set are combined. Through dynamic simulation analysis of the operating status under multiple scenarios, it is verified whether the temperature control deviation of the corresponding area is controlled within the set range, and the optimization parameter set of each area is generated; Based on the optimization parameter set of each area, combined with the evaluated temperature control deviation and gas consumption rate, the adjustment range of the combustion valve opening and the gas supply pressure is refined, and a heat exchange path optimization sequence is generated for each area to obtain the optimization result of the constant temperature energy-saving control of the wall-mounted boiler.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, through real-time collection and multi-dimensional classification analysis of the wall-mounted boiler operation data, combined with the event triggering characteristics, gas flow data is generated, providing accurate data support for subsequent heating demand prediction and combustion control. The combustion valve opening and gas supply pressure are dynamically adjusted using the reinforced autonomous learning algorithm, which realizes the dynamic adaptation of gas flow and heating demand, and significantly improves the gas utilization efficiency. Combined with the dynamic division and adjustment of the heat exchange area, the distribution characteristics of the heat flow path are optimized, the heat exchange efficiency is improved and the temperature difference deviation range is reduced. By combining the heat demand of different areas with the dynamic adjustment of gas supply, the overall operation logic is optimized and the gas consumption rate is reduced. Simulating constant temperature control under various operation scenarios, by adjusting the gas flow and heat exchange path, the accuracy of the temperature control response is improved, and the refined valve opening and pressure adjustment logic ensure the stability of constant temperature operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a system flow chart of the present invention; Figure 2 A flow chart for calculating the frequency of occurrence of each event for the present invention; Figure 3 A flow chart for obtaining a gas flow data set according to the present invention; Figure 4 A flow chart for performing strategy network training for the present invention; Figure 5 A flow chart for obtaining a combustion valve and pressure optimization parameter set for the present invention; Figure 6 A flow chart for obtaining a heat exchange area optimization parameter set for the present invention; Figure 7 A flow chart of the present invention for predicting the current heating demand of a wall-mounted boiler; Figure 8 The present invention is a flow chart for obtaining optimization results of constant temperature energy-saving control of a wall-mounted boiler. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0019] See also Figure 1 The constant temperature energy-saving control system of the wall-mounted boiler based on enhanced autonomous learning and SAC algorithm includes: The gas flow optimization module collects the heating operation data of the wall-mounted boiler, classifies the heating operation data according to the change characteristics of multiple events, calculates the occurrence frequency of each event, analyzes the change of the operation data according to the frequency of event occurrence, and generates a gas flow data set; The dynamic combustion control module uses the SAC algorithm in reinforcement learning to input the information of the gas flow data set into the strategy network of the SAC algorithm and conduct strategy network training. The gas energy saving and temperature control response of the strategy network are weighed by the entropy adjustment coefficient of the SAC algorithm to generate the combustion valve and pressure optimization parameter set. The heat exchange optimization and adjustment module optimizes the heat flow distribution inside the heat exchanger based on the combustion valve and pressure optimization parameter set and combines the operating status of the wall-mounted boiler heat exchanger to generate the heat exchange area optimization parameter set; The constant temperature energy-saving evaluation module predicts the current heating demand of the wall-mounted boiler based on the heating operation data of the wall-mounted boiler, combines the predicted heat demand with the temperature difference adjustment and flow rate optimization results in the heat exchange area optimization parameter set, refines the combustion valve opening and heat exchange path optimization parameters, and generates the wall-mounted boiler constant temperature energy-saving control optimization results; The gas flow data set specifically includes the gas valve pressure fluctuation characteristics, the outlet water temperature change rate characteristics, and the temperature difference change characteristics between the target water temperature and the current outlet water temperature; the combustion valve and pressure optimization parameter set includes the combustion valve opening adjustment parameters and the combustion chamber gas supply pressure adjustment parameters; the heat exchange area optimization parameter set specifically includes the temperature difference distribution range of the heat flow control area, the flow rate adjustment target of the heat flow control area, and the heat flow path optimization parameters; the wall-mounted boiler constant temperature energy-saving control optimization results include temperature control deviation adjustment parameters, gas consumption rate optimization parameters, and heat exchange path optimization refinement parameters.

[0020] See also Figure 2 , the steps to calculate the frequency of occurrence of each event are as follows: Collect the heating operation data of the wall-mounted boiler, including gas valve pressure, water temperature changes, and the target water temperature value set by the user. Use sensors to monitor and record the gas valve pressure, water temperature changes, and corresponding operation timestamps. Classify the heating operation data according to the change characteristics of multiple events such as combustion start, combustion maintenance, heating load changes, and combustion stop, and obtain a classified data set. The heating operation data of the wall-mounted boiler is collected through monitoring equipment, including a pressure sensor installed on the gas valve to record the pressure value of the gas valve in real time. The range of the pressure sensor is 0-5MPa, and the resolution is 0.01MPa. At the same time, temperature sensors are set at the inlet and outlet of the water pipe to monitor the changes in the outlet and inlet temperatures. The measurement range of the temperature sensor is 0-100°C, and the resolution is 0.1°C. The pressure and temperature data collected by the sensor are transmitted to the storage database through the data acquisition module. At the same time, the target water temperature value set by the user is manually input and used as a fixed parameter in the data set. The collected operation data is then classified according to the operating status of the wall-mounted boiler. By detecting the gas Significant changes in the gas valve pressure (for example, the pressure value increases from 0.2MPa to 1.0MPa with a time span of less than 5 seconds) are marked as combustion start events. The combustion state stabilization time (pressure fluctuation is less than 0.02MPa and the duration is greater than 10 minutes) is used to determine the combustion maintenance event. The heating load change is analyzed by the water temperature change rate. For example, if the water temperature change rate increases from 2°C / minute to 4°C / minute, it is classified as a heating load change event. The combustion stop event is marked when the gas valve pressure drops below 0.2MPa. Based on the above classification results, the operating data are stored in independent data sets for combustion start, combustion maintenance, heating load change and combustion stop.

