An intelligent heating network optimization control method and system based on the Internet of Things

Abstract

The present invention discloses an intelligent heating network optimization control method and system based on the Internet of Things, which relates to the field of heating network optimization control technology, including environmental parameter acquisition modules, which are distributed in different areas covered by the heating network and are used to collect environmental data of each area in real time; an Internet of Things control platform, which is used to receive environmental data from the environmental parameter acquisition module. The intelligent heating network optimization control method and system based on the Internet of Things monitors flow and temperature data in real time, accurately controls valve opening according to a PID control algorithm, forms a closed-loop control system, quickly responds to changes in heating demand, reduces energy waste, reduces energy consumption while ensuring heating quality, and improves energy utilization efficiency; monitors changes in environmental parameters and user heating habits in real time, adjusts heating strategies in a timely manner, quickly responds to fluctuations in heating demand, and ensures that heating always meets actual user needs.

Description

Technical Field

[0001] The present invention relates to the technical field of heating pipe network optimization control, and in particular to an intelligent heating pipe network optimization control method and system based on the Internet of Things. Background Art

[0002] With the acceleration of urbanization and the improvement of people's living standards, centralized heating systems play a vital role in winter heating. However, when the heating network system has large differences in user demand in different regions, the temperature in some areas is often too high while the temperature in other areas is insufficient, resulting in a waste of heating resources and a reduction in residents' heating experience. Taking the heating of typical residential areas in the city as an example, the differences in building locations, floor heights and sunlight conditions will make the heat demand of each household different. In this case, if the heating network fails to dynamically adjust the heat supply, it will be difficult to achieve uniform and effective heat distribution. Summary of the invention

[0003] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: an intelligent heating pipe network optimization control method and system based on the Internet of Things, including environmental parameter acquisition modules, which are distributed in different areas covered by the heating pipe network. The environmental parameter acquisition modules include temperature sensors, humidity sensors and flow sensors, which are used to collect environmental data of each area in real time; the environmental data include outdoor temperature, humidity and flow, and each sensor exchanges data with the Internet of Things control platform through wireless communication;

[0004] The Internet of Things control platform is used to receive environmental data from the environmental parameter acquisition module, analyze the environmental data based on the data analysis algorithm, identify and predict the heating demand of different areas, obtain integrated prediction results, and generate control instructions based on the integrated prediction results;

[0005] The execution layer control module includes electric regulating valves and variable frequency circulation pumps distributed in each branch of the heating network. The execution layer control module is connected to the Internet of Things control platform for communication, receives control instructions from the Internet of Things control platform, and dynamically adjusts the hot water flow and heating temperature of each regional branch to achieve a heating balance in each region.

[0006] User terminal devices interact with the IoT control platform. Users can view the real-time heating status and set individual indoor heating through user terminal devices;

[0007] Preferably, the Internet of Things control platform receives environmental data of each area, and processes the collected environmental data based on a preset data analysis algorithm, identifies differences in heating demand in different areas, and obtains heating demand in different areas;

[0008] The data analysis algorithm includes a heat demand prediction model based on machine learning, which is used to predict the heat demand of each area in the next week or month based on historical environmental data and external weather forecast data;

[0009] It also combines the user's heating habit data to make heating forecasts based on user behavior patterns, obtains integrated forecast results, and adjusts heating parameters based on the integrated forecast results, thereby more accurately reflecting the actual heating needs of each user and further improving the responsiveness and adaptability of the heating system;

[0010] Preferably, the process of the heating demand prediction model predicting the heating demand of each region in the next week or month is as follows: acquiring historical environmental data from temperature sensors, humidity sensors and flow sensors arranged in each region of the heating pipe network system, including outdoor temperature, humidity and flow data in different time periods; and classifying and cleaning the historical environmental data, removing abnormal values ​​and erroneous data, and ensuring the accuracy and completeness of the data; storing the sorted historical environmental data according to region and time series to form a historical environmental data set for analysis and model training; the time series includes daily, weekly and monthly;

[0011] Obtain weather forecast data for the next week or month from professional meteorological agencies through the data receiving unit of the IoT control platform, including predicted outdoor temperature and humidity information; use the sliding average method to remove short-term fluctuation noise in the data to obtain a weather forecast data set, and use data verification rules to check the rationality of the data to ensure that the data meets the requirements of subsequent analysis and model input; at the same time, convert the data into a format that matches historical environmental data for unified processing, including converting temperature data into Celsius units and adjusting the time format to a timestamp format consistent with historical data;

[0012] Extract features related to heating demand from the historical environmental data set and the weather forecast data set, obtain relevant features, perform normalization processing, and map them to [0,1] to obtain historical environmental data features and weather forecast data features respectively; make different features have the same dimension to avoid affecting the model training effect due to large differences in feature values;

[0013] A neural network model is selected as the basic framework of the heating demand prediction model, and the number of hidden layers, the number of neurons in each layer, and the activation function are determined; including: setting a neural network with three hidden layers, the number of neurons in each layer is [64, 32, 16], and the activation function uses the ReLU corrected linear unit function; through empirical rules and multiple experiments, a model structure that can achieve good performance on the training set and is not overfitting is selected;

[0014] The processed historical environmental data features and the corresponding historical heating demand data are used as training samples and input into the constructed model for training. Stochastic gradient descent (SGD) is used to optimize and adjust the model parameters so that the model can learn the mapping relationship between environmental data and heating demand. The historical heating demand data is obtained through the actual heating amount or the comfort feedback from users.

[0015] During the training process, the data set is divided into a training set, a validation set, and a test set, with the ratio set to 70%:15%:15%. The performance indicators of the model are evaluated on the validation set: mean square error (MSE) and mean absolute error (MAE) to monitor whether the model is overfitting or underfitting. If the performance indicators on the validation set no longer improve or begin to deteriorate, the training is stopped to obtain a heating demand prediction model to prevent the model from over-learning the noise in the training data and losing its generalization ability.

[0016] Input the weather forecast data features into the heating demand prediction model; the heating demand prediction model calculates and outputs the prediction results based on the learned patterns and rules, and the prediction results include the predicted heating demand values ​​of each area in the corresponding time period in the future;

[0017] Preferably, the heating prediction is performed based on the user behavior pattern in combination with the user's heating habit data, and the process is as follows: the heating habit information actively input by the user is collected through the user terminal device, the heating habit information includes the indoor temperature preference value set by the user on a daily basis, the heating demand change pattern in different time periods, and whether there is a specific heating time pattern, where the different time periods are daytime, evening, and weekends on weekdays; the specific heating time pattern includes the timed switching of the heating equipment; the indirect data generated by the interaction between the user and the heating system is collected, and the heating habit data is analyzed, including the frequency, adjustment range and adjustment time point operation records of the user adjusting the heating temperature on the terminal device, which can reflect the user's The real-time changes in users' demand for heating comfort; mining potential heating habit data from users' past heating bill information, including the peak and trough periods of monthly heating consumption, and the trend of total heating consumption in different seasons, to understand users' heating behavior patterns from the perspective of heating consumption; cleaning the collected user heating habit data to remove invalid, erroneous or duplicate data records; correcting or deleting obviously unreasonable temperature preference values ​​to ensure the accuracy and reliability of the data; and converting data of different formats and types into a unified format suitable for analysis; extracting relevant features based on the characteristics of user heating habit data and prediction results; and constructing composite features by combining environmental data and time information;

[0018] Through cluster analysis, users with similar heating habits are grouped into the same category, and a typical behavior pattern model is established for each cluster. The sequence pattern mining algorithm is used to discover the sequence patterns of users' heating operation at different time points, analyze their frequency and conditions of occurrence, and clarify the changing rules of users' dynamic heating demand under different circumstances.

