Multi-gas boiler collaborative group control method and device based on gas consumption prediction and scheduling optimization

Abstract

The application discloses a kind of based on steam prediction and scheduling optimization's multi-gas boiler collaborative group control method, comprising: real-time acquisition industrial heating field main pipe pressure, steam flow, each gas boiler operating condition data;Based on historical steam consumption data establishes the load prediction model based on extreme gradient boosting tree XGBoost and one-dimensional Kalman filter fusion, combined with feedback control, output total control quantity;According to the current boiler operating state and the total control quantity obtained, the number of boilers based on the operating boundary is used to make a decision, the shortest time constraint and gradual adjustment strategy are set, and the number of boilers required at present is output;The multi-objective load optimization allocation problem is solved by using the algorithm of the dung beetle optimization, and the optimal load set value of each online unit is dynamically output with load tracking accuracy, operation efficiency, control smoothness and sparsity as optimization objectives and the number of boilers as constraints.The application can effectively suppress the main pipe pressure fluctuation and improve the system operation efficiency.

Description

Technical Field

[0001] This invention belongs to the field of industrial heating equipment control, and particularly relates to a method and device for collaborative group control of multiple gas-fired boilers based on steam consumption prediction and scheduling optimization. Background Technology

[0002] In modern light industrial manufacturing, heat pipe steam turbines are gradually replacing traditional small and medium-sized coal-fired or gas-fired boilers due to their advantages such as small size, rapid steam production, and high thermal efficiency. In actual plant projects, the rated steam production of a single heat pipe steam turbine is usually around 1 ton / hour. To meet the large-scale and significantly fluctuating steam demand, heating systems generally adopt a clustered topology structure in which multiple heat pipe steam turbines are connected in parallel to produce steam, which is then aggregated into a steam header and supplied to end users.

[0003] However, due to the diverse types of downstream steam users and the complexity of their processes, the total steam load of the system often exhibits strong randomness, abrupt changes, and severe fluctuations. Currently, most industrial plants still use the most basic single-unit independent proportional-integral-derivative (PID) control strategy for controlling multiple heat pipe steam turbines. Under this control method, there is a lack of information exchange and coordination mechanisms between the units. When faced with sudden changes in downstream steam demand, this often leads to severe supply-demand mismatch, causing a significant drop in mains pressure or multiple boilers competing for load and oscillating.

[0004] Meanwhile, heat pipe steam turbines have an engineering drawback of approximately 2 minutes of cold-start delay. Once shut down and entering a cold state, even if a start-up command is received, it still takes at least 2 minutes of physical delay to heat up to the rated steam production level. Without coordinated planning, when the end-user steam load suddenly increases, the boiler in a shutdown state cannot instantly fill the gap, and the start-up delay will directly lead to a severe drop in the main pipe pressure.

[0005] To overcome pressure fluctuations and energy efficiency issues in boiler control, numerous scholars have conducted in-depth research in recent years. Regarding advanced control of individual boilers, related studies have proposed methods such as improved implicit generalized predictive control, the introduction of autoregressive moving average models, internal model control, and adaptive automatic power generation control. These methods are used to overcome large system inertia, mitigate the effects of large delays, stabilize main steam pressure, and improve the robustness of individual systems. However, these methods are mainly designed for large gas-fired power boilers or coal-fired units and are difficult to adapt to the characteristics of small-capacity heat pipe steam turbine clusters.

[0006] In the area of ​​multi-boiler collaborative control under a main control system, existing research employs a hybrid strategy combining fmincon constrained nonlinear optimization solvers and ant colony optimization algorithms. This strategy incorporates multilayer perceptron prediction models, energy loss mathematical models, model-based predictive control optimization strategies, multivariate decoupling algorithms, bias allocation function blocks, and P-graph frameworks. These methods have achieved load optimization and energy consumption reduction to some extent. However, most of these approaches are designed for large industrial boilers with significant thermal inertia and strong coupling, or interdisciplinary slow-response equipment. Directly applying these methods to small-capacity heat pipe steam turbine clusters presents certain limitations.