[0021] Based on the classified data set, the formula is used: ; Calculate the Frequency of occurrence of similar events (times / hour); in, The first The number of occurrences of such events is counted through the event detection algorithm, and the number of switching times of combustion start, combustion maintenance, heating load change and combustion stop is counted. For example, by monitoring the pressure mutation, the specific time point of the combustion start event can be marked, and then the total number of events can be counted. The first The total duration of this type of event (hours) is obtained by recording the timestamp difference of the boiler operation log. For example, if the operation time of a period is 10 hours, it will be recorded as 10 hours. The first The gas valve pressure change value (MPa) of such events is obtained by calculating the difference between the maximum pressure and the minimum pressure in the event, for example, if the maximum pressure is 1.2MPa and the minimum pressure is 0.8MPa in a certain period of time, then MPa, It is the reference pressure value (MPa) of the gas valve, which is used to normalize the gas pressure change value. It is determined by the factory calibration of the equipment. For example, the standard gas pressure value set in the equipment calibration is 1.0MPa. It is The adaptive adjustment coefficient (dimensionless) of the event is used to correct the influence of pressure change on the event occurrence rate. It is obtained by fitting experimental data. For example, the influence of gas valve pressure change on the occurrence frequency of different events is observed through multiple experiments. A representative pressure change range is selected, such as from 0.2MPa to 1.2MPa. The corresponding event occurrence frequency in each pressure change range is recorded respectively, and a data set of the relationship between pressure change and event occurrence frequency is established. The data set is used for fitting analysis. The fitting process is divided into the following steps: Under a variety of typical operating conditions, including the influence of different user target water temperature setting values ​​and external ambient temperature, the gas valve pressure change and its corresponding event occurrence frequency are recorded to generate a A group of experimental data was collected, and the collected pressure change data was normalized to between 0 and 1 to ensure the consistency of the magnitude of the data during the fitting process. According to the distribution pattern of the experimental data, a polynomial fitting model was selected for analysis. In the fitting process, the pressure change value was used as the independent variable, and the event frequency was used as the dependent variable. The parameters of the fitting model were solved by the least squares method, and the fitting results were compared with the experimental data to evaluate the accuracy of the fitting model. The order of the fitting model was adjusted to achieve a higher degree of fit. The fitting model was verified using an independent experimental data set. If the fitting error was less than the preset threshold (for example, 5%), the fitting result of the adaptive adjustment coefficient was confirmed, and the adaptive adjustment coefficient was fitted to be 0.5.

[0022] Assume that the observation period is 10 hours. If the operation of the wall-mounted boiler is as follows, the combustion start event: (obtained by counting the time points of pressure mutation). Hours (calculated from timestamp records). MPa (recorded by pressure sensor). MPa (equipment calibration value). (Fitted by experiment). Calculation: Times / hour (rounded); combustion maintenance events: (Obtained by recording the changing points during the period of stable combustion). Hour. MPa. MPa. .calculate: Times / hour (rounded); Heating load change events: . Hour. MPa. MPa. .calculate: Times / hour (rounded); Burning stop event: . Hour. MPa. MPa. .calculate: times / hour (rounded); the result shows that the average frequency of combustion start events is 5 times per hour. The average frequency of combustion maintenance events is 10 times per hour. The average frequency of heating load change events is 2 times per hour. The average frequency of combustion stop events is 3 times per hour. By normalizing the pressure change value, the units of physical quantities in the formula are consistent. The introduction of the adaptive adjustment coefficient further improves the practical application value of the calculation results, enabling it to dynamically reflect the frequency change characteristics under different combustion events.

[0023] See also Figure 3 ,The specific steps for obtaining the gas flow data set are: Based on the frequency of combustion start-up, combustion maintenance, heating load change and combustion stop events, the gas valve pressure fluctuation, outlet water temperature change rate and the temperature difference between the target water temperature and the outlet water temperature are analyzed, and the dynamic change characteristics of the gas flow rate triggered by the event are extracted to obtain the gas flow rate change analysis results; The combustion start event occurs 5 times per hour, which means it is triggered once every 12 minutes. Combined with the actual monitoring data, the valve pressure rises significantly at each start, with a fluctuation range of 0.5 to 0.8MPa, and the pressure rise rate is an average of 0.1MPa / second. The outlet water temperature rises rapidly after startup, rising by about 2°C per minute. The temperature difference between the target water temperature and the actual outlet water temperature at this stage is initially large, about 5°C, and the gas flow rate needs to be increased quickly to narrow the temperature difference. The average frequency of combustion maintenance events is 10 times per hour, that is, the maintenance conditions are re-evaluated every 6 minutes. At this stage, the valve pressure fluctuation range is small, 0.1 to 0.2MPa, the pressure fluctuation rate drops to 0.02MPa / second, and the outlet water temperature change rate tends to be stable, maintained within 0.5°C / minute. The temperature difference between the target water temperature and the outlet water temperature gradually decreases to 1°C or less. At this time, the stability of the gas flow is the key and should be maintained at a medium level to avoid fluctuations. The heating load change event occurs 2 times per hour, which means it is triggered once every 30 minutes. The valve pressure fluctuation of this event is quite violent, ranging from 0.4 to 0.6MPa, and the pressure rise rate is about 0.08MPa / second. At this time, the outlet water temperature change rate increases rapidly to 1.5 to 2°C per minute, and the temperature difference between the target water temperature and the outlet water temperature is rapidly reduced to less than 2°C at the beginning of heating. At this stage, the gas flow rate needs to be quickly increased to a high level to meet the load change requirements. The combustion stop event occurs an average of 3 times per hour, that is, it is triggered once every 20 minutes. During the stop phase, the valve pressure drops rapidly to below 0.2MPa, with a drop rate of 0.2MPa / second, and the outlet water temperature change rate slows down rapidly to less than 0.3°C per minute. The temperature difference between the target water temperature and the outlet water temperature gradually increases to more than 4°C, and the gas flow rate needs to be quickly reduced to avoid gas waste. Based on the above analysis, the dynamic change characteristics of the gas flow rate under the triggering of each event are as follows: the combustion start-up and heating load change events require a rapid increase in gas flow rate to meet the temperature difference and load change requirements, the combustion maintenance phase requires a stable gas flow rate, and the combustion stop phase requires a rapid reduction in flow rate to save gas consumption. By analyzing the relationship between frequency, pressure fluctuation and temperature difference changes.