[0019] The average heating demand and heating demand fluctuation range under different behavior patterns are calculated through statistical analysis methods, and a mapping relationship model between behavior patterns and heating demand is established to clarify the quantitative relationship between different user behavior patterns and heating demand.

[0020] At the same time, the impact of environmental factors on the relationship between user behavior patterns and heating demand is considered. A comprehensive model including outdoor temperature, humidity, user behavior pattern characteristics and heating demand in environmental variables is established through multivariate regression analysis. This model analyzes how user behavior patterns affect changes in heating demand under different environmental conditions, and obtains heating demand prediction results based on user behavior patterns, thereby more accurately predicting heating demand.

[0021] Determine the strategy for fusing the heating demand forecast results based on user behavior patterns with the forecast model based on historical environmental data and weather forecast data; the fusion strategies include: weighted average method: assigning weights according to the accuracy and importance of different models; stacked ensemble learning: using the forecast results of one model as the input features of another model for secondary prediction;

[0022] Evaluate the performance of different fusion strategies on the training data set and select the fusion strategy that optimizes the overall heating forecasting model performance, including: the fusion strategy with the smallest prediction error and the highest stability; determine the final fusion strategy;

[0023] The heating demand forecast results obtained based on the user behavior pattern analysis are integrated with the output of the heating demand forecast model based on historical environmental data and weather forecast data according to the final fusion strategy to obtain the integrated heating demand forecast model and obtain the final heating demand forecast value;

[0024] The integrated heating demand forecasting model is used to forecast the heating demand of each region in the next week or month, and the integrated forecasting results are obtained. The integrated forecasting results are then applied to the optimization control decision of the heating network system. The hot water flow and heating temperature of each regional branch are dynamically adjusted according to the predicted heating demand to ensure the heating balance of each region. At the same time, accurate heating is achieved by combining the user's personalized settings, including the indoor temperature target value and heating schedule set by the user through the terminal device, to improve the user's comfort and energy efficiency.

[0025] Preferably, the execution layer control module is used to receive control instructions and dynamically adjust the heating parameters of the heating network according to the control instructions; including: adjusting the hot water flow and heating temperature of each regional branch by controlling the execution layer equipment to ensure the heating balance of each region, wherein the execution layer equipment includes an electric control valve, a variable frequency circulation pump and a heat pump; wherein the electric control valve has an automatic calibration function, which is used to automatically adjust to the optimal state according to real-time flow and temperature data to improve the control accuracy and response speed of the system; the electric control valve also has a self-learning function, which is used to optimize its own adjustment process by analyzing historical control data and current heating demand, thereby achieving more efficient heating flow control and gradually improving the control accuracy in long-term operation;

[0026] The self-learning function is based on a reinforcement learning algorithm. It optimizes the adjustment strategy by learning the feedback between each adjustment operation and the heating result. Especially when facing complex changes in heating demand, the self-learning function can better adapt to different working conditions and achieve refined control of the heating process, thereby effectively reducing energy consumption while ensuring the quality of heating.

[0027] Preferably, the process of the execution layer control module dynamically adjusting the heating parameters of the heating network according to the control instructions is as follows: the execution layer control module first receives the control instructions from the central control system through wireless communication; the control instructions include the set target heating temperature, flow distribution ratio and heating strategy, and the control instructions are parsed by the built-in microprocessor of the execution layer control module, and the control instructions are converted into operation signals; based on the operation signals, the execution layer control module sends control signals to the electric regulating valves and variable frequency circulation pumps in each area, wherein the electric regulating valves are installed at the hot water inlet of each branch, and are used to automatically adjust the valve opening and the output temperature of the heat pump according to the flow distribution ratio and the target heating temperature in the control signal, respectively, to obtain the opening adjustment parameters and the temperature adjustment parameters, thereby changing the amount of hot water flowing through the branch; this adjustment method can very finely control the water supply and supply of each area. Water temperature, to adapt to different heat load requirements; the variable frequency circulation pump automatically adjusts the speed according to the heating strategy, obtains the speed adjustment parameter, and forms the primary heating adjustment parameter based on the opening adjustment parameter, temperature adjustment parameter and speed adjustment parameter, which not only ensures sufficient water circulation power, but also avoids unnecessary energy consumption; by monitoring the system pressure and flow, the control module can intelligently adjust the working state of the pump to maintain the stable operation of the heating system; and through the sensor unit equipped in the execution layer control module, the key parameters of the heating network are monitored in real time, including temperature parameters, pressure parameters and flow parameters; and the key parameters are fed back to the execution layer control module for evaluating the current heating results, obtaining heating result feedback, and adjusting the control instructions based on the heating result feedback, generating new control instructions, and re-dynamically adjusting the heating parameters of the heating network based on the new control instructions to obtain secondary heating adjustment parameters;

[0028] Preferably, the process of the electric regulating valve automatically adjusting to the optimal state according to the real-time flow and temperature data is as follows: the temperature and flow data of the hot water are monitored in real time by the temperature sensors and flow sensors installed in each branch of the heating pipe network, the temperature sensor accurately measures the real-time temperature of the hot water at different positions in the pipe, and the flow sensor measures the flow of hot water flowing through the branch where the electric regulating valve is located; the sensor transmits the collected temperature and flow data to the Internet of Things control platform in the form of electrical signals to ensure the timely acquisition and transmission accuracy of the data, and provide basic data support for the adjustment of the electric regulating valve; according to the heating demand prediction model and the indoor temperature target value and heating schedule set by the user terminal device, the Internet of Things control platform calculates the ideal hot water flow and heating temperature target value corresponding to the current moment for each area or branch; after receiving the real-time flow and temperature data, the control platform compares it with the target value and calculates the flow deviation and temperature deviation, the flow deviation is the difference between the actual flow and the target flow, and the temperature deviation is the difference between the actual temperature and the target temperature;

[0029] The built-in control algorithm of the electric control valve is based on the PID control algorithm in classical control theory. It calculates according to the flow deviation and temperature deviation to determine the opening adjustment of the control valve; the output control signal is calculated according to the three parts of the deviation: proportional P, integral I and differential D. The proportional term reflects the size of the current deviation and responds to the deviation in a timely manner; the integral term is used to eliminate the steady-state error of the system and integrates the accumulated deviation over time; the differential term predicts the system trend in advance according to the rate of change of the deviation and suppresses the overshoot of the system; by adjusting these three parameters, the control of the electric control valve can be made more precise and stable; the opening adjustment signal calculated by the control algorithm is transmitted to the drive actuator of the electric control valve: the motor; the motor controls the position of the valve core of the electric control valve according to the received signal and changes the opening size of the valve; after adjusting the opening, the electric control valve continues to monitor the flow and temperature data in real time, repeats the above deviation calculation, control algorithm operation and drive actuator adjustment process to form a closed-loop control system;

[0030] Preferably, the process of realizing more efficient heating flow control by the electric regulating valve is as follows: Feature extraction: extracting features related to heating flow control from historical control data and current heating demand information; including: taking the outdoor temperature change rate, the user-set temperature adjustment amplitude, and the difference in flow changes of adjacent branches as input features, and taking the opening adjustment amount of the electric regulating valve as output features;

[0031] Model training and optimization: Q-learning in the reinforcement learning algorithm is used to build a self-learning model for the electric control valve. The extracted feature data is input into the model. The model learns the optimal adjustment strategy under different conditions to achieve efficient heating flow control actions to maximize long-term heating results and energy utilization efficiency. During the training process, the model parameters are continuously optimized so that the model can accurately predict the opening adjustment actions that the electric control valve should take under various working conditions to achieve a more efficient heating flow control goal. This includes: using the reward function in the Q-learning algorithm to evaluate the effect of each adjustment operation. If the heating flow is closer to the target value and energy consumption is reduced after adjustment, a positive reward is given, otherwise a negative reward is given to guide the model to learn the optimal adjustment strategy.

