AI-driven steam-natural gas co-heating intelligent switching method
By employing an AI-driven intelligent switching method for steam-natural gas co-heating, which utilizes real-time data to predict load sequences and dynamically allocate strategies, the problem of response lag and insufficient economy in traditional control methods is solved, achieving efficient and stable switching between steam and natural gas.
Patent Information
- Application Number
- CN202610352790.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-23
- Publication Date
- 2026-07-03
AI Technical Summary
Traditional control methods cannot simultaneously take into account multiple economic and environmental indicators such as energy costs and carbon emissions. They are slow to respond and cannot predict load changes or energy price fluctuations, resulting in passive and uneconomical switching actions between steam and natural gas heating systems.
By employing an AI-driven intelligent switching method for steam-natural gas co-heating, real-time data is collected through an industrial gateway to construct an input feature vector. A time-series prediction model is used to predict the load sequence, and a reward function and intelligent decision-making model are designed to dynamically allocate steam and natural gas opening commands. Combined with dominant factors, heat source patterns are identified to achieve dynamic optimal allocation and smooth transition.
It achieves predictive feedforward control, dynamically and optimally allocates steam and natural gas, reduces temperature shocks and pressure fluctuations, optimizes energy costs, and improves response speed and system stability.
Smart Images
Figure CN122331380A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent energy control technology, specifically to an AI-driven intelligent switching method for steam-natural gas co-heating. Background Technology
[0002] In industrial heating sectors, such as petrochemical, food processing, and papermaking processes, steam heating and natural gas heating are the two main methods of heat energy supply. Steam heating typically originates from coal-fired boilers, gas-fired boilers, or combined heat and power (CHP) systems, offering advantages such as relatively stable costs and high safety, but it also suffers from inherent drawbacks like high thermal inertia and slow response speed. Natural gas heating, on the other hand, boasts advantages such as fast response speed, high control precision, and flexible start-up and shutdown, but its operating costs are significantly affected by fluctuations in natural gas market prices, and the combustion process generates emissions such as nitrogen oxides.
[0003] However, traditional control methods only aim to maintain temperature stability and cannot simultaneously take into account multiple economic and environmental indicators such as energy costs and carbon emissions. They are slow to respond and are reactive, adjusting only after temperature or pressure deviates. For steam systems with high thermal inertia, this can easily lead to overshoot or fluctuations. They also lack predictive capabilities and cannot anticipate future load changes or energy price fluctuations, resulting in passive and uneconomical switching actions. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an AI-driven intelligent switching method for steam-natural gas co-heating, which solves the problems of achieving prediction-based feedforward control and dynamic optimal allocation of steam and natural gas.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI-driven intelligent switching method for steam-natural gas co-heating, which specifically includes the following steps: Step 1: Collect historical heat load data, real-time operating condition data, and external constraint data in real time through an industrial gateway, and resample, filter, and fill in missing values in the collected data to obtain preprocessed data; Step 2: Construct an input feature vector based on the preprocessed data, input the input feature vector into the pre-trained time series prediction model, output the future heat prediction load sequence, and construct the heat prediction load curve based on the heat prediction load sequence; Step 3: Combine the real-time data at the current moment with the heat prediction load curve to form a state space, define the action space, and design a reward function. Based on the reward function, perform pre-training in the digital twin environment to obtain an intelligent decision model. In each control cycle, input the current state into the intelligent decision model and output the opening command of the steam valve and natural gas valve at the next moment. Step 4: Based on the opening command at the next moment, control the steam valve and natural gas valve to execute the corresponding opening degree, and monitor the deviation between the current heat load and the predicted value in real time. According to the magnitude of the deviation, perform graded regulation, calculate the total energy adjustment amount, and allocate the total energy adjustment amount to steam and natural gas according to the cost-response weighted allocation principle, and generate the corresponding valve control command. Step 5: Calculate the dominant factor based on real-time operating conditions, identify the current dominant heat source mode based on the dominant factor, dynamically adjust the target allocation ratio according to the identified mode, and execute the corresponding collaborative control strategy.