[0007] Although existing studies have partially incorporated load forecasting or equipment life constraints, they have not fully considered the supply and demand time difference caused by the cold start delay of heat pipe steam engines, as well as the coordination problem between the discrete start-up and shutdown planning of multiple boilers and the continuous load distribution logic. As a result, under extreme variable load conditions, the system still has problems such as large fluctuations in main pipe pressure, frequent equipment start-up and shutdown, and low energy efficiency, which cannot meet the actual needs of industrial sites. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a method and device for collaborative group control of multiple gas-fired boilers based on steam consumption prediction and scheduling optimization. This method can solve the technical problems of lack of coordination in existing multi-gas-fired boiler cluster control, delayed response to load changes, cold start delay leading to fluctuations in main pipe pressure, and low operating efficiency. It achieves stable main pipe pressure, efficient collaborative operation of boilers, and energy saving and carbon reduction.

[0009] The objective of this invention is achieved through the following technical solution: A collaborative group control method for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization includes the following steps: S1: Real-time collection of main pipe pressure, steam flow, and operating status data of various gas boilers in industrial heating sites; S2: Based on historical steam consumption data, a load prediction model based on the fusion of extreme gradient boosting tree XGBoost and one-dimensional Kalman filter is established. Combined with feedback control, the total control quantity is output to realize the early perception of sudden load changes at the back end. S3: Based on the current boiler operating status and the total control quantity obtained in S2, adopt the boiler number redundancy decision method based on the operating boundary, set the minimum start-up and shutdown time constraint and the gradual boiler number adjustment strategy, and output the current required number of boilers. S4: The dung beetle optimization algorithm is used to solve the multi-objective load optimization allocation problem. The optimization objectives are load tracking accuracy, operating efficiency, control smoothness and sparsity. The number of boilers output by S3 is used as a constraint to dynamically output the optimal load setpoint for each online unit. S5: Sends real-time load control commands to each boiler unit via PLC.

[0010] A collaborative group control device for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization includes one or more processors for implementing a collaborative group control method for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization.

[0011] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements a collaborative group control method for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization.

[0012] The beneficial effects of this invention are as follows: This invention addresses the shortcomings of existing multi-gas-fired boiler cluster control technologies by proposing a three-layer collaborative control approach centered on "prediction-scheduling-optimization." First, a load prediction model based on the fusion of extreme gradient boosting tree (XGBoost) and one-dimensional Kalman filtering, combined with PID feedback control, is used to anticipate sudden changes in downstream steam load and output precise total control values. Then, a boiler redundancy decision logic based on operational boundaries is introduced to avoid frequent boiler start-ups and shutdowns and compensate for cold start delays. Finally, the Dung Beetle Optimization (DBO) algorithm is employed to solve the multi-objective load optimization allocation problem, dynamically allocating the load to each boiler. This provides crucial technical support for the stable and efficient operation of multi-gas-fired boiler clusters. This method eliminates the need for complex control logic reconfiguration, is suitable for the operating characteristics of small-capacity gas-fired heat pipe steam turbine clusters, effectively suppresses main pipe pressure fluctuations, extends equipment lifespan, improves system operating efficiency, and achieves energy conservation and carbon reduction. Attached Figure Description

[0013] Figure 1 The flowchart shows the collaborative group control method for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization according to the present invention.

[0014] Figure 2 This is a schematic diagram of a boiler system to be controlled and optimized in an embodiment of the present invention.

[0015] Figure 3 This is a schematic diagram illustrating the prediction effect of the prediction model on steam consumption in an embodiment of the present invention.

[0016] Figure 4 This is a schematic diagram illustrating the effect of sparse loss and regularization in the objective function on boiler load commands over time in an embodiment of the present invention.

[0017] Figure 5 This is a schematic diagram illustrating the change in the number of boiler start-ups and shutdowns over time under the control of the group control algorithm and the traditional single-machine independent PID control method in this embodiment of the invention.

[0018] Figure 6 This is a schematic diagram illustrating the changes in the head pipe pressure control effect and load tracking performance of the group control algorithm over time in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The purpose and effects of the present invention will become clearer. It should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0020] like Figure 1 As shown, on the one hand, the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization of the present invention includes the following steps one to five.