[0032] Working condition judgment and strategy selection: Based on the current real-time monitored heating demand data and environmental parameters, the trained self-learning model is used to judge the current working condition; the model selects the most suitable adjustment strategy for the current situation in the learned strategy space according to the input feature data: current outdoor temperature, user set temperature, and flow conditions of other branches, and determines the opening adjustment direction and amplitude of the electric control valve; Execute adjustment action: The electric control valve controls the valve core position by driving the actuator according to the adjustment strategy determined by the model to achieve real-time adjustment of the heating flow; during the adjustment process, the flow change is continuously monitored to ensure that the flow adjustment is carried out in the direction close to the target value; at the same time, it works in coordination with the variable frequency circulation pump in the heating pipe network system to dynamically balance the flow distribution of each branch according to the overall heating demand of the system, avoid local overheating or overcooling, and achieve efficient heating flow control;

[0033] Effect evaluation and feedback collection: After completing a heating flow adjustment, immediately collect feedback information on the adjusted heating results; the heating result feedback information includes the actual indoor temperature changes, the energy consumption changes of the heating system, and whether the user adjusts the heating settings again; by comparing the various indicators before and after the adjustment, evaluate the effectiveness and efficiency of this adjustment operation on the control of the heating flow; if the indoor temperature quickly reaches the target value after the adjustment and the energy consumption does not increase significantly, it is considered that the adjustment effect is good; on the contrary, if there is a large temperature fluctuation or the energy consumption is too high, it is necessary to further analyze the cause;

[0034] Model parameter update and optimization: Based on the feedback evaluation results, the learning mechanism in the reinforcement learning algorithm is used to update the parameters of the self-learning model; if a certain adjustment operation obtains a good heating result and lower energy consumption, the model will strengthen the parameters related to the adjustment strategy that leads to the result, making it more likely to be selected under similar working conditions in the future; conversely, if the adjustment effect is not good, the model will adjust the parameters accordingly to avoid using similar undesirable adjustment strategies again; by constantly repeating the above data collection, analysis, decision-making, feedback and model update process, the electric control valve can gradually improve the control accuracy in the long-term operation, constantly adapt to various changes in the heating system, achieve more efficient and accurate heating flow control, and improve the stability and energy efficiency of the entire heating system;

[0035] Preferably, the process of realizing more efficient heating flow control by the electric regulating valve also includes:

[0036] Historical control data accumulation: During the operation of the heating system, the electric control valve continuously records the adjustment time of each adjustment operation, the flow value before adjustment, the flow value after adjustment, the corresponding temperature data, the outdoor ambient temperature at that time, the humidity data and the heating parameters set by the user: target temperature, heating schedule, and obtains historical control data;

[0037] Current heating demand monitoring: The current heating demand information is obtained in real time through the IoT control platform, including the heating demand predicted for each area based on environmental parameters: outdoor temperature, humidity and user heating habits, as well as the heating settings adjusted by users in real time through terminal devices. The heating settings include temporarily increasing or decreasing the indoor temperature. At the same time, the operating status of the variable frequency circulation pump in the heating network system and the opening of the electric regulating valves in other branches are closely monitored, and their impact on the current heating demand. The above information is combined to determine the actual heating demand changes of each branch where the electric regulating valve is located at the current moment, and obtain the current heating demand information.

[0038] An intelligent heating network optimization control method based on the Internet of Things comprises the following steps:

[0039] Step 1: Environmental data collection: Collect environmental data in real time in different areas covered by the heating network, including outdoor temperature, humidity and flow, and transmit the collected environmental data to the IoT control platform via wireless communication;

[0040] Step 2: Prediction and analysis of heating demand: Collect historical environmental data and obtain future weather forecast data, then extract relevant features of heating demand, select a neural network model for training to obtain a heating demand prediction model; collect user heating habit data, extract features to establish a user behavior pattern model, clarify the quantitative relationship between behavior patterns and heating demand, and obtain heating demand prediction results based on user behavior;

[0041] Step 3: Based on the fusion strategy, the user behavior prediction results are integrated with the output of the historical environment and weather forecast data prediction model to obtain the final heating demand prediction model, predict future heating demand, obtain the integrated prediction results, and generate control instructions based on the integrated prediction results;

[0042] Step 4: Based on the control instruction, the electric regulating valve is controlled to automatically adjust the valve opening and the heat pump output temperature according to the flow distribution ratio and the target heating temperature. The variable frequency circulation pump adjusts the speed according to the heating strategy to achieve dynamic adjustment of the hot water flow and heating temperature of each regional branch to ensure the balance of heating supply;

[0043] Step 5: Users interact with the IoT control platform through user terminal devices to view real-time heating status and set individual indoor heating.

[0044] The present invention provides an intelligent heating pipe network optimization control method and system based on the Internet of Things, which has the following beneficial effects:

[0045] 1. The intelligent heating network optimization control method and system based on the Internet of Things predicts heating demand by combining historical environmental data, weather forecast data and user heating habit data; and collects data by arranging multiple sensors in various areas of the heating network, and classifies, cleans, extracts features and normalizes the data to provide a rich and high-quality data basis for accurate prediction; uses a neural network model to train the heating demand prediction model, and improves the prediction accuracy from multiple dimensions to better adapt to the changes in heating demand in different regions and users, and provide strong support for precise heating; and according to the control instructions generated by the Internet of Things control platform, dynamically adjusts the hot water flow and heating temperature of each regional branch through electric regulating valves and variable frequency circulation pumps; achieves heating balance in each area, ensures that the user's indoor temperature is stable and comfortable, and improves the stability and reliability of the heating system.

[0046] 2. The intelligent heating pipe network optimization control method and system based on the Internet of Things monitors flow and temperature data in real time, accurately controls valve opening according to the PID control algorithm, forms a closed-loop control system, quickly responds to changes in heating demand, reduces energy waste, reduces energy consumption while ensuring heating quality, and improves energy utilization efficiency; monitors environmental parameters and changes in user heating habits in real time, adjusts heating strategies in a timely manner, quickly responds to fluctuations in heating demand, copes with sudden changes in outdoor temperature or temporary adjustments to temperature settings by users, ensures that heating always meets actual user needs, and improves users' overall experience of the heating system; can accurately identify differences in heating demand in different regions, generate scientific and reasonable control instructions, enable the heating system to adapt to complex and changing environments and user needs, and improve the system's adaptability and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1It is the flowchart of an intelligent heating pipe network optimization control system based on the Internet of Things according to the present invention;

[0048] Figure 2 It is the step block diagram of an intelligent heating pipe network optimization control method according to the present invention. Specific embodiments

[0049] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are given for purposes of illustration and description, and are not exhaustive or limit the present invention to the disclosed form. Many modifications and variations are obvious to those of ordinary skill in the art. The embodiments are selected and described to better illustrate the principles and practical applications of the present invention, and enable those of ordinary skill in the art to understand the present invention and thus design various embodiments with various modifications suitable for specific purposes.