[0006] As a further aspect of the present invention, the historical heat load data includes temperature, flow rate, and pressure over the past N hours; the real-time operating condition data includes instantaneous pressure and temperature of the steam pipeline network, upper limit of available flow rate, pressure after the natural gas pressure regulating valve, calorific value, and current burner opening; and the external constraint data includes real-time energy unit price, carbon emission factor, and production schedule.
[0007] As a further aspect of the present invention, the method for constructing the input feature vector based on preprocessed data is as follows: Dynamic continuous features, discrete state features, and periodic time features are extracted from the preprocessed data. Dynamic continuous features include historical heat load sequences and ambient temperature and humidity. Discrete state features include production cycle signals and equipment start-stop status bits. The production cycle signal represents the ratio of the current production line operating speed to the rated speed. The periodic time features represent the mapping of hours and days of the week to continuous vectors using sine-cosine position encoding. All continuous features are Z-score standardized, and the processed static, dynamic and temporal feature vectors are concatenated along the channel dimension to construct the final input feature vector Xt.
[0008] As a further aspect of the present invention, the operating space includes the increment of the steam regulating valve opening. Increment of the opening of the natural gas regulating valve , specific T is the transpose sign, with a value range of [-1, 1]. reward function =-[ ],in , and These are the weighting coefficients. Indicates real-time energy cost, This indicates a penalty for temperature deviation. This indicates a penalty for smoothness of motion.
[0009] As a further aspect of the present invention, the ,in , These represent the real-time unit prices of natural gas and steam, respectively. , These represent the instantaneous flow rates of natural gas and steam, respectively. The temperature deviation penalty term is designed using a piecewise function, which applies when the actual heating temperature deviates from the set target temperature. When within the range of [-1℃, 1℃], =0, when the deviation exceeds this range. = , This is the temperature penalty coefficient, ranging from 0.5 to 2.0; This is the dead zone threshold allowed by the process.
[0010] As a further aspect of the present invention, the specific method for calculating the total energy adjustment amount by performing graded control based on the magnitude of the deviation is as follows: Obtain the corresponding real-time data, which includes the current heat load, natural gas flow / pressure, steam flow / pressure, system temperature, and equipment operating status. The specific current heat load is the deviation between the actual value and the predicted value. Compare the current heat load with the predicted value. If the current heat load is higher than the predicted value, it means that the energy supply intensity needs to be increased. Conversely, if the current heat load is lower than the predicted value, it means that the energy supply intensity needs to be reduced.
[0011] As a further aspect of the present invention, regarding the situation of increasing energy supply intensity: Obtaining heat load deviation And the heat load deviation = |actual heat load value - predicted heat load value|, and is graded based on the heat load deviation. If 0 < ≤ ,and , If the predicted heat load is indicated, it is considered a minor deviation; if < ≤ ,and If so, it is judged as a moderate deviation. > If so, it will be judged as a serious deviation; Determining the total increment After, and = The allocation is performed on steam and natural gas according to the cost-response weighted allocation principle, and an allocation function is constructed. , ,and The value range is [0, 1], where This indicates the change in the amount of heat required to be provided by the combustion of natural gas. This indicates the change in heat that needs to be provided by steam heat exchange. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented.
[0012] As a further aspect of the present invention, regarding the situation of reduced functional strength: Obtain the negative deviation amplitude And based on the negative deviation amplitude Perform hierarchical processing, if ≤ If <0, it is judged as a small negative deviation. < ≤ If it is, then it is judged as a moderate negative deviation. < If so, it is judged as a serious negative deviation; In determining the total reduction Then, according to the cost-response weighted allocation principle, the gas and natural gas are allocated, and an allocation function is constructed. , ,in This indicates the amount of natural gas change that needs to be reduced. This indicates the amount of change in steam heat that needs to be reduced. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented.