[0024] Based on the results of the gas flow rate change analysis, the gas flow rate ranges for combustion start, combustion maintenance, heating load change, and combustion stop events are set, and the flow rate ranges are converted into time series data to generate a gas flow data set; By analyzing the dynamic change characteristics of the four events of combustion start, combustion maintenance, heating load change and combustion stop, the event frequency is matched with the adjustment range of the gas flow rate. The specific setting is that the combustion start event is 5 times per hour. When the event occurs, the gas flow rate is rapidly increased to between 90% and 100% of the maximum flow rate. The combustion maintenance event is 10 times per hour, and the flow rate is stabilized between 40% and 60% of the maximum flow rate. When the heating load change event occurs twice per hour, the gas flow rate is rapidly increased to 70% to 90% of the maximum flow rate. When the combustion stop event occurs three times per hour, the gas flow rate is gradually reduced from the current value to less than 20% of the maximum flow rate. The above-set flow range is converted into a flow data sequence matching the time. The time series is generated with a recording interval of 5 seconds, and matched and verified with the real-time detected gas valve pressure data to ensure that the flow change conforms to the dynamic adjustment law and complete the generation and storage of the gas flow data set.

[0025] See also Figure 4 , the specific steps for policy network training are: Extract the gas valve pressure fluctuation, outlet water temperature change rate, and temperature difference between the target water temperature and the current outlet water temperature from the gas flow data set, and use the extracted information to construct the state and action range of the combustion valve opening adjustment amount and the combustion chamber gas supply pressure change value to generate gas flow dynamic characteristic information; The gas valve pressure fluctuation, outlet water temperature change rate, and temperature difference between the target water temperature and the current outlet water temperature are extracted from the generated gas flow data set. The specific method is to segment the gas flow data, calculate the difference between the maximum and minimum gas valve pressure in each time period to characterize the pressure fluctuation range, and combine the water temperature change data to calculate the temperature change per unit time to obtain the outlet water temperature change rate. For example, the pressure fluctuation in a certain period is 0.4MPa, and the temperature change rate is 1.8°C per minute. The absolute value of the difference between the target water temperature and the current outlet water temperature is taken to characterize the temperature difference change, and the extracted information is input into the SAC in reinforcement learning. Algorithm strategy network, defines the state space of the strategy network, including the current gas valve pressure fluctuation, outlet water temperature change rate and temperature difference change value. The action space is the combustion valve opening adjustment amount and the change value of the combustion chamber gas supply pressure. In the state space, the gas valve pressure fluctuation range is defined as 0 to 1.0MPa, the outlet water temperature change rate range is defined as 0 to 5°C / minute, and the temperature difference change range is defined as 0 to 10°C. In the action space, the combustion valve opening adjustment amount is set to -10% to +10%, and the combustion chamber gas supply pressure change value range is -0.5MPa to +0.5MPa, providing clear input and adjustment range for subsequent training.

[0026] Based on the dynamic characteristic information of gas flow, the control of the combustion valve opening and gas supply pressure is adjusted using all state records of gas valve pressure fluctuation, outlet water temperature change rate, and temperature difference between target water temperature and current outlet water temperature, as well as action data records of combustion valve opening adjustment and gas supply pressure change value, and multiple rounds of optimization training are performed to obtain the trained control strategy network; The SAC algorithm in reinforcement learning is adopted. The strategy function and value function are derived from the design and training process of deep neural network. The strategy function is used to generate the optimal action under a given state. Its function is to fit the conditional probability distribution of state and action through neural network, so as to select the optimal combustion valve opening and gas supply pressure adjustment value in the complex wall-mounted boiler operation state. The value function is used to evaluate the value of each state. Its function is to fit the value function of the current state through deep neural network to provide evaluation basis for the strategy function. The specific value includes the following aspects: Energy saving value: state evaluation can quantify the gas consumption level of the current combustion state and judge whether adjusting the gas valve pressure and opening can effectively reduce gas consumption; Temperature control stability: state value evaluates the deviation between the target water temperature and the actual water outlet temperature, and obtains a higher value through a state with a lower temperature difference, ensuring the accuracy and response speed of temperature control; Combustion efficiency: by evaluating the pressure fluctuation range and the change value of the gas supply pressure, it is determined whether the combustion process is in the predetermined range to reduce the occurrence of incomplete combustion; System responsiveness: evaluate the rate of change of the state after the action is executed to ensure the real-time and dynamic adjustment of the combustion valve and pressure; Safety: identify the state of abnormal pressure or temperature changes, and prompt the safety risks in the strategy adjustment process. The policy function is optimized by minimizing the KL divergence between the action output by the policy function and the target action. The KL divergence is used to measure the difference between two probability distributions. The optimization process ensures that the action output by the policy gradually converges to the optimal action. The value function is updated by minimizing the difference between the predicted value and the actual reward plus the discounted future reward. The actual reward is calculated by monitoring the difference between the target water temperature and the actual water outlet temperature. The discounted future reward value is obtained based on the prediction of the future state change trend. The parameters of the policy function and the value function are randomly initialized at the beginning of training. The experience replay buffer is used to store the historical state-action-reward-next state four-tuple data generated by the policy function. The experience replay buffer The buffer is used to break the time correlation of the data. Small batches of data are randomly extracted from the buffer each time for training. The reward function is designed as a negative correlation value between the absolute value of the temperature difference between the target water temperature and the actual water outlet temperature. Its role is to reward the superiority of the strategy output by reducing the temperature difference. The value function uses the gradient descent method to adjust the weight parameters of the neural network so that its estimate of the state value is closer to the actual reward value. The output action of the strategy function is corrected and improved through the feedback of the value function to continuously optimize the combustion valve opening adjustment amount and the gas supply pressure change value. After multiple rounds of training, an optimization strategy network that can accurately control the combustion state of the wall-mounted boiler is generated, and the network is ultimately used to dynamically adjust the operating parameters.