[0021] Step 1: Real-time collection of data on main pipe pressure, steam flow, and operating conditions of each gas-fired boiler at the industrial heating site.

[0022] Acquire historical steam flow data of the main pipe of the industrial heating system at the second level, with a data sampling interval of no more than 1 minute.

[0023] Step 2: Based on historical steam consumption data, establish a load prediction model that combines extreme gradient boosting tree (XGBoost) with one-dimensional Kalman filtering. Combined with feedback control, output the total control quantity to achieve early detection of sudden load changes at the back end.

[0024] The output of the load forecasting model is the predicted steam consumption at a future time, such as the predicted steam consumption at the 10th second in the future.

[0025] During the training process of the load prediction model, a combination of grid search and cross-validation is used to optimize the hyperparameters.

[0026] Step two includes the following sub-steps: S2.1: Use a sliding window method to extract data and predict the steam consumption at a certain time after the current time.

[0027] Let the prediction step size be... Then, at control time t, the predicted output obtained based on the historical steam usage data up to the current time is denoted as... , indicating that at time t, Predicted steam consumption at all times.

[0028] S2.2: Predicted values Considering this as a "noisy observation" of future steam consumption, a one-dimensional Kalman filter is introduced to perform online correction and smoothing of the XGBoost prediction results. The specific steps are as follows: S2.2.1: By statistically analyzing adjacent timestamps in history... The variance of the difference between the actual steam consumption and the actual steam usage is used as the process noise variance. Q This characterizes the degree of fluctuation in gas consumption itself. in, This represents the actual amount of steam used at time t. express Real-time steam consumption; This is the operator for calculating variance.

[0029] By statistically analyzing the effective samples from the W most recent control cycles prior to the current time t, the residual variance R is calculated within the sliding window: Where 's' represents a historical moment in the sliding window. Indicates in The output of the XGBoost model at each moment is targeted The predicted value of steam consumption at any given time; W represents the length of the sliding window.

[0030] S2.2.2: Real-time iterative update algorithm. Within each control cycle, the fusion process consists of two steps: time update (prediction) and state update (correction). First, calculate the prior estimate of the error covariance. : In the formula, M t Let be the posterior error covariance at time t.

[0031] Then calculate the Kalman gain (dynamic weight α). t+1 This gain dynamically measures the confidence level between the predicted value and the current empirical value: Then the optimal fusion value is calculated, where This is the fused prediction value after filtering: Finally, update the posterior error covariance at time t+1: S2.3: To eliminate the steady-state error of the system and cope with unmeasurable external disturbances, the feedback regulation of the positional PID controller is superimposed on the optimal fused feedforward quantity, and the total load requirement of the system is finally calculated. P t : In the formula, k ff For feedforward gain, u PID (t) is t The discretization formula for the output of a time-position PID controller is as follows: in, for t The difference between the set value of the main pipe pressure or temperature and the current actual measured value at any given time; The sampling period of the control system; K p , K i , K d These are the proportional, integral, and derivative coefficients of the controller, respectively.

[0032] Step 3: Based on the current boiler operating status and the total control quantity obtained in Step 2, adopt the boiler redundancy decision method based on the operating boundary, set the minimum start-up and shutdown time constraint and the gradual boiler number adjustment strategy, and output the current required number of boilers.

[0033] Step S3 specifically includes the following sub-steps: S3.1: Based on the load range of a single boiler Calculate the minimum number of units required for full-load operation. N min Maximum number of units operating at minimum load N max : In the formula, and These represent rounding up and rounding down, respectively.

[0034] S3.2: Adopt a gradual strategy of adding or removing furnaces and a redundancy judgment mechanism to adjust the target number of operating furnaces. N t : Furnace addition judgment (continuous) Triggered in seconds): if continued Then let Furnace Reduction Judgment (Continuing) Triggered in seconds): if continued Then let in, N run This represents the current number of running machines. T min This is the minimum start / stop time threshold (unit: seconds). n This is the margin coefficient.