[0013] As a further aspect of the present invention, the specific method for calculating the dominant factor based on real-time operating conditions and identifying the current dominant heat source mode based on the dominant factor is as follows: According to the formula The dominant factor was calculated. ,in The rate of change of heat load, The maximum allowable / historical maximum load change rate, This is the upper limit of the safe pressure for the steam pipeline network. This represents the current real-time pressure of the steam pipeline network. Indicates the pressure safety margin, This indicates the cost of steam per unit of calorific value. This indicates the cost of natural gas per unit of calorific value. This indicates the maximum possible price difference reference value. , and The corresponding weight coefficients will be used to obtain the dominant factors. Comparison threshold and If a comparison is made, > If natural gas is determined to be the primary heat source, < If steam is determined to be the dominant heat source, < < If the ratio is not specified, it is determined to be a mixed synergistic mode, and the current ratio is maintained.
[0014] As a further aspect of the present invention, if natural gas is the dominant heat source, based on the calculated dominant factor... Calculate the target natural gas allocation ratio, specifically according to the formula. Perform calculations, where To achieve the minimum stable combustion ratio, This refers to the mapping gain coefficient; If steam is the dominant heat source, based on the calculated dominant factor Calculate the target steam load ratio, specifically according to the formula. Perform calculations. Indicates the minimum stable combustion ratio. Indicates the current base ratio. This represents the descent mapping gain coefficient.
[0015] This invention provides an AI-driven intelligent switching method for steam-natural gas co-heating. Compared with existing technologies, it has the following advantages: This invention dynamically adjusts the allocation coefficient to distribute the total energy adjustment amount to steam and natural gas according to the cost-response weighted allocation principle. When the price of natural gas is low or the response speed requirement is high, the value is increased, increasing the proportion of natural gas bearing the load; when the cost advantage of steam is obvious or a stable base load is required, the value is decreased, increasing the proportion of steam bearing the load. This dynamic allocation mechanism fully leverages the complementary advantages of natural gas's fast response and steam's low cost, minimizing the overall energy cost while ensuring process requirements. By calculating the dominant factor and comparing thresholds, the current mode (steam-dominated, natural gas-dominated, or hybrid collaborative) is accurately identified. When switching between different modes is required, a smooth transition trajectory is planned through a preset ramp function or first-order inertial link, avoiding temperature shocks and pressure fluctuations in traditional switching methods. Especially in the natural gas-dominated mode, a minimum stable opening maintenance and hot standby management strategy for the steam system is designed to ensure a rapid response when switching back to the collaborative mode is required, achieving seamless switching. Attached Figure Description
[0016] Figure 1 This is a flowchart of the AI-driven intelligent switching method for steam-natural gas co-heating according to the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Please see Figure 1 This application provides an AI-driven intelligent switching method for steam-natural gas co-heating, which specifically includes the following steps: Step 1: Collect historical heat load data, real-time operating condition data, and external constraint data in real time through an industrial gateway. The historical heat load data includes temperature, flow rate, and pressure over the past N hours. The real-time operating condition data includes instantaneous pressure, temperature, available flow limit, pressure after the natural gas pressure regulating valve, calorific value, and current burner opening. The external constraint data includes real-time energy unit price, carbon emission factor, and production schedule. Then, the collected data is resampled to a standard time step using an interpolation algorithm. At the same time, a sliding window filter is used to remove abnormal noise from the sensors, and missing data is filled in based on the mean of the nearest time point to obtain preprocessed data. Step 2: Based on the obtained preprocessed data, extract dynamic continuous features, discrete state features, and periodic time features. Dynamic continuous features include historical heat load sequences and ambient temperature and humidity. Discrete state features include production cycle signals and equipment start / stop status positions. The production cycle signal represents the ratio of the current production line operating speed to the rated speed. Periodic time features are represented by using sine-cosine position encoding to map hours and days of the week into continuous vectors. Then, Z-Score normalization is performed on all continuous features. The processed static, dynamic, and time feature vectors are concatenated along the channel dimension to construct the final input feature vector Xt. The obtained feature vector Xt is input into a pre-trained temporal prediction model, which is a Long Short-Term Memory network optimized based on an attention mechanism. This temporal prediction model consists of an input layer, an LSTM layer, an attention layer, and a fully connected output layer. The input layer accepts data of shape (batch size). sizeThe tensor is defined as follows: L, F, where F is the number of features. The first LSTM layer has 64 neurons, the second LSTM layer has 64 neurons, the attention layer performs a weighted summation of the output of the second LSTM layer, the dropout layer adds dropout after the attention layer with a dropout rate of 0.2, and the fully connected output layer has the same number of neurons as the prediction step size, uses a linear activation function, and outputs the predicted heat load values for the next H time steps. A sliding window mechanism is used to output the predicted heat load sequence for the next M time steps at once, and the corresponding predicted heat load curve is constructed based on the obtained predicted heat load sequence.