[0027] See also Figure 5 , the steps for obtaining the combustion valve and pressure optimization parameter set are as follows: Based on the trained control strategy network, the formula is: ; Calculate the value of the reinforcement learning objective function , according to the reinforcement learning objective function value Evaluate the quality of the current strategy. The larger the value, the better the strategy. The range is assumed to be [8, 12] during the training phase. A value lower than 8 indicates that the strategy is not effective enough for gas energy saving or temperature control. A value higher than 12 indicates that the strategy may over-optimize one goal and ignore another goal. The strategy evaluation result is obtained. in, Is the expectation symbol, indicating strategy The weighted average of all possible actions is obtained by sampling the historical data stored in the experience replay buffer. For example, 100 state-action pairs are randomly selected from the buffer each time, the distribution probability of these samples is calculated and weighted averaged, and finally the expected value is approximated. is the state and action value function, which means that in state Next action The accumulated reward value is obtained by deep neural network fitting. For example, input: state variable , including gas valve pressure fluctuation (MPa), outlet water temperature change rate (°C / minute) and temperature difference change (°C), and the data is collected in real time through sensors. For example, in the current state, the pressure fluctuation is 0.3MPa, the temperature change rate is 2.5°C / minute, and the temperature difference is 3°C. A three-layer fully connected neural network is used. The input layer contains 3 nodes (corresponding to state variables), the hidden layer is set to 64 nodes, the activation function uses ReLU, and the output layer is a single node (representing the reward value). The network is trained by state-action-reward data collected from the environment. The goal is to minimize the error between the predicted value and the actual reward. The mean square error is used as the loss function, and Adam is selected as the optimization algorithm. A batch size of 64 is used during training, and the network parameters are updated at each iteration. Given an input state and actions After that, the network predicts the corresponding cumulative reward value, such as the network output . is the policy function, which means that in the state Next select action The probability distribution of is obtained by deep neural network fitting, for example, input: state variable , which is the same as the state-action value function, including pressure fluctuations, temperature change rate, and temperature difference changes. A three-layer fully connected network similar to the state-action value function is used, but the number of nodes in the output layer is equal to the number of action spaces (for example, the number of combinations of the two actions of combustion valve opening adjustment and gas supply pressure change). The output layer activation function is Softmax, and a probability distribution is generated. Through the policy gradient method optimization, the goal is to maximize the expected reward value of the policy function on all state-action pairs. The loss function is the negative objective function value, and the optimization algorithm still chooses Adam. Given a state After that, the network outputs the probability distribution of each action, such as . is the entropy adjustment coefficient, the control strategy is exploratory, and the range is set to [0.1, 0.5]. The acquisition process is as follows: the initial value is set to 0.3, and its contribution to the objective function is verified through multiple rounds of experiments. The changing trends of gas consumption and objective function values ​​under different entropy adjustment coefficients are observed. When it is too large, the strategy may over-explore and waste gas; when If it is too small, the strategy may be too conservative and affect the temperature control response. Through experimental iterative adjustment, when the value of the entropy adjustment coefficient leads to the objective function value in the range of [8, 12] and the energy saving effect is optimal, its value range is determined. It is the logarithm of the action probability used to calculate the entropy term of the strategy, which represents the exploratory nature of the strategy and is obtained by directly calculating the logarithm of the action probability.

[0028] If the current status :The gas valve pressure fluctuation is 0.3MPa, the outlet water temperature change rate is 2.5°C / minute, and the temperature difference is 3°C. Strategy function The output action is that the combustion valve opening increases by 5%, the gas supply pressure increases by 0.2MPa, and the action probability . Cumulative Reward Value , obtained by evaluating the combustion state and temperature control effect under simulated environment. Entropy regulation coefficient , determined according to the experimental adjustment process. Substitute the above parameters into the formula: ; The results show that the objective function value is 10.0669, which is within the reasonable range of [8, 12], indicating that the strategy is superior in the current state and balances the gas energy saving demand and temperature control response capability. The target value is in the middle of the interval, indicating that the strategy network has a good balance between energy saving and response and does not need further adjustment.