[0035] The furnace reduction strategy is a gradual reduction strategy, aiming to reduce the target number of furnaces in operation. Nt Adjust the load to the average of the theoretical maximum value and the current number of units to avoid drastic load fluctuations.

[0036] When the total system load approaches 40% of the upper limit supported by the current number of operating boilers, a redundancy judgment mechanism is triggered to increase the number of operating boilers in advance to ensure that the system has sufficient adjustment margin. if continued Then let in, p i It is the first i The load value of the boiler. T threshold It is the redundancy judgment time threshold.

[0037] Step 4: Use the dung beetle optimization algorithm to solve the multi-objective load optimization allocation problem. The optimization objectives are load tracking accuracy, operating efficiency, control smoothness and sparsity. The number of boilers output in Step 3 is used as a constraint to dynamically output the optimal load setpoint for each online unit.

[0038] The objective optimization function includes four terms: load tracking loss, efficiency loss, distribution loss, and L1 regularization, expressed as follows: In the formula, J To optimize the function value for multiple objectives, k 1. k 2. k 3 represents the weighting coefficients for each loss term; Load tracking loss characterizes the deviation between the actual allocated total load and the target total load; η avg For efficiency loss; η ( p i ) represents the fitted value of the efficiency curve for a single boiler; l div To mitigate losses, adopt The function takes its maximum value when the independent variable is 0; This represents the difference between the current control cycle and the previous control cycle for the i-th boiler; l L1 This is a regularization term.

[0039] The constraints of the objective function are: Where I(⋅) is an indicator function, which takes the value 1 when the condition in the parentheses is true, and takes the value 0 otherwise.

[0040] In the design of this objective function, η avg Punish the boiler by operating it at its inefficient point. l div Punish "changes that shouldn't have changed" (small fluctuations). l L1 By punishing "too many" changes, the optimal distribution of loads across multiple boilers can ultimately be achieved under stable main pipe pressure.

[0041] The Dung Beetle Optimization (DBO) algorithm, in each iteration, first calculates the population fitness and then assigns four roles—rolling dung beetles, female dung beetles, small dung beetles, and thieving dung beetles—according to a preset ratio, updating their positions accordingly. In this algorithm, the proportions of the four species are 20%, 20%, 25%, and 35%, respectively.

[0042] The first type of role is the dung beetle, which is responsible for extensive exploration throughout the search space. It updates its position based on the current worst-case position (simulating light source offset); when it encounters an obstacle, it redirects by "dancing".

[0043] The formula for updating the position of the rolling dung beetle is as follows: The update formula for the dung beetle's dance when it encounters an obstacle is: in, For the first i Only the rolling dung beetle in the first tj The position of the next iteration corresponds to the total The load of the boiler is from one 3D vector representation; X w This is the worst position globally. The direction coefficient is used to simulate the deviation in motion direction caused by natural environmental factors; its value is determined by random numbers. Probability threshold of deviation from direction Decide: in, It is a random number. This is the probability threshold for directional deviation. In this algorithm, it is set... ; The deflection coefficient;b A constant between (0,1); It is a random deflection angle.

[0044] The second type of female dung beetle will be in the optimal position in the current global situation. Establish a dynamic security boundary nearby They lay their eggs here. The location for breeding dung beetle eggs has been updated as follows: in, b 1 and b 2 is a vector containing independent random variables; For the first i Only female dung beetles on the first tj The position of the next iteration.

[0045] The third type of role is the dung beetle. The hatched dung beetle will move to the current locally optimal position. Nearby smaller boundaries They forage for food within the area, and further explore and optimize their search: in, For the first i A small dung beetle in the first tj The position of the next iteration; C 1 is a random number that follows a normal distribution. C 2 is a random number between (0, 1).

[0046] The fourth role is that of the dung beetle, a thieving insect. Dung beetles, with their thieving habits, will randomly wander around local optima and steal food. This provides the algorithm with a perturbation mechanism to escape local optima: in, g Let them be random vectors that follow a normal distribution. S It is a constant.