[0019] Step 3: Construct a state space by combining the real-time data at the current time t with the predicted heat load curve, and define the action A(t). T is the transpose symbol, where This represents the increment of the steam control valve opening, with a value range of [-1, 1]. Here, 1 represents the maximum normalized unit of valve opening change, i.e., the maximum rate of change when fully open or fully closed, and -1 represents the maximum rate of change in the opposite direction. Design a reward function for the increment of the natural gas regulating valve opening. =-[ ],in , and These are the weighting coefficients. Indicates immediate energy cost, and Real-time energy costs The calculation can be further extended to include the cost of carbon emissions, i.e. ,in For carbon price, , Here are the carbon emission factors per unit calorific value for natural gas and steam, respectively. , These represent the real-time unit prices of natural gas and steam, respectively. , These represent the instantaneous flow rates of natural gas and steam, respectively. This indicates a penalty for temperature deviation. This represents the motion smoothness penalty, where the temperature deviation penalty uses a piecewise function design. This applies when the actual heating temperature deviates from the set target temperature. When within the range of [-1℃, 1℃], =0, when the deviation exceeds this range. = , This is the temperature penalty coefficient, ranging from 0.5 to 2.0; To determine the dead zone threshold allowed by the process, the system parses and maps the production schedule into a state space. The core feature of the model is the production rhythm signal, which serves as a key input to the intelligent decision-making model. It participates in the model's training iteration and real-time inference, guiding the model to generate a heat source switching strategy that matches the production rhythm. Based on the above reward function, pre-training is performed in the digital twin environment. The training objective is to maximize the cumulative discount reward. After training, the model is deployed to the actual control system. In each control cycle, the model outputs the optimal action A(t) based on the current state S(t), which is the opening command of the steam valve and the natural gas valve at the next moment. Step 4: Based on the obtained opening command for the next moment, the system issues corresponding execution actions to control the steam valve and natural gas valve to open accordingly. At the same time, it acquires the corresponding real-time data, which includes the current heat load, natural gas flow / pressure, steam flow / pressure, system temperature, and equipment operating status. The specific current heat load is the deviation between the actual value and the predicted value. The current heat load is compared with the predicted value. If the current heat load is higher than the predicted value, it means that the energy supply intensity needs to be increased. Conversely, if the current heat load is lower than the predicted value, it means that the energy supply intensity needs to be reduced. For situations requiring enhanced functional strength, obtain the heat load deviation. And the heat load deviation = |actual heat load value - predicted heat load value|, and is graded based on the heat load deviation. If 0 < ≤ ,and , If the predicted heat load is considered a minor deviation, then only the natural gas side will be adjusted according to the formula. The gas valve opening is calculated and adjusted accordingly; if < ≤ ,and If the deviation is too large, the natural gas valve opening will increase by a larger percentage, while the steam valve opening will increase slowly according to a preset ramp function. > If this is not observed, it is considered a serious deviation, and both the natural gas valve and the steam valve open simultaneously at their maximum permissible rate of change, while simultaneously introducing a feedforward overshoot coefficient. Set target for energy supply ,in This represents the actual heat load demand at the current moment. Determining the total increment After, and = The allocation is performed on steam and natural gas according to the cost-response weighted allocation principle, and an allocation function is constructed. , ,in This indicates the change in the amount of heat required to be provided by the combustion of natural gas. This indicates the change in heat that needs to be provided by steam heat exchange. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented. For situations requiring a reduction in functional strength, calculate the negative deviation amplitude. , here This represents the negative deviation amplitude, that is, its value relative to the heat load deviation. Instead, it is a calculated value based on the negative deviation magnitude. Perform hierarchical processing, if ≤ If the value is less than 0, it is considered a small negative deviation. The valve opening is calculated, and the gas valve opening is reduced based on the calculated valve opening, while the natural gas valve opening remains unchanged. < ≤ If the deviation is deemed as a moderate negative deviation, the natural gas valve will be rapidly closed to below the baseline load, or even temporarily closed, while the steam valve will be slowly closed according to a preset ramp function. < If the deviation is too large, it is determined to be a serious negative deviation. The natural gas valve will immediately execute a rapid shut-off command, and the steam valve will close at the maximum permissible shut-off rate. In determining the total reduction Then, according to the cost-response weighted allocation principle, the gas and natural gas are allocated, and an allocation function is constructed. , ,in This indicates the amount of natural gas change that needs to be reduced. This indicates the amount of change in steam heat that needs to be reduced. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented.
[0020] Step 5: Based on the calculated target allocation, priority arbitration is first performed according to the formula. The dominant factor was calculated. ,in The rate of change of heat load, The maximum allowable / historical maximum load change rate, This is the upper limit of the safe pressure for the steam pipeline network. This represents the current real-time pressure of the steam pipeline network. Indicates the pressure safety margin, This indicates the cost of steam per unit of calorific value. This indicates the cost of natural gas per unit of calorific value. This indicates the maximum possible price difference reference value. , and These are the corresponding weight coefficients, satisfying... + + =1, and its specific value is set according to the priority requirements of safety, response speed and economic cost in the actual production process, or it is obtained through training and optimization using historical data. The resulting dominant factor is... Comparison threshold and If a comparison is made, > If natural gas is determined to be the primary heat source and steam is the secondary / base load heat source, then... < If steam is determined to be the primary heat source and natural gas is the regulating / peak-shaving heat source, then... < < If it is determined to be a mixed synergistic mode, the current ratio will be maintained, and the dominant factor obtained will be determined. Comparison threshold and If a comparison is made, > If natural gas is determined to be the primary heat source and steam is the secondary / base load heat source, then... < If steam is determined to be the primary heat source and natural gas is the regulating / peak-shaving heat source, then... < < If so, it is determined to be a mixed synergistic mode, and the current ratio is maintained; If natural gas is the dominant heat source, based on the calculated dominant factor Calculate the target natural gas allocation ratio, specifically according to the formula. Perform calculations, where To achieve the minimum stable combustion ratio, This refers to the mapping gain coefficient; If steam is the dominant heat source, based on the calculated dominant factor Calculate the target steam load ratio, specifically according to the formula. Perform calculations. Indicates the minimum stable combustion ratio. Indicates the current base ratio. This represents the decrease in mapping gain coefficient, and the minimum stable combustion ratio. It is an empirical value determined based on the lower limit of safe operation of the steam generator, with a preferred range of [0.25, 0.4]. The specific value can be set according to the minimum stable output curve provided by the equipment manufacturer, or obtained through regression analysis of long-term operating data. Decrease mapping gain coefficient This is used to adjust the influence of the degree of deviation of the dominant factor on the steam load ratio. Its value needs to take into account both response speed and system stability, and the value range is [0.8, 1.5]. Specifically, it can be set through the following steps: The initial value is set to 1.0. A step test is performed under typical negative deviation conditions to observe whether the steam ratio adjustment converges smoothly. If oscillation occurs, the value is reduced. If the response is delayed, increase the value until optimal dynamic performance is achieved. Alternatively, it can be determined based on physical parameters such as steam network volume and valve response time, using empirical formulas. A preliminary estimate was made, among which Indicates system response time. Indicates the control cycle.