[0029] Based on the strategy evaluation results, the pressure fluctuation, temperature difference and temperature change rate are monitored under various fluctuation conditions, the combustion valve opening and pressure parameters are dynamically adjusted, and the adjusted parameter range is recorded to generate the combustion valve and pressure optimization parameter set; Dynamically adjust the combustion valve opening and pressure parameters under various fluctuation conditions. According to paragraph 1, the objective function value is 10.0669 and is within the reasonable range of [8, 12]. This shows that the strategy has balanced the gas energy saving demand and temperature control response capability under the current state. Based on this, the specific adjustment steps are as follows: Monitoring and analysis of the initial state: Real-time monitoring of gas valve pressure fluctuations, outlet water temperature change rate, and the temperature difference between the target water temperature and the actual outlet water temperature. According to the data collected by the pressure sensor and the temperature sensor, analyze the range of change of the current state variables. For example, the current pressure fluctuation is 0.3MPa, the temperature change rate is 2.5°C / minute, and the temperature difference is 3°C. Action selection and strategy output: Use the objective function value to feed back to the strategy network, and execute the strategy function under the current state. The generated actions include the adjustment of the combustion valve opening and the gas supply pressure. For example, if the valve opening increases by 5%, the pressure increases by 0.2MPa. When the action is output, the objective function value is combined to verify whether the action is within a reasonable range. Dynamic adjustment logic: Classify the pressure fluctuation range and temperature difference change: When the pressure fluctuation range is large (for example, above 0.5MPa) and the temperature difference changes by more than 5°C, the strategy network tends to output the action of increasing the gas supply pressure and valve opening to ensure a rapid increase in heat supply. When the pressure fluctuation is small (for example, below 0.2MPa) and the temperature difference changes by less than 2°C, the strategy network outputs the action of reducing the gas supply pressure or maintaining the current valve opening to avoid excessive gas consumption. Parameter optimization and recording: According to the action output by the strategy network, the actual adjusted combustion valve opening and gas supply pressure are recorded as the optimized parameter set. For example, the combustion valve opening range is adjusted to [40%, 60%], and the gas supply pressure range is adjusted to [0.8MPa, 1.0MPa]. These optimized parameters are stored in the control module for subsequent dynamic use. In the case of various fluctuations, such as when users frequently adjust the target water temperature or when the external temperature changes significantly, the strategy network is corrected online through the objective function value, and the weight parameters of the strategy network are updated according to the real-time monitored state variables, so as to further optimize the combustion valve opening and gas pressure adjustment logic and form a more refined parameter optimization set. Through the above process, the generated combustion valve and pressure optimization parameter set can not only dynamically respond to various fluctuations, but also find a balance between energy saving and temperature control response capabilities, ensuring the stability and efficiency of the wall-mounted boiler operation.

[0030] See also Figure 6 , the specific steps for obtaining the heat exchange area optimization parameter set are: Based on the combustion valve and pressure optimization parameter set, combined with the operating status of the wall-mounted boiler heat exchanger, the heat exchanger is divided into zones according to the influence of the combustion valve opening and the combustion chamber pressure change on the temperature difference distribution and fluid flow rate in the heat exchange area, and the temperature difference distribution range and flow rate target value of each area are dynamically adjusted to generate the heat flow control area division result; First, the interior of the heat exchanger is divided into regions. By extracting the combustion valve opening and combustion chamber pressure change data, combined with the heat flow parameters monitored by the temperature sensor and flow rate sensor, the initial temperature difference and flow rate distribution of each region are clarified. For example, the temperature difference range of region 1 is [10°C, 15°C], and the flow rate range is [0.8m / s, 1.2m / s], and the temperature difference range of region 2 is [5°C, 10°C], and the flow rate range is [0.5m / s, 1.0m / s]. Then, each region is taken as a neighborhood unit, and the temperature difference and flow rate target values ​​are adjusted using dynamic neighborhood search rules. The specific rules are as follows: neighborhood search initialization: set the initial search range, and limit the target value of the temperature difference distribution in the region to ±20% of the current temperature difference. For example, the target temperature difference range of region 1 is initially set to [12°C, 18°C], and the flow rate adjustment target range is set to ±15%, that is, [0.92m / s, 1.38m / s]. Dynamic search and adjustment: gradually adjust the target value range, analyze the changing trend of heat conduction efficiency according to the iterative results of neighborhood search, for example, when the temperature difference target of area 1 is adjusted to [13°C, 17°C], the heat conduction efficiency is improved by 5%, continue to refine the search within this range, and gradually narrow the target value range. Determine the termination condition: take the change amplitude of heat conduction efficiency as the termination condition, when the efficiency improvement amplitude is less than the set threshold (for example, <1%) after adjusting the target value, determine the current target value range as the final optimization value, for example, the final temperature difference target value of area 1 is [13.5°C, 16.5°C], and the flow rate target value is [1.0m / s, 1.2m / s]. Through the above process, the dynamic neighborhood search algorithm defines specific temperature difference distribution target values ​​and flow rate adjustment targets for each heat flow control area in the step-by-step optimization, and these targets can more accurately guide the adjustment of the heat flow path.

[0031] Based on the heat flow control area division results, the formula is used: ; Calculate the total heat exchange of the entire heat exchanger within the target time , in W, based on the total amount of heat exchange Dynamically adjust the heat flow path of each temperature zone to obtain the optimized parameter set of the heat exchange area; in, Indicates the accumulation of all heat flow control areas in the heat exchanger. It is The temperature difference of each heat flow control area, in °C, is obtained by monitoring the temperature difference between the inlet and outlet of the heat exchanger through the temperature sensor. For example, if the inlet and outlet temperatures of area 1 are 80°C and 65°C respectively, then , It is The flow rate of each heat flow control area is in m / s. The velocity of the fluid is measured by the flow velocity sensor. For example, the flow velocity of area 1 is 1.2 m / s. It is The thermal resistance of each heat flow control area, in K / W, is calculated by combining the thermal conductivity of the material and the heat exchange area. The calculation formula is: , is the length of the heat flow path (in m), is the thermal conductivity (unit: W / (m·K)), is the heat exchange area (in m²), Represents each independent heat flow control area. The heat exchanger is divided into multiple areas, such as area 1, area 2 and area 3, through the dynamic neighborhood search algorithm. is the total number of heat flow control areas, which are automatically divided by the dynamic neighborhood search algorithm. For example, three areas are divided by optimizing the temperature difference distribution and flow velocity distribution.