[0050] As Figure 1 and Figure 2 shown, the present invention provides a technical solution: an intelligent heating pipe network optimization control method and system based on the Internet of Things, including an environmental parameter acquisition module, which is distributed in different areas covered by the heating pipe network. The environmental parameter acquisition module includes a temperature sensor, a humidity sensor and a flow sensor, and is used to collect the environmental data of each area in real time; the environmental data includes outdoor temperature, humidity, and flow, and each sensor communicates with the Internet of Things control platform through wireless communication to perform data interaction;

[0051] The Internet of Things control platform is used to receive the environmental data from the environmental parameter acquisition module, analyze the environmental data based on the data analysis algorithm, identify and predict the heating demand of different areas, obtain the integrated prediction result, and generate a control instruction based on the integrated prediction result;

[0052] The execution layer control module includes electric control valves and variable frequency circulating pumps distributed in each branch of the heating pipe network. The execution layer control module is communicatively connected to the Internet of Things control platform, receives the control instructions of the Internet of Things control platform, and dynamically adjusts the hot water flow and heating temperature of each area branch to achieve the heating balance of each area;

[0053] The user terminal device interacts with the Internet of Things control platform. The user views the real-time heating status through the user terminal device and sets the individual indoor heating.

[0054] The Internet of Things control platform receives the environmental data of each area, processes the collected environmental data based on the preset data analysis algorithm, identifies the heating demand differences of different areas, and obtains the heating demands of different areas;

[0055] The data analysis algorithm includes a heat demand forecasting model based on machine learning. The heat demand forecasting model is used to predict the heat demand of each area in the next week or month based on historical environmental data and external weather forecast data;

[0056] It also combines the user's heating habit data to make heating forecasts based on user behavior patterns, obtains integrated forecast results, and adjusts heating parameters based on the integrated forecast results, thereby more accurately reflecting the actual heating needs of each user and further improving the responsiveness and adaptability of the heating system;

[0057] The heating demand forecasting model predicts the heating demand of each region in the next week or month as follows:

[0058] Obtain historical environmental data from temperature sensors, humidity sensors, and flow sensors installed in various areas of the heating network system, including outdoor temperature, humidity, and flow data for different time periods, such as outdoor temperature, humidity, and flow data for the past few years; classify and clean the historical environmental data, remove outliers and erroneous data, and ensure the accuracy and completeness of the data; store the sorted historical environmental data by region and time series to form a historical environmental data set that can be used for analysis and model training; time series include daily, weekly, and monthly;

[0059] Obtain weather forecast data for the next week or month from professional meteorological agencies through the data receiving unit of the IoT control platform, including predicted outdoor temperature and humidity information; use the sliding average method to remove short-term fluctuation noise in the data to obtain a weather forecast data set, and use data verification rules to check the rationality of the data to ensure that the data meets the requirements of subsequent analysis and model input; at the same time, convert the data into a format that matches historical environmental data for unified processing, including converting temperature data into Celsius units and adjusting the time format to a timestamp format consistent with historical data;

[0060] Extract features related to heating demand from the historical environmental data set and the weather forecast data set, obtain relevant features, perform normalization processing, and map them to [0,1] to obtain historical environmental data features and weather forecast data features respectively; make different features have the same dimension to avoid affecting the model training effect due to large differences in feature values;

[0061] A neural network model is selected as the basic framework of the heating demand prediction model, and the number of hidden layers, the number of neurons in each layer, and the activation function are determined; including: setting a neural network with three hidden layers, the number of neurons in each layer is [64, 32, 16], and the activation function uses the ReLU corrected linear unit function; through empirical rules and multiple experiments, a model structure that can achieve good performance on the training set and is not overfitting is selected;

[0062] The processed historical environmental data features and the corresponding historical heating demand data are used as training samples and input into the constructed model for training. Stochastic gradient descent (SGD) is used to optimize and adjust the model parameters so that the model can learn the mapping relationship between environmental data and heating demand. The historical heating demand data is obtained through the actual heating amount or the comfort feedback from users.

[0063] During the training process, the data set is divided into a training set, a validation set, and a test set, with the ratio set to 70%:15%:15%. The performance indicators of the model are evaluated on the validation set: mean square error (MSE) and mean absolute error (MAE) to monitor whether the model is overfitting or underfitting. If the performance indicators on the validation set no longer improve or begin to deteriorate, the training is stopped to obtain a heating demand prediction model to prevent the model from over-learning the noise in the training data and losing its generalization ability.

[0064] Input the weather forecast data features into the heat demand prediction model; the heat demand prediction model calculates and outputs the prediction results based on the learned patterns and rules, and the prediction results include the predicted heat demand values ​​of each region in the corresponding time period in the future; for example, for a certain region, the model predicts that the daily heat demand in the next week is approximately [Q1, Q2, Q3, Q4, Q5, Q6, Q7], where Q is the heat demand, and the unit is set according to the actual situation, such as megawatts and gigajoules;

[0065] Combined with the user's heating habit data, heating supply prediction is performed based on the user's behavior pattern. The process is as follows:

[0066] The user terminal device collects the heating habit information actively input by the user. The heating habit information includes the indoor temperature preference value set by the user on a daily basis, the heating demand change pattern in different time periods, and whether there is a specific heating time pattern, where the different time periods are daytime, evening, and weekends on weekdays; the specific heating time pattern includes the timed switching of heating equipment; the indirect data generated by the interaction between the user and the heating system is collected, and the heating habit data is analyzed, including the frequency, adjustment range, and adjustment time point operation records of the user adjusting the heating temperature on the terminal device. These data can reflect the real-time changes in the user's demand for heating comfort; potential heating habit data is mined from the user's past heating bill information, including the peak and trough periods of monthly heating consumption, and the trend of total heating consumption in different seasons, so as to understand the user's heating behavior pattern from the perspective of heating consumption; the collected user heating habit data is cleaned to remove invalid, erroneous or duplicate data records; and obviously unreasonable temperature preference values ​​are corrected or deleted. For example, if the temperature exceeds the temperature range that the heating equipment can provide, ensure the accuracy and reliability of the data; and convert data of different formats and types into a unified format suitable for analysis; such as converting time-related data into a standard time format: timestamp, and normalizing the numerical data of temperature preference so that it falls within the numerical range [0,1] or [-1,1] to facilitate subsequent calculations and model processing; extract relevant features based on the characteristics and prediction results of user heating habit data; for example, create "average set temperature during the day on weekdays", "frequency of temperature adjustment at weekend nights", and "duration of monthly peak heating hours" features to quantify users' heating behavior patterns in different scenarios; combine environmental data and time information to construct composite features; such as calculating "user-set temperature increase when outdoor temperature is below 5°C" and "correlation between user-adjusted heating temperature and outdoor humidity in a specific time period (such as 7pm-10pm)" features to further explore the relationship between users' heating behavior and environmental factors, and provide richer information for heating prediction;

[0067] Through cluster analysis, users with similar heating habits are grouped into the same category. A typical behavior pattern model is established for each cluster. The sequence pattern mining algorithm is used to discover the user's heating operation sequence patterns at different time points, such as "lower the temperature first, keep it for a while, and then increase the temperature". The frequency and conditions of the operation sequence are analyzed to clarify the dynamic heating demand change rules of users in different situations.

[0068] The average heating demand and fluctuation range of heating demand under different behavior patterns are calculated through statistical analysis methods, and a mapping relationship model between behavior patterns and heating demand is established to clarify the quantitative relationship between different user behavior patterns and heating demand. For example, it is found that users with the behavior pattern of "high temperature preference and frequent adjustment of heating time" usually have a certain percentage higher heating demand than other users in cold weather, and this relationship is incorporated into the heating prediction model.