[0021] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.
[0022] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.
Claims
1. An AI-driven intelligent switching method for steam-natural gas co-heating, characterized in that, The method specifically includes the following steps: Step 1: Collect historical heat load data, real-time operating condition data, and external constraint data in real time through an industrial gateway, and resample, filter, and fill in missing values in the collected data to obtain preprocessed data; Step 2: Construct an input feature vector based on the preprocessed data, input the input feature vector into the pre-trained time series prediction model, output the future heat prediction load sequence, and construct the heat prediction load curve based on the heat prediction load sequence; Step 3: Combine the real-time data at the current moment with the heat prediction load curve to form a state space, define the action space, and design a reward function. Based on the reward function, perform pre-training in the digital twin environment to obtain an intelligent decision model. In each control cycle, input the current state into the intelligent decision model and output the opening command of the steam valve and natural gas valve at the next moment. Step 4: Based on the opening command at the next moment, control the steam valve and natural gas valve to execute the corresponding opening degree, and monitor the deviation between the current heat load and the predicted value in real time. According to the magnitude of the deviation, perform graded regulation, calculate the total energy adjustment amount, and allocate the total energy adjustment amount to steam and natural gas according to the cost-response weighted allocation principle, and generate the corresponding valve control command. Step 5: Calculate the dominant factor based on real-time operating conditions, identify the current dominant heat source mode based on the dominant factor, dynamically adjust the target allocation ratio according to the identified mode, and execute the corresponding collaborative control strategy.
2. The AI-driven intelligent switching method for steam-natural gas co-heating as described in claim 1, characterized in that, The historical heat load data includes temperature, flow rate, and pressure over the past N hours. The real-time operating condition data includes instantaneous pressure and temperature of the steam pipeline network, upper limit of available flow rate, pressure after the natural gas pressure regulating valve, calorific value, and current burner opening. External constraint data includes real-time energy unit price, carbon emission factor, and production schedule.
3. The AI-driven intelligent switching method for steam-natural gas co-heating as described in claim 1, characterized in that, The method for constructing the input feature vector based on preprocessed data is as follows: Dynamic continuous features, discrete state features, and periodic time features are extracted from the preprocessed data. Dynamic continuous features include historical heat load sequences and ambient temperature and humidity. Discrete state features include production cycle signals and equipment start-stop status bits. The production cycle signal represents the ratio of the current production line operating speed to the rated speed. The periodic time features represent the mapping of hours and days of the week to continuous vectors using sine-cosine position encoding. All continuous features are Z-score standardized, and the processed static, dynamic and temporal feature vectors are concatenated along the channel dimension to construct the final input feature vector Xt.
4. The AI-driven intelligent switching method for steam-natural gas co-heating as described in claim 1, characterized in that, The operating space includes the increment of the steam regulating valve opening. Increment of the opening of the natural gas regulating valve , specific T is the transpose sign, with a value range of [-1, 1]. reward function =-[ ],in , and These are the weighting coefficients. Indicates real-time energy cost, This indicates a penalty for temperature deviation. This indicates a penalty for smoothness of motion.
5. The AI-driven intelligent switching method for steam-natural gas co-heating as described in claim 4, characterized in that, The ,in , These represent the real-time unit prices of natural gas and steam, respectively. , These represent the instantaneous flow rates of natural gas and steam, respectively. The temperature deviation penalty term is designed using a piecewise function, which applies when the actual heating temperature deviates from the set target temperature. When within the range of [-1℃, 1℃], =0, when the deviation exceeds this range. = , This is the temperature penalty coefficient, ranging from 0.5 to 2.0; This is the dead zone threshold allowed by the process.
