Combustion optimization method, system, electronic device and storage medium for coal-fired boiler

Abstract

The application provides a coal-fired boiler combustion optimization method, system, electronic equipment and storage medium. The coal-fired boiler combustion optimization method comprises the following steps: obtaining historical operation data and operation condition data of a boiler; determining a boiler combustion efficiency according to the historical operation data of the boiler; performing condition division according to the boiler combustion efficiency and the operation condition data to obtain each condition interval information; constructing a time series data-driven combustion efficiency prediction model according to the each condition interval information; embedding the combustion efficiency prediction model into a deep reinforcement learning method framework to perform training and generate an optimization model; and inputting real-time operation data of the boiler into the optimization model to generate optimization decision data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal-fired power plant boiler control, and in particular to a coal-fired boiler combustion optimization method and system, an electronic device, and a storage medium. BACKGROUND

[0002] As the core energy pillar supporting the operation of modern social economy, coal-fired power generation plays an irreplaceable fundamental role in ensuring the safe and stable operation of the power system. Although the installed capacity of new energy power generation represented by wind power and photovoltaic power continues to grow rapidly, there is significant instability and intermittency in its output characteristics. This strong random power characteristic brings serious challenges to real-time balancing and precise scheduling of the power grid. It is worth noting that in the process of frequent participation in deep peak shaving, the mismatch of combustion dynamics parameters and the attenuation of heat transfer system efficiency caused by variable load conditions result in a loss of thermal efficiency of a typical 350MW subcritical unit of up to 12%. The waste of standard coal caused by a single unit per year is as high as 18,200 tons, equivalent to the emission of 47,600 tons of carbon dioxide.

[0003] The existing coal-fired boiler combustion optimization method has significant noise in the boiler operating parameters collected by the DCS system (Distributed Control System) and missing data in some measurement points, resulting in large output error fluctuations (±1.8%) of the data-driven combustion prediction model, which affects the stability of the optimization control. SUMMARY

[0004] The present application provides a coal-fired boiler combustion optimization method, system, electronic device and storage medium to solve the problems of related technologies. The technical solution is as follows:

[0005] In a first aspect, the present application provides a coal-fired boiler combustion optimization method, comprising:

[0006] obtaining historical operation data and operating condition data of the boiler;

[0007] determining the boiler combustion efficiency according to the historical operation data of the boiler;

[0008] dividing the operating conditions according to the boiler combustion efficiency and the operating condition data to obtain each operating condition interval information;

[0009] constructing a time series data-driven combustion efficiency prediction model according to the each operating condition interval information;

[0010] embedding the combustion efficiency prediction model into a deep reinforcement learning method framework for training to generate an optimization model;

[0011] The real-time operation data of the boiler is input into the optimization model to generate optimization decision data.

[0012] In a second aspect, the embodiments of the present application provide a combustion optimization system of a coal-fired boiler, comprising:

[0013] An acquisition module is configured to acquire historical operation data and operation condition data of the boiler.

[0014] A determination module is configured to determine a combustion efficiency of the boiler according to the historical operation data of the boiler.

[0015] An obtaining module is configured to divide conditions according to the combustion efficiency of the boiler and the operation condition data to obtain information of each condition interval.

[0016] A construction module is configured to construct a time-series data driven combustion efficiency prediction model according to the information of each condition interval.

[0017] A generation module is configured to embed the combustion efficiency prediction model into a deep reinforcement learning method framework to train an optimization model.

[0018] An input module is configured to input real-time operation data of the boiler into the optimization model to generate optimization decision data.

[0019] In a third aspect, the embodiments of the present application provide an electronic device, which comprises at least one processor and a memory connected with the at least one processor in communication; the memory stores instructions executable by the at least one processor, so that the at least one processor can execute the combustion optimization method of the coal-fired boiler.

[0020] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores computer instructions; when the computer instructions are executed on a computer, the method in any one of the embodiments of the aspects is executed.

[0021] The above technical solution has at least the following advantages or beneficial effects:

[0022] In the combustion optimization method of the embodiment, the state intervals of stable and unstable working conditions in the working condition are identified through the historical operation data and operation condition data of the boiler and the working condition interval information obtained, and a combustion efficiency prediction model is constructed through the identified working condition interval information, the deep reinforcement learning method framework is trained offline in the virtual environment of the combustion prediction model, an optimization model is generated, the optimization model is deployed in the real-time data stream processing system of the boiler, the real-time operation data is input into the optimization model, optimization decision data is generated, and the boiler is optimized through the optimization decision data, which can effectively reduce the coal consumption and improve the combustion thermal efficiency, realizes online dynamic optimization control of the coal-fired boiler, significantly reduces the coal consumption while improving the thermal efficiency of the unit, and can comprehensively analyze each variable in the operation process of the boiler, effectively solving the problem that the output error of the combustion prediction model based on data driving fluctuates greatly due to the significant noise of the boiler operation parameters and the missing of part of the measured point data, and affecting the stability of the optimization control.

[0023] The above summary is intended to illustrate only and is not intended to be limiting in any way. Still other aspects, embodiments and features of the application are readily apparent to one skilled in the art and are intended to be encompassed by this application. Descriptions of illustrative aspects, embodiments and features are intended to be illustrative, and not restrictive, of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0024] In the drawings, like numerals refer to like elements throughout the various drawings. The drawings are not necessarily to scale, the emphasis instead being placed on the principles of the application. It should be understood that the drawings are merely intended to depict some of the embodiments of the application. The application should not be considered limited to the precise drawing illustrations.

[0025] Figure 1 A flowchart of a combustion optimization method for a coal-fired boiler according to an embodiment of the application.

[0026] Figure 2 A block diagram of an electronic device according to an embodiment of the application. DETAILED DESCRIPTION

[0027] In the following, only some example embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the application. Therefore, the drawings and the