[0047] The optimal solution obtained by this optimization algorithm is the control quantity of each boiler at the next moment.

[0048] Step 5: Send real-time load control commands to each boiler unit via PLC.

[0049] The real-time load vector is split into multiple scalars and output to each boiler PLC.

[0050] Corresponding to the aforementioned embodiments of the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization, the present invention also provides embodiments of a multi-gas boiler collaborative group control device based on steam consumption prediction and scheduling optimization.

[0051] The embodiment of the multi-gas boiler collaborative group control device based on steam consumption prediction and scheduling optimization provided by the present invention includes one or more processors for implementing the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization in the above embodiment.

[0052] The embodiments of the multi-gas boiler collaborative group control device based on steam consumption prediction and scheduling optimization of this invention can be applied to any device with data processing capabilities, such as a computer. The device embodiments can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of any data-processing device reading the corresponding computer program instructions from non-volatile memory into memory and executing them. From a hardware perspective, in addition to the processor, memory, network interface, and non-volatile memory, the data-processing device in the embodiments may also include other hardware depending on its actual functions, which will not be elaborated further.

[0053] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0054] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0055] This invention also provides a computer-readable storage medium storing a program that, when executed by a processor, implements the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization described in the above embodiments.

[0056] The computer-readable storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0057] The following specific implementation case demonstrates the effectiveness of the method and apparatus of the present invention.

[0058] To avoid impacting actual production, this embodiment uses appropriate simulation modeling to moderately simplify the production process of multiple boilers generating steam in parallel through main pipes, and takes into account factors such as the influence of main pipe pressure on boiler exhaust volume, start-up delay after boiler shutdown, and boiler efficiency curve.

[0059] Modeling a real-world multi-boiler mother-pipe parallel steam generation process, such as... Figure 2 As shown in the figure, there are multiple heat pipe steam turbines that receive instructions from the controller on the left to produce steam, which is then delivered to the main pipe. The main pipe discharges steam according to the actual steam consumption data on site. The full-load output evaporation capacity of each boiler is 1000 kg / h, the pressure setpoint is a fixed value of 0.7 MPa, the control cycle is selected as 5 seconds, and the simulation duration is selected as 1200 seconds.

[0060] In this embodiment, the key hyperparameters of the load prediction model that integrates XGBoost and Kalman filtering include: number of trees n_estimators=120, maximum depth max_depth=5, learning rate=0.06, and subsample ratio subsample=0.9.

[0061] In addition, for comparison, a load forecasting model based on XGBoost was used separately for forecasting.

[0062] The prediction performance of XGBoost and XGBoost models with fused Kalman filtering is as follows: Figure 3As shown, the XGBoost model with fused Kalman filtering in this embodiment has a significantly improved prediction performance compared to the single XGBoost model, and it better matches the actual steam flow curve. Its mean absolute error (MAE) is 0.030881, root mean square error (RMSE) is 0.050393, and coefficient of determination (R²) is 0.999487.

[0063] Based on the total control quantity and boiler status, a redundancy decision and adjustment for the number of boilers is performed, where the margin coefficient n is taken as 1, and T min Take 20 seconds.

[0064] Then, using the number of boilers as a constraint and combining it with the total control quantity from the upper level, a multi-objective optimization function is constructed, and the DBO algorithm is used to solve the optimization problem. The hyperparameters of the DBO algorithm are set as follows: population size 100, maximum number of iterations 20, and other parameters are selected as: k=0.1, b=0.3, S=0.5. λ =0.1. The coefficients of each term in the objective function are selected as follows: k 1 = 5 k 2 = 10 k 3=0.5, feedforward gain k ff =100, the coefficients of the PID controller K p =2000, K i =12, K d =0.

[0065] Finally, real-time load control commands output by the lower-level controllers are sent to each boiler unit.