[0069] At the same time, the impact of environmental factors on the relationship between user behavior patterns and heating demand is considered. A comprehensive model including outdoor temperature, humidity, user behavior pattern characteristics and heating demand in environmental variables is established through multivariate regression analysis. This model analyzes how user behavior patterns affect changes in heating demand under different environmental conditions, and obtains heating demand prediction results based on user behavior patterns, thereby more accurately predicting heating demand.

[0070] Determine the strategy for fusing the heating demand forecast results based on user behavior patterns with the forecast model based on historical environmental data and weather forecast data; the fusion strategies include: weighted average method: assigning weights according to the accuracy and importance of different models; stacked ensemble learning: using the forecast results of one model as the input features of another model for secondary prediction;

[0071] Evaluate the performance of different fusion strategies on the training data set and select the fusion strategy that optimizes the overall heating forecasting model, including: the fusion strategy with the smallest prediction error and the highest stability; for example, compare the performance of the weighted average method (the weight is determined by the model evaluation index) and stacked ensemble learning (using different basic model combinations) in terms of prediction accuracy and recall rate through cross-validation experiments to determine the final fusion strategy;

[0072] The heating demand forecast results obtained based on the user behavior pattern analysis are integrated with the output of the heating demand forecast model based on historical environmental data and weather forecast data according to the final fusion strategy to obtain the integrated heating demand forecast model and the final heating demand forecast value; if the weighted average method is used, the weight of the user behavior pattern model forecast result is determined according to the previous evaluation as , the weights of the prediction results of the historical environment and weather forecast data model are ,in, ;

[0073] The final predicted value , the calculation formula is:

[0074] ;in The heating demand predicted by the user behavior pattern model, heating demand predicted by models of historical environmental and weather forecast data;

[0075] The integrated heating demand forecasting model is used to forecast the heating demand of each region in the next week or month, and the integrated forecasting results are obtained. The integrated forecasting results are then applied to the optimization control decision of the heating network system. The hot water flow and heating temperature of each regional branch are dynamically adjusted according to the predicted heating demand to ensure the heating balance of each region. At the same time, accurate heating is achieved by combining the user's personalized settings, including the indoor temperature target value and heating schedule set by the user through the terminal device, to improve the user's comfort and energy efficiency.

[0076] The execution layer control module is used to receive control instructions and dynamically adjust the heating parameters of the heating network according to the control instructions; including: adjusting the hot water flow and heating temperature of each regional branch by controlling the execution layer equipment to ensure the heating balance of each region, wherein the execution layer equipment includes electric control valves, variable frequency circulation pumps and heat pumps; wherein the electric control valve has an automatic calibration function, which is used to automatically adjust to the optimal state according to real-time flow and temperature data to improve the control accuracy and response speed of the system; the electric control valve also has a self-learning function, which is used to optimize its own adjustment process by analyzing historical control data and current heating demand, thereby achieving more efficient heating flow control and gradually improving the control accuracy in long-term operation;

[0077] The self-learning function is based on a reinforcement learning algorithm. It optimizes the adjustment strategy by learning the feedback between each adjustment operation and the heating result. Especially when facing complex changes in heating demand, the self-learning function can better adapt to different working conditions and achieve refined control of the heating process, thereby effectively reducing energy consumption while ensuring heating quality;

[0078] The process of the execution layer control module dynamically adjusting the heating parameters of the heating network according to the control instructions is as follows: the execution layer control module first receives the control instructions from the central control system through wireless communication; the control instructions include the set target heating temperature, flow distribution ratio and heating strategy, and the control instructions are parsed by the built-in microprocessor of the execution layer control module, and the control instructions are converted into operation signals; based on the operation signal, the execution layer control module sends control signals to the electric regulating valves and variable frequency circulation pumps in each area, wherein the electric regulating valves are installed at the hot water inlet of each branch, and are used to automatically adjust the valve opening and the output temperature of the heat pump according to the flow distribution ratio and the target heating temperature in the control signal, and obtain the opening adjustment parameters and the temperature adjustment parameters respectively, thereby changing the amount of hot water flowing through the branch; this adjustment method can very finely control the water supply and water temperature of each area. degree to adapt to different heat load requirements; the variable frequency circulation pump automatically adjusts the speed according to the heating strategy to obtain the speed adjustment parameters, and the primary heating adjustment parameters are formed based on the opening adjustment parameters, temperature adjustment parameters and speed adjustment parameters, which not only ensures sufficient water circulation power, but also avoids unnecessary energy consumption; by monitoring the system pressure and flow, the control module can intelligently adjust the working state of the pump to maintain the stable operation of the heating system; and through the sensor unit equipped in the execution layer control module, the key parameters of the heating network are monitored in real time, and the key parameters include temperature parameters, pressure parameters and flow parameters; and the key parameters are fed back to the execution layer control module for evaluating the current heating results, obtaining heating result feedback, and adjusting the control instructions based on the heating result feedback, generating new control instructions, and re-dynamically adjusting the heating parameters of the heating network based on the new control instructions to obtain secondary heating adjustment parameters;

[0079] The process of the electric regulating valve automatically adjusting to the optimal state according to the real-time flow and temperature data is as follows: the temperature and flow data of the hot water are monitored in real time by the temperature sensors and flow sensors installed in each branch of the heating network. The temperature sensor accurately measures the real-time temperature of the hot water at different positions in the pipeline, and the flow sensor measures the flow of hot water flowing through the branch where the electric regulating valve is located; the sensor transmits the collected temperature and flow data to the Internet of Things control platform in the form of electrical signals to ensure the timely acquisition and transmission accuracy of the data, and provide basic data support for the adjustment of the electric regulating valve; according to the heating demand prediction model and the indoor temperature target value and heating schedule set by the user terminal device, the Internet of Things control platform calculates the ideal hot water flow and heating temperature target value corresponding to the current moment for each area or branch; after receiving the real-time flow and temperature data, the control platform compares it with the target value and calculates the flow deviation and temperature deviation. The flow deviation is the difference between the actual flow and the target flow, and the temperature deviation is the difference between the actual temperature and the target temperature.

[0080] The built-in control algorithm of the electric control valve is based on the PID control algorithm in the classical control theory. It calculates according to the flow deviation and temperature deviation to determine the opening adjustment of the control valve; the output control signal is calculated according to the three parts of the deviation: proportional P, integral I and differential D. The proportional term reflects the size of the current deviation and responds to the deviation in a timely manner; the integral term is used to eliminate the steady-state error of the system and integrates the accumulated deviation over time; the differential term predicts the system trend in advance according to the rate of change of the deviation and suppresses the overshoot of the system; by adjusting these three parameters, the control of the electric control valve can be made more accurate and stable; the opening adjustment signal calculated by the control algorithm is The signal is transmitted to the driving actuator of the electric control valve: the motor; the motor controls the position of the valve core of the electric control valve according to the received signal and changes the valve opening; for example, if the calculation result shows that the hot water flow needs to be increased to reach the target temperature, the driving actuator will move the valve core in the opening direction to increase the valve opening, thereby increasing the amount of hot water passing through; conversely, if the flow or temperature needs to be reduced, the valve core will move in the closing direction to reduce the opening; after adjusting the opening, the electric control valve continues to monitor the flow and temperature data in real time, repeating the above deviation calculation, control algorithm operation and driving actuator adjustment process to form a closed-loop control system;

[0081] Through continuous dynamic adjustment, the electric control valve quickly responds to various factors in the heating system, such as outdoor temperature changes, changes in user heating habits, and fluctuations in heating demand from other branches, which cause flow and temperature changes, so that the hot water flow and heating temperature gradually approach and stabilize near the target value, achieving precise control of the system, improving the control accuracy and response speed of the heating system, ensuring balanced heating in each area, and effectively reducing energy consumption;