6. The AI-driven intelligent switching method for steam-natural gas co-heating according to claim 1, characterized in that, The specific method for calculating the total energy adjustment amount by performing graded control based on the magnitude of the deviation is as follows: Obtain the corresponding real-time data, which includes the current heat load, natural gas flow / pressure, steam flow / pressure, system temperature, and equipment operating status. The specific current heat load is the deviation between the actual value and the predicted value. Compare the current heat load with the predicted value. If the current heat load is higher than the predicted value, it means that the energy supply intensity needs to be increased. Conversely, if the current heat load is lower than the predicted value, it means that the energy supply intensity needs to be reduced.
7. The AI-driven intelligent switching method for steam-natural gas co-heating according to claim 6, characterized in that, Regarding the situation of increasing energy supply intensity: Obtaining heat load deviation And the heat load deviation = |actual heat load value - predicted heat load value|, and is graded based on the heat load deviation. If 0 < ≤ ,and , If the predicted heat load is indicated, it is considered a minor deviation; if < ≤ ,and If so, it is judged as a moderate deviation. > If so, it will be judged as a serious deviation; Determining the total increment After, and = The allocation is performed on steam and natural gas according to the cost-response weighted allocation principle, and an allocation function is constructed. , ,and The value range is [0, 1], where This indicates the change in the amount of heat required to be provided by the combustion of natural gas. This indicates the change in heat that needs to be provided by steam heat exchange. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented.
8. The AI-driven intelligent switching method for steam-natural gas co-heating according to claim 6, characterized in that, Regarding situations where functional strength is reduced: Obtain the negative deviation amplitude And based on the negative deviation amplitude Perform hierarchical processing, if ≤ If <0, it is judged as a small negative deviation. < ≤ If it is, then it is judged as a moderate negative deviation. < If so, it is judged as a serious negative deviation; In determining the total reduction Then, according to the cost-response weighted allocation principle, the gas and natural gas are allocated, and an allocation function is constructed. , ,in This indicates the amount of natural gas change that needs to be reduced. This indicates the amount of change in steam heat that needs to be reduced. This represents the proportion of natural gas supplied at time t, based on calculations. and Adjustments and controls were implemented.
9. The AI-driven intelligent switching method for steam-natural gas co-heating as described in claim 1, characterized in that, The specific method for calculating the dominant factor based on real-time operating conditions and identifying the current dominant heat source mode based on the dominant factor is as follows: According to the formula The dominant factor was calculated. ,in The rate of change of heat load, The maximum allowable / historical maximum load change rate, This is the upper limit of the safe pressure for the steam pipeline network. This represents the current real-time pressure of the steam pipeline network. Indicates the pressure safety margin, This indicates the cost of steam per unit of calorific value. This indicates the cost of natural gas per unit of calorific value. This indicates the maximum possible price difference reference value. , and The corresponding weight coefficients will be used to obtain the dominant factors. Comparison threshold and If a comparison is made, > If natural gas is determined to be the primary heat source, < If steam is determined to be the dominant heat source, < < If the ratio is not specified, it is determined to be a mixed synergistic mode, and the current ratio is maintained.
10. The AI-driven intelligent switching method for steam-natural gas co-heating according to claim 9, characterized in that, If natural gas is the dominant heat source, based on the calculated dominant factor Calculate the target natural gas allocation ratio, specifically according to the formula. Perform calculations, where To achieve the minimum stable combustion ratio, This refers to the mapping gain coefficient; If steam is the dominant heat source, based on the calculated dominant factor Calculate the target steam load ratio, specifically according to the formula. Perform calculations. Indicates the minimum stable combustion ratio. Indicates the current base ratio. This represents the descent mapping gain coefficient.