[0066] The impact of sparse loss and regularization of the objective function on boiler load commands is as follows: Figure 4 As shown. A comparison of the method in this embodiment of the invention with the traditional single-machine independent PID control for the number of boiler start-ups and shutdowns is provided. Figure 5 As shown, the traditional single-unit independent PID control method refers to each boiler independently adjusting its PID control based solely on its own load or header pressure deviation, without implementing upper-level steam consumption prediction, middle-level boiler redundancy decision-making, or lower-level DBO load optimization allocation. Therefore, the method in this embodiment reduces the number of boiler start-ups and shutdowns. The final control effect is as follows... Figure 6 As shown. From Figure 6 As can be seen, the method in this embodiment has a good control effect when faced with an extreme step increase in steam flow rate, and can control the maximum deviation of the main pipe pressure within 0.1151 MPa.

[0067] It will be understood by those skilled in the art that the above descriptions are merely preferred examples of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or make equivalent substitutions for some of the technical features. All modifications and equivalent substitutions made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A collaborative group control method for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization, characterized in that, The method includes the following steps: S1: Real-time collection of main pipe pressure, steam flow, and operating status data of various gas boilers in industrial heating sites; S2: Based on historical steam consumption data, a load prediction model based on the fusion of extreme gradient boosting tree XGBoost and one-dimensional Kalman filter is established. Combined with feedback control, the total control quantity is output to realize the early perception of sudden load changes at the back end. S3: Based on the current boiler operating status and the total control quantity obtained in S2, adopt the boiler number redundancy decision method based on the operating boundary, set the minimum start-up and shutdown time constraint and the gradual boiler number adjustment strategy, and output the current required number of boilers. S4: The dung beetle optimization algorithm is used to solve the multi-objective load optimization allocation problem. The optimization objectives are load tracking accuracy, operating efficiency, control smoothness and sparsity. The number of boilers output by S3 is used as a constraint to dynamically output the optimal load setpoint for each online unit. S5: Sends real-time load control commands to each boiler unit via PLC.

2. The multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization according to claim 1, characterized in that, During the training process of the load prediction model, a combination of grid search and cross-validation is used to optimize the hyperparameters.

3. The multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization according to claim 1, characterized in that, S2 includes the following sub-steps: S2.1: Use a sliding window method to extract data and predict the steam consumption at a certain time after the current time; S2.2: Treat the predicted value as a "noisy observation" of future steam consumption status, and introduce a one-dimensional Kalman filter to perform online correction and smoothing of the XGBoost prediction results; S2.3: Based on the optimal fusion feedforward quantity, the feedback adjustment quantity of the position PID controller is superimposed to finally calculate the total load requirement required by the system.

4. The method for collaborative group control of multiple gas-fired boilers based on steam consumption prediction and scheduling optimization according to claim 1, characterized in that, S3 includes the following sub-steps: S3.1: Based on the load range of a single boiler Calculate the minimum number of units required for full-load operation. N min Maximum number of units operating at minimum load N max : In the formula, and These represent rounding up and rounding down, respectively. S3.2: Adopt a gradual strategy of adding or removing furnaces and a redundancy judgment mechanism to adjust the target number of operating furnaces. N t : if continued Then let ; if continued Then let ; if continued Then let ; in, N run This represents the current number of running machines. N min The minimum number of units required for full-load operation. N max Maximum number of units operating at minimum load N max ; T min This is the minimum start / stop time threshold; n This is the margin coefficient; p i It is the first i The load value of the boiler. T threshold It is the redundancy judgment time threshold.

5. The multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization according to claim 1, characterized in that, In each iteration, the dung beetle optimization algorithm first calculates the population fitness and then assigns four roles—rolling dung beetle, female dung beetle, small dung beetle, and thieving dung beetle—according to a preset ratio, and updates their positions accordingly.

6. The method for collaborative group control of multiple gas-fired boilers based on steam consumption prediction and scheduling optimization according to claim 1, characterized in that, In step S1, the historical steam flow data of the main pipe of the industrial heating system is acquired at the second level, and the data sampling interval is no longer than 1 minute.

7. A collaborative group control device for multiple gas-fired boilers based on steam consumption prediction and scheduling optimization, characterized in that, It includes one or more processors for implementing the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements the multi-gas boiler collaborative group control method based on steam consumption prediction and scheduling optimization as described in any one of claims 1 to 5.

Images