[0082] The process of electric regulating valve to achieve more efficient heating flow control is as follows:

[0083] Feature extraction: Extract features related to heating flow control from historical control data and current heating demand information; including: taking the outdoor temperature change rate, user-set temperature adjustment range, and adjacent branch flow change difference as input features, and taking the opening adjustment amount of the electric control valve as output features;

[0084] Model training and optimization: Q-learning in the reinforcement learning algorithm is used to build a self-learning model for the electric control valve. The extracted feature data is input into the model. The model learns the optimal adjustment strategy under different conditions to achieve efficient heating flow control actions to maximize long-term heating results and energy utilization efficiency. During the training process, the model parameters are continuously optimized so that the model can accurately predict the opening adjustment actions that the electric control valve should take under various working conditions to achieve a more efficient heating flow control goal. This includes: using the reward function in the Q-learning algorithm to evaluate the effect of each adjustment operation. If the heating flow is closer to the target value and energy consumption is reduced after adjustment, a positive reward is given, otherwise a negative reward is given to guide the model to learn the optimal adjustment strategy.

[0085] Working condition judgment and strategy selection: Based on the current real-time monitored heating demand data and environmental parameters, the trained self-learning model is used to judge the current working condition. The model selects the most suitable adjustment strategy for the current situation in the learned strategy space according to the input feature data: current outdoor temperature, user set temperature, and other branch flow conditions, and determines the opening adjustment direction and amplitude of the electric control valve. For example, if the current outdoor temperature drops sharply and the user increases the indoor temperature target value, the model will judge that the heating flow needs to be greatly increased, and thus select a larger opening increase.

[0086] Execute regulation action: The electric regulating valve controls the valve core position by driving the actuator according to the regulation strategy determined by the model to achieve real-time adjustment of the heating flow rate; during the regulation process, the flow rate changes are continuously monitored to ensure that the flow rate adjustment is in the direction close to the target value; at the same time, it works in coordination with the variable frequency circulation pump in the heating pipe network system to dynamically balance the flow distribution of each branch according to the overall heating demand of the system, avoid local overheating or overcooling, and achieve efficient heating flow control;

[0087] Effect evaluation and feedback collection: After completing a heating flow adjustment, immediately collect feedback information on the adjusted heating results; the heating result feedback information includes the actual indoor temperature changes, the energy consumption changes of the heating system, and whether the user adjusts the heating settings again; by comparing the various indicators before and after the adjustment, evaluate the effectiveness and efficiency of this adjustment operation on the control of the heating flow; if the indoor temperature quickly reaches the target value after the adjustment and the energy consumption does not increase significantly, it is considered that the adjustment effect is good; on the contrary, if there is a large temperature fluctuation or the energy consumption is too high, it is necessary to further analyze the cause;

[0088] Model parameter update and optimization: Based on the feedback evaluation results, the learning mechanism in the reinforcement learning algorithm is used to update the parameters of the self-learning model; if a certain adjustment operation obtains a good heating result and lower energy consumption, the model will strengthen the parameters related to the adjustment strategy that leads to the result, making it more likely to be selected under similar working conditions in the future; conversely, if the adjustment effect is not good, the model will adjust the parameters accordingly to avoid using similar undesirable adjustment strategies again; by constantly repeating the above data collection, analysis, decision-making, feedback and model update process, the electric control valve can gradually improve the control accuracy in the long-term operation, constantly adapt to various changes in the heating system, achieve more efficient and accurate heating flow control, and improve the stability and energy efficiency of the entire heating system;

[0089] The process of achieving more efficient heating flow control by electric control valve also includes:

[0090] Historical control data accumulation: During the operation of the heating system, the electric control valve continuously records the adjustment time of each adjustment operation, the flow value before adjustment, the flow value after adjustment, the corresponding temperature data, the outdoor ambient temperature at that time, the humidity data and the heating parameters set by the user: target temperature, heating schedule, and obtains historical control data;

[0091] Current heating demand monitoring: The current heating demand information is obtained in real time through the IoT control platform, including the heating demand predicted for each area based on environmental parameters: outdoor temperature, humidity and user heating habits, as well as the heating settings adjusted by users in real time through terminal devices. The heating settings include temporarily increasing or decreasing the indoor temperature. At the same time, the operating status of the variable frequency circulation pump in the heating network system and the opening of the electric regulating valves in other branches are closely monitored, and their impact on the current heating demand. The above information is combined to determine the actual heating demand changes of each branch where the electric regulating valve is located at the current moment, and obtain the current heating demand information.

[0092] It should be further explained that, in the specific implementation process, environmental data collection is first carried out: environmental data including outdoor temperature, humidity and flow are collected in real time in different areas covered by the heating network, and the collected environmental data are transmitted to the Internet of Things control platform through wireless communication; then heating demand prediction and analysis are carried out: historical environmental data are collected and future weather forecast data is obtained, and then the relevant features of heating demand are extracted, and a neural network model is selected for training to obtain a heating demand prediction model; user heating habit data is collected, features are extracted to establish a user behavior pattern model, the quantitative relationship between behavior pattern and heating demand is clarified, and the heating demand prediction result based on user behavior is obtained;

[0093] Subsequently, based on the fusion strategy, the user behavior prediction results are integrated with the output of the historical environment and weather forecast data prediction model to obtain the final heating demand prediction model, predict future heating demand, obtain integrated prediction results, and generate control instructions based on the integrated prediction results; then, based on the control instructions, the electric regulating valve is controlled to automatically adjust the valve opening and the heat pump output temperature according to the flow distribution ratio and the target heating temperature, and the variable frequency circulation pump adjusts the speed according to the heating strategy to achieve dynamic adjustment of the hot water flow and heating temperature of each regional branch to ensure heating balance; finally, the user interacts with the Internet of Things control platform through the user terminal device to view the real-time heating status and set the individual indoor heating.

[0094] It should be further explained that, in the specific implementation process, multi-source data is integrated, combined with historical environmental data, weather forecast data and user heating habit data to predict heating demand; and multiple sensors are arranged in various areas of the heating network to collect data, and the data is classified, cleaned, feature extracted and normalized, providing a rich and high-quality data basis for accurate prediction; it can obtain outdoor temperature, humidity, flow data in different time periods, as well as user's daily set temperature preferences, heating time patterns and other information, to fully reflect the factors affecting heating demand;

[0095] The heating demand prediction model is trained using a neural network model, and the complex mapping relationship between environmental data and heating demand is learned based on a machine learning algorithm to improve prediction accuracy. At the same time, the sliding average method is used to process weather forecast data, cluster analysis and sequence pattern mining algorithms are used to analyze user behavior patterns, and multivariate regression analysis is used to establish a comprehensive model to improve prediction accuracy from multiple dimensions, better adapt to changes in heating demand in different regions and users, and provide strong support for precise heating.

[0096] The execution layer control module dynamically adjusts the hot water flow and heating temperature of each regional branch through electric control valves and variable frequency circulation pumps according to the control instructions generated by the IoT control platform. The electric control valve automatically adjusts the valve opening and the heat pump output temperature according to the flow distribution ratio and the target heating temperature, and the variable frequency circulation pump adjusts the speed according to the heating strategy to achieve heating balance in each area, ensure stable and comfortable indoor temperature for users, and improve the stability and reliability of the heating system.

[0097] The electric control valve has automatic calibration and self-learning functions, and continuously optimizes the adjustment strategy based on the reinforcement learning algorithm. By real-time monitoring of flow and temperature data, the valve opening is accurately controlled according to the PID control algorithm to form a closed-loop control system, which can quickly respond to changes in heating demand, reduce energy waste, reduce energy consumption while ensuring heating quality, and improve energy utilization efficiency.