[0032] For example, the thermal resistance is 0.01K / W. Assume that the heat exchanger is divided into three heat flow control areas, and the area parameters are as follows: Area 1: , , . Area 2: , , . Area 3: , , . Substitute the parameters of each region into the formula: ; The results show that the total heat exchange efficiency of the heat exchanger under the current optimization conditions is 557W. This value is used to evaluate the effect of the optimization of the heat flow control area to ensure that the heat exchanger can achieve more efficient heat conduction after dynamic adjustment. value, such as 557W, combined with the temperature difference of each heat flow control area , flow rate and thermal resistance The specific distribution of heat in each area is determined by the specific distribution of heat in each area. For example, the contribution of area 1 is 360W, area 2 is 125W, and area 3 is 72W. Compare the heat conduction efficiency of each area with the preset target value to analyze the high-efficiency area and the low-efficiency area. For example, if the temperature difference target of area 1 is 15°C but the actual value is 13°C, it means that the heat flow path of area 1 needs to be further optimized; if the flow rate of area 2 is only 1.0m / s, which is lower than the target value of 1.2m / s, it is necessary to increase the regional flow rate. For areas with low temperature difference, increase the heat flow input to increase the temperature difference. For example, by adjusting the combustion valve opening to introduce more heat into area 1, the temperature difference is increased to the target value. For areas with low flow rate, optimize the fluid flow rate. For example, by adjusting the inlet and outlet pressure difference, the flow rate is within the target range. Use the dynamic neighborhood search algorithm to further refine the adjustment and gradually narrow the adjustment range, for example, adjust the temperature difference target from [13°C, 17°C] to [14°C, 16°C] in area 1. Recalculate the heat exchange amount of each area according to the adjusted heat flow path and verify the adjusted total heat exchange amount Whether there is any improvement. The dynamically adjusted parameters of each area are recorded as an optimized parameter set, including the temperature difference target value, flow rate target value and thermal resistance value of each area. See also Figure 7 , the specific steps for predicting the current heating demand of the wall-mounted boiler are: According to the heating operation data of the wall-mounted boiler, the formula is used: ; Calculate the heat required for heating demand in the future target time period , unit is kWh; in, Indicates that each hour in the target time period is accumulated. is the time index, is the total number of sampling hours. For example, if a sampling period is 12 hours, then , It is The hourly heat loss, in kWh, is calculated using the boiler monitoring data and the building heat dissipation coefficient. The formula is: , It is the building thermal conductivity coefficient, in W / (m²·K), obtained by experimentally measuring the thermal conductivity of wall and window materials. For example, the building thermal conductivity coefficient is 0.8W / (m²·K). is the heating area in m², is the temperature difference between indoor and outdoor, in K, collected in real time by temperature sensors. It is The output energy efficiency of the hourly device is in kWh. The heat generated by the gas combustion of the device is monitored by the wall-mounted boiler sensor. For example, the output energy efficiency is 0.9kWh. It is The effective area of ​​hourly heating is in m², which is set according to the actual heating area of ​​the room. For example, if the heating area in a certain period is 80 m², It is the thermal efficiency coefficient of the wall-mounted boiler. It is dimensionless and represents the efficiency of converting unit gas into heat. It is obtained through experiments or equipment calibration. For example, the thermal efficiency is 0.92 (92%).

[0033] If the sampling period is 12 hours ( ), the hourly heat loss of the building is kWh, the equipment output energy efficiency per hour is kWh, heating area The thermal efficiency of the wall-mounted boiler is . Substituting the parameters into the formula: ; Calculate the calorie requirement at a single time point: ; Accumulate all time points: ; The results show that the calculated heating demand forecast value is 313.08kWh, which represents the total gas heating capacity required to provide sufficient heat for a room with a heating area of ​​80m² within 12 hours.

[0034] Based on the heat required for heating demand in the future target time period, the demand trend data for each time period is sorted out, the dynamic parameters of the combustion valve opening and gas supply pressure are adjusted in combination with the operating time period, and regional integration is performed to obtain the forecast results of the heating demand of the wall-mounted boiler; Arrange the calculated hourly calorie demand values ​​in chronological order, for example, the individual predicted values ​​from hour 1 to hour 12 are , forming a set of heat demand sequences, and summarizing all data for subsequent analysis. According to the changing trend of hourly heat demand data, identify peak and valley demand periods. For example, the demand is high in the early stage of heating (1st to 3rd hour), and then tends to stabilize. The data is visualized through trend charts to clarify the demand characteristics of different time periods. Adjust the equipment operating parameters in combination with the heat demand trend, such as increasing the combustion valve opening (adjusted to 60%-80%) and gas supply pressure (increased to 1.2-1.5MPa) during peak hours, and reducing the valve opening (adjusted to 40%-50%) and pressure (reduced to 0.8-1.0MPa) during valley hours to ensure gas utilization and heating stability. For scenarios with a large heating area, the total heating demand is divided by region, for example, the demand for region 1 is 120kWh, the demand for region 2 is 100kWh, and the demand for region 3 is 93.08kWh. The demand of each region is further optimized in combination with its independent heating parameters. Integrate all forecast data to generate an optimization plan for the operation of the wall-mounted boiler, such as clarifying parameters such as valve opening, gas supply pressure, combustion time, etc. in each time period, and input these optimization data into the control module for real-time adjustment of the operating status of the wall-mounted boiler.