[0098] The system can monitor environmental parameters and user heating habits in real time, adjust heating strategies in a timely manner, quickly respond to fluctuations in heating demand, and deal with sudden changes in outdoor temperature or temporary adjustments to temperature settings by users, ensuring that heating always meets users' actual needs and improving users' overall experience of the heating system.

[0099] Based on big data analysis and intelligent algorithms, the IoT control platform accurately identifies differences in heating demand in different regions and generates scientific and reasonable control instructions, enabling the heating system to adapt to complex and changing environments and user needs, thereby improving the system's adaptability and robustness. The execution layer control module monitors key parameters of the heating network in real time and feeds back to the control platform. The control platform adjusts the control instructions based on the heating results, forming a dynamic optimization and adjustment mechanism to ensure the stable operation of the heating system, effectively respond to various internal and external interference factors, and ensure long-term stability of the heating quality.

[0100] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. An intelligent heating network optimization control system based on the Internet of Things, characterized in that: The system includes: Environmental parameter acquisition modules are distributed in different areas covered by the heating network. The environmental parameter acquisition modules include temperature sensors, humidity sensors and flow sensors, which are used to collect environmental data of each area in real time. The environmental data include outdoor temperature, humidity and flow. Each sensor interacts with the Internet of Things control platform through wireless communication. The Internet of Things control platform is used to receive environmental data from the environmental parameter acquisition module, analyze the environmental data based on the data analysis algorithm, identify and predict the heating demand of different areas, obtain integrated prediction results, and generate control instructions based on the integrated prediction results; The execution layer control module includes electric regulating valves and variable frequency circulation pumps distributed in each branch of the heating network. The execution layer control module is connected to the Internet of Things control platform for communication, receives control instructions from the Internet of Things control platform, and dynamically adjusts the hot water flow and heating temperature of each regional branch; User terminal devices interact with the IoT control platform. Users can view the real-time heating status and set individual indoor heating through user terminal devices; It also combines the user's heating habit data to make heating forecasts based on the user's behavior patterns, obtains integrated forecast results, and adjusts the heating parameters based on the integrated forecast results; The heating demand forecast results obtained based on the user behavior pattern analysis are integrated with the output of the heating demand forecast model based on historical environmental data and weather forecast data according to the final fusion strategy to obtain the integrated heating demand forecast model and obtain the final heating demand forecast value; The weighted average method is used to determine the weight of the prediction results of the user behavior pattern model based on the previous evaluation: , the weights of the prediction results of the historical environment and weather forecast data model are ,in, ; The final predicted value , the calculation formula is: ;in The heating demand predicted by the user behavior pattern model, heating demand predicted by models of historical environmental and weather forecast data; The integrated heat demand forecasting model is used to forecast the heat demand of each region in the next week or month to obtain the integrated forecasting results, which are then applied to the optimal control decision-making of the heating network system.

2. According to the intelligent heating network optimization control system based on the Internet of Things according to claim 1, it is characterized by: The Internet of Things control platform receives environmental data from each area, and processes the collected environmental data based on a preset data analysis algorithm to identify differences in heating demand in different areas and obtain heating demand in different areas; The data analysis algorithm includes a heating demand prediction model based on machine learning, and the heating demand prediction model is used to predict the heating demand of various regions in the next week or month based on historical environmental data and external weather forecast data.

3. According to the Internet of Things-based intelligent heating network optimization control system of claim 2, it is characterized by: The heating demand forecasting model predicts the heating demand of each region in the next week or month as follows: Obtain historical environmental data from temperature sensors, humidity sensors, and flow sensors installed in various areas of the heating network system, including outdoor temperature, humidity, and flow data in different time periods; classify and clean the historical environmental data to remove outliers and erroneous data; store the sorted historical environmental data by region and time series to form a historical environmental data set that can be used for analysis and model training; Obtain weather forecast data for the next week or month from professional meteorological agencies through the data receiving unit of the IoT control platform, including predicted outdoor temperature and humidity information; and use the sliding average method to remove short-term fluctuation noise in the data to obtain a weather forecast data set; Extract features related to heating demand from the historical environmental data set and the weather forecast data set to obtain relevant features, perform normalization processing, and map them to [0,1] to obtain historical environmental data features and weather forecast data features respectively; A neural network model is selected as the basic framework of the heating demand prediction model, and the number of hidden layers, the number of neurons in each layer, and the activation function are determined; including: setting a neural network with three hidden layers, the number of neurons in each layer is [64, 32, 16] respectively, and the activation function uses the ReLU corrected linear unit function; The processed historical environmental data features and the corresponding historical heating demand data are used as training samples and input into the constructed model for training. Stochastic gradient descent (SGD) is used to optimize and adjust the model parameters so that the model can learn the mapping relationship between environmental data and heating demand. During the training process, the data set is divided into a training set, a validation set, and a test set, with the ratio set to 70%:15%:15%. The performance indicators of the model are evaluated on the validation set: mean square error (MSE) and mean absolute error (MAE) to monitor whether the model is overfitting or underfitting. If the performance indicators on the validation set no longer improve or begin to deteriorate, the training is stopped to obtain a heating demand prediction model. Input weather forecast data features into the heating demand prediction model; The heat demand forecasting model calculates and outputs forecast results based on the learned patterns and rules. The forecast results include the forecast value of heat demand for each region in the corresponding time period in the future.

4. According to claim 3, the intelligent heating network optimization control system based on the Internet of Things is characterized by: Combined with the user's heating habit data, heating supply prediction is performed based on the user's behavior pattern. The process is as follows: Collect the heating habit information actively input by users through user terminal devices, collect indirect data generated by the interaction between users and the heating system, analyze the heating habit data, mine potential heating habit data from users' past heating bill information, clean the collected user heating habit data, remove invalid, erroneous or duplicate data records; and convert data of different formats and types into a unified format; extract relevant features based on the characteristics of user heating habit data and prediction results; combine environmental data and time information to construct composite features; Through cluster analysis, users with similar heating habits are grouped into the same category, and a typical behavior pattern model is established for each cluster. The sequence pattern mining algorithm is used to discover the user's heating operation sequence patterns at different time points, analyze their frequency and conditions of occurrence, and clarify the changing rules of users' dynamic heating demand under different situations; the average heating demand and heating demand fluctuation range under different behavior patterns are calculated through statistical analysis methods, and a mapping relationship model between behavior patterns and heating demand is established to clarify the quantitative relationship between different user behavior patterns and heating demand; At the same time, the impact of environmental factors on the relationship between user behavior patterns and heating demand is considered. A comprehensive model including outdoor temperature, humidity, user behavior pattern characteristics and heating demand in environmental variables is established through multivariate regression analysis. The model analyzes how user behavior patterns affect changes in heating demand under different environmental conditions, and obtains heating demand prediction results based on user behavior patterns. Determine strategies for integrating heating demand forecasts based on user behavior patterns with forecast models based on historical environmental data and weather forecast data; Fusion strategies include: weighted average method: assign weights according to the accuracy and importance of different models; stacked ensemble learning: use the prediction results of one model as the input features of another model for secondary prediction; Evaluate the performance of different fusion strategies on the training data set, and select the fusion strategy that optimizes the performance of the overall heating prediction model, including: the fusion strategy with the smallest prediction error and the highest stability; determine the final fusion strategy.

5. According to claim 4, an intelligent heating network optimization control system based on the Internet of Things is characterized in that: The execution layer control module is used to receive control instructions and dynamically adjust the heating parameters of the heating network according to the control instructions; including: adjusting the hot water flow and heating temperature of each regional branch by controlling the execution layer equipment to ensure the heating balance of each region, wherein the execution layer equipment includes an electric regulating valve, a variable frequency circulation pump and a heat pump; Wherein, the electric regulating valve has an automatic calibration function, which is used to automatically adjust to the optimal state according to real-time flow and temperature data; The electric regulating valve also has a self-learning function for optimizing its own regulating process by analyzing historical control data and current heating demand; The self-learning function is based on a reinforcement learning algorithm, which optimizes the adjustment strategy by learning the feedback between each adjustment operation and the heating result.

6. According to the Internet of Things-based intelligent heating network optimization control system of claim 5, it is characterized by: The process of the execution layer control module dynamically adjusting the heating parameters of the heating network according to the control instructions is as follows: The execution layer control module first receives the control instructions from the central control system through wireless communication; the control instructions include the set target heating temperature, flow distribution ratio and heating strategy, and the control instructions are parsed by the built-in microprocessor of the execution layer control module, and the control instructions are converted into operation signals; based on the operation signals, the execution layer control module sends control signals to the electric regulating valves and variable frequency circulation pumps in each area, where the electric regulating valves are installed at the hot water inlet of each branch, and are used to automatically adjust the valve opening and the output temperature of the heat pump according to the flow distribution ratio and the target heating temperature in the control signal, and obtain the opening adjustment parameters and temperature adjustment parameters respectively. The variable frequency circulation pump automatically adjusts the speed according to the heating strategy to obtain the speed adjustment parameters, and the primary heating adjustment parameters are formed based on the opening adjustment parameters, the temperature adjustment parameters and the speed adjustment parameters; and through the sensor unit equipped in the execution layer control module, the key parameters of the heating network are monitored in real time, and the key parameters include temperature parameters, pressure parameters and flow parameters; and the key parameters are fed back to the execution layer control module to evaluate the current heating result, obtain the heating result feedback, and adjust the control instructions based on the heating result feedback, generate new control instructions, and dynamically adjust the heating parameters of the heating network based on the new control instructions to obtain the secondary heating adjustment parameters.

7. According to claim 6, an intelligent heating network optimization control system based on the Internet of Things is characterized by: The process of the electric control valve automatically adjusting to the optimal state according to the real-time flow and temperature data is as follows: The temperature and flow data of hot water are monitored in real time by temperature sensors and flow sensors installed in each branch of the heating network. According to the heating demand prediction model and the indoor temperature target value and heating schedule set by the user terminal device, the IoT control platform calculates the ideal hot water flow and heating temperature target value corresponding to the current moment for each area or branch. After receiving the real-time flow and temperature data, the control platform compares it with the target value and calculates the flow deviation and temperature deviation. The flow deviation is the difference between the actual flow and the target flow, and the temperature deviation is the difference between the actual temperature and the target temperature. The built-in control algorithm of the electric control valve is based on the PID control algorithm in classical control theory. It calculates according to the flow deviation and temperature deviation to determine the opening adjustment of the control valve; the output control signal is calculated according to the three parts of the deviation: proportional P, integral I and differential D. The proportional term reflects the size of the current deviation and responds to the deviation in a timely manner; the integral term is used to eliminate the steady-state error of the system and integrates the accumulated deviation over time; the differential term predicts the system trend in advance according to the rate of change of the deviation and suppresses the overshoot of the system; the opening adjustment signal calculated by the control algorithm is transmitted to the drive actuator of the electric control valve: the motor; the motor controls the position of the valve core of the electric control valve according to the received signal and changes the opening size of the valve; after adjusting the opening, the electric control valve continues to monitor the flow and temperature data in real time, repeats the above deviation calculation, control algorithm operation and drive actuator adjustment process to form a closed-loop control system.

8. According to claim 7, the intelligent heating network optimization control system based on the Internet of Things is characterized by: The process of electric regulating valve to achieve more efficient heating flow control is as follows: Feature extraction: Extract features related to heating flow control from historical control data and current heating demand information; including: taking the outdoor temperature change rate, user-set temperature adjustment range, and adjacent branch flow change difference as input features, and taking the opening adjustment amount of the electric control valve as output features; Model training and optimization: Q-learning in the reinforcement learning algorithm is used to build a self-learning model for the electric control valve. The extracted feature data is input into the model, and the model is used to learn the optimal adjustment strategy under different conditions to maximize the long-term heating results and energy utilization efficiency. During the training process, the model parameters are continuously optimized so that the model can accurately predict the opening adjustment actions that the electric control valve should take under various working conditions to achieve a more efficient heating flow control goal. Working condition judgment and strategy selection: Based on the current real-time monitored heating demand data and environmental parameters, the trained self-learning model is used to judge the current working condition. The model selects the most suitable adjustment strategy for the current situation in the learned strategy space according to the input feature data: current outdoor temperature, user set temperature, and other branch flow conditions, and determines the opening adjustment direction and amplitude of the electric control valve. Execute regulation action: The electric regulating valve controls the valve core position by driving the actuator according to the regulation strategy determined by the model, thereby realizing real-time adjustment of the heating flow rate; at the same time, it works in coordination with the variable frequency circulation pump in the heating pipe network system to dynamically balance the flow distribution of each branch according to the overall heating demand of the system.

9. The intelligent heating network optimization control system based on the Internet of Things according to claim 8 is characterized in that: The process of achieving more efficient heating flow control by electric control valve also includes: Historical control data accumulation: During the operation of the heating system, the electric control valve continuously records the adjustment time of each adjustment operation, the flow value before adjustment, the flow value after adjustment, the corresponding temperature data, the outdoor ambient temperature at that time, the humidity data and the heating parameters set by the user: target temperature, heating schedule, and obtains historical control data; Current heating demand monitoring: The current heating demand information is obtained in real time through the IoT control platform, including the heating demand predicted for each area based on environmental parameters: outdoor temperature, humidity and user heating habits, as well as the heating settings adjusted in real time by the user through the terminal device. The heating settings include temporarily increasing or lowering the indoor temperature. The above information is combined to determine the actual heating demand changes in the branch where each electric control valve is located at the current moment, and obtain the current heating demand information.

10. A control method for an intelligent heating network optimization control system based on the Internet of Things according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Environmental data collection: Collect environmental data in real time in different areas covered by the heating network, including outdoor temperature, humidity and flow, and transmit the collected environmental data to the IoT control platform via wireless communication; Step 2: Prediction and analysis of heating demand: Collect historical environmental data and obtain future weather forecast data, then extract relevant features of heating demand, select a neural network model for training to obtain a heating demand prediction model; collect user heating habit data, extract features to establish a user behavior pattern model, clarify the quantitative relationship between behavior patterns and heating demand, and obtain heating demand prediction results based on user behavior; Step 3: Based on the fusion strategy, the user behavior prediction results are integrated with the output of the historical environment and weather forecast data prediction model to obtain the final heating demand prediction model, predict future heating demand, obtain the integrated prediction results, and generate control instructions based on the integrated prediction results; Step 4: Based on the control instruction, the electric regulating valve is controlled to automatically adjust the valve opening and the heat pump output temperature according to the flow distribution ratio and the target heating temperature. The variable frequency circulation pump adjusts the speed according to the heating strategy to achieve dynamic adjustment of the hot water flow and heating temperature of each regional branch to ensure the balance of heating supply; Step 5: Users interact with the IoT control platform through user terminal devices to view real-time heating status and set individual indoor heating.

Images