[0035] See also Figure 8 The specific steps for obtaining the optimization results of constant temperature energy-saving control of wall-mounted boilers are as follows: Based on the prediction results of heating demand of wall-mounted boilers, the heat demand is decomposed into multiple areas. Combined with the temperature difference adjustment range and flow rate optimization range of each area in the heat exchange area optimization parameter set, the operation status under multiple scenarios is analyzed through dynamic simulation to verify whether the temperature control deviation of the corresponding area is controlled within the set range, and the optimization parameter set of each area is generated; The predicted heat demand is combined with the temperature difference adjustment and flow rate optimization results in the heat exchange area optimization parameter set. First, according to the calculated heat demand prediction value of 313.08kWh, it is decomposed into the heat demand of each area. For example, the demand of area 1 is 120kWh, and the corresponding target temperature difference range is [12°C, 15°C], the demand of area 2 is 100kWh, and the target temperature difference range is [10°C, 12°C], and the demand of area 3 is 93.08kWh, and the target temperature difference range is [8°C, 10°C]. At the same time, the flow rate target value is combined for optimization. For example, the flow rate range of area 1 is [1.1m / s, 1.3m / s], the flow rate range of area 2 is [1.0m / s, 1.2m / s], and the flow rate range of area 3 is [0.9m / s, 1.1m / s]. The specific process of dynamic simulation analysis is as follows: Under the initial conditions of the simulation, according to the predicted value of heat demand and real-time monitoring data, the initial opening of the gas valve is set to 50%, the gas supply pressure is set to 1.0MPa, and the initial value of heat distribution in each area is evenly distributed, for example, 105kWh is allocated to area 1, area 2 and area 3 respectively. During the simulation process, the gas supply parameters and heat exchange paths of each area are adjusted in real time to monitor whether the temperature difference change value reaches the target range. For example, if the actual temperature difference of area 1 is lower than 12°C, the combustion valve opening is increased to 60%, and the gas pressure is adjusted to 1.2MPa to observe whether the temperature difference change enters the target range [12°C, 15°C]. Simulation analysis is performed for peak and valley periods respectively. During the peak period, the gas consumption rate is set to 0.6m³ / h, and the combustion valve opening fluctuates between [60%, 80%]. During the valley period, the gas consumption rate is set to 0.4m³ / h, and the combustion valve opening fluctuates between [40%, 60%]. The corresponding temperature control deviation is monitored in real time to be controlled within ±0.5°C. The temperature difference, flow rate, gas consumption rate and other data of each area are integrated to verify the temperature control stability and gas energy saving level in different scenarios. For example, by comparison, it is found that when operating within the target temperature difference range of area 1, the gas consumption rate is 15% lower than when it is not optimized. Through the above dynamic simulation analysis process, the optimized parameter set of each area in different operating scenarios is generated, and the performance of the wall-mounted boiler in constant temperature control effect and energy saving level is comprehensively evaluated based on these parameters.

[0036] Based on the optimized parameter set of each area, combined with the evaluated temperature control deviation and gas consumption rate, the adjustment range of the combustion valve opening and the gas supply pressure is refined, and a heat exchange path optimization sequence is generated for each area to obtain the optimization result of the constant temperature energy-saving control of the wall-mounted boiler; According to the results of the evaluated temperature control deviation of ±0.5°C and the gas consumption rate in the range of [0.4m³ / h, 0.6m³ / h], the combustion valve opening and heat exchange path optimization parameters are refined. The combustion valve opening in area 1 is refined to [60%, 80%], the combustion valve opening in area 2 is [50%, 70%], and the combustion valve opening in area 3 is [40%, 60%]. At the same time, the pressure optimization range in area 1 is [1.1MPa, 1.3MPa], the pressure range in area 2 is [1.0MPa, 1.2MPa], and the pressure range in area 3 is [0.9MPa, 1.1MPa]. Combined with the optimized heat exchange path parameters, the target temperature difference range for area 1 is [12°C, 15°C], for area 2 is [10°C, 12°C], and for area 3 is [8°C, 10°C]. An optimized heat exchange path sequence is generated in each area, and the applicability of the optimized parameters is further verified through dynamic simulation. Finally, the optimization result of constant temperature energy-saving control of wall-mounted boiler is formed, providing efficient data support for equipment operation.

[0037] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm, characterized in that: The system comprises: The gas flow optimization module collects the heating operation data of the wall-mounted boiler, classifies the heating operation data according to the change characteristics of multiple events, calculates the occurrence frequency of each event, analyzes the change of the operation data according to the frequency of event occurrence, and generates a gas flow data set; The dynamic combustion control module adopts the SAC algorithm in reinforcement learning, inputs the information of the gas flow data set into the strategy network of the SAC algorithm, and performs strategy network training. The gas energy saving and temperature control response of the strategy network are weighed by the entropy adjustment coefficient of the SAC algorithm to generate a combustion valve and pressure optimization parameter set; The heat exchange optimization and adjustment module optimizes and adjusts the internal heat flow distribution of the heat exchanger based on the combustion valve and pressure optimization parameter set and the operating state of the wall-mounted boiler heat exchanger to generate a heat exchange area optimization parameter set; The constant temperature energy-saving evaluation module predicts the current wall-mounted boiler heating demand based on the wall-mounted boiler heating operation data, combines the predicted heat demand with the temperature difference adjustment and flow rate optimization results in the heat exchange area optimization parameter set, refines the combustion valve opening and heat exchange path optimization parameters, and generates the wall-mounted boiler constant temperature energy-saving control optimization results.

2. The wall-mounted boiler constant temperature energy-saving control system based on enhanced autonomous learning and SAC algorithm according to claim 1 is characterized in that: The steps of calculating the occurrence frequency of each event are specifically as follows: Collect the heating operation data of the wall-mounted boiler, including gas valve pressure, water temperature changes, and the target water temperature value set by the user. Use sensors to monitor and record the gas valve pressure, water temperature changes, and corresponding operation timestamps. Classify the heating operation data according to the change characteristics of multiple events such as combustion start, combustion maintenance, heating load changes, and combustion stop, and obtain a classified data set. Based on the classified data set, the formula is adopted: ； Calculate the Frequency of occurrence of similar events ; in, The first The number of occurrences of this type of event, The first The total duration of the event, The first The gas valve pressure change value of this type of event, is the reference pressure value of the gas valve, It is Adaptive adjustment coefficient for class events.

3. The constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm according to claim 2 is characterized in that: The steps for obtaining the gas flow data set are specifically as follows: Based on the occurrence frequency of the events of combustion start, combustion maintenance, heating load change and combustion stop, the gas valve pressure fluctuation, the outlet water temperature change rate and the temperature difference change between the target water temperature and the outlet water temperature are analyzed, and the dynamic change characteristics of the gas flow rate triggered by the event are extracted to obtain the gas flow rate change analysis results; Based on the gas flow rate change analysis results, the gas flow rate ranges for combustion start, combustion maintenance, heating load change and combustion stop events are set, and the flow rate ranges are converted into time series data to generate a gas flow rate data set.

4. The constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm according to claim 3 is characterized in that: The steps of performing strategy network training are specifically as follows: Extract the gas valve pressure fluctuation, outlet water temperature change rate and temperature difference change between the target water temperature and the current outlet water temperature from the gas flow data set, construct the state and action range of the combustion valve opening adjustment amount and the combustion chamber gas supply pressure change value with the extracted information, and generate the gas flow dynamic characteristic information; Based on the dynamic characteristic information of the gas flow, all state records of gas valve pressure fluctuations, outlet water temperature change rate, and the temperature difference between the target water temperature and the current outlet water temperature, as well as action data records of the combustion valve opening adjustment amount and the gas supply pressure change value are used to adjust the control of the combustion valve opening and the gas supply pressure, and multiple rounds of optimization training are performed to obtain the trained control strategy network.

5. The constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm according to claim 4 is characterized in that: The steps for obtaining the combustion valve and pressure optimization parameter set are specifically as follows: Based on the trained control strategy network, the formula is adopted: ； Calculate the value of the reinforcement learning objective function , according to the reinforcement learning objective function value Evaluate the pros and cons of the current strategy and obtain the strategy evaluation results; in, Representation strategy The weighted average of all actions, Indicates in status Next action The accumulated reward value is Indicates in status Next select action The probability distribution of is the entropy adjustment coefficient, is the logarithm of the action probability used to calculate the entropy term of the policy; Based on the strategy evaluation results, the pressure fluctuation, temperature difference and temperature change rate are monitored under various fluctuation conditions, the combustion valve opening and pressure parameters are dynamically adjusted, and the adjusted parameter range is recorded to generate a combustion valve and pressure optimization parameter set.

6. The constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm according to claim 5 is characterized in that: The steps for obtaining the heat exchange area optimization parameter set are specifically as follows: Based on the combustion valve and pressure optimization parameter set, combined with the operating state of the wall-mounted boiler heat exchanger, according to the influence of the combustion valve opening and the combustion chamber pressure change on the temperature difference distribution and fluid flow rate of the heat exchange area, the heat exchanger is divided into zones, and the temperature difference distribution range and flow rate target value of each zone are dynamically adjusted to generate the heat flow control zone division result; Based on the heat flow control area division results, the formula is adopted: ； Calculate the total heat exchange of the entire heat exchanger within the target time , in W, based on the total amount of heat exchange Dynamically adjust the heat flow path of each temperature zone to obtain the optimized parameter set of the heat exchange area; in, is the total number of regions, It is The temperature difference of the heat flow control area is It is The flow rate in the heat flow control area is It is The thermal resistance of a heat flow control area.

7. The wall-mounted boiler constant temperature energy-saving control system based on enhanced autonomous learning and SAC algorithm according to claim 1 is characterized in that: The steps of predicting the current heating demand of the wall-mounted boiler are specifically as follows: According to the heating operation data of the wall-mounted boiler, the formula is used: ； Calculate the heat required for heating demand in the future target time period , unit is kWh; in, is the time index, is the total sampling hours, It is Heat loss per hour, It is Output energy efficiency of the equipment per hour, It is Hourly heating area, is the thermal efficiency coefficient of the wall-mounted boiler; Based on the heat required for heating demand in the future target time period, the demand trend data for each time period is sorted out, the dynamic parameters of the combustion valve opening and the gas supply pressure are adjusted in combination with the operating period, and regional integration is performed to obtain the wall-mounted boiler heating demand forecast result.

8. The constant temperature energy-saving control system for a wall-mounted boiler based on enhanced autonomous learning and SAC algorithm according to claim 7 is characterized in that: The steps for obtaining the optimization result of the constant temperature energy-saving control of the wall-mounted boiler are specifically as follows: Based on the prediction result of the heating demand of the wall-mounted boiler, the heat demand is decomposed into multiple areas, and the temperature difference adjustment range and flow rate optimization range of each area in the heat exchange area optimization parameter set are combined. Through dynamic simulation analysis of the operating status under multiple scenarios, it is verified whether the temperature control deviation of the corresponding area is controlled within the set range, and the optimization parameter set of each area is generated; Based on the optimization parameter set of each area, combined with the evaluated temperature control deviation and gas consumption rate, the adjustment range of the combustion valve opening and the gas supply pressure is refined, and a heat exchange path optimization sequence is generated for each area to obtain the optimization result of the constant temperature energy-saving control of the wall-mounted boiler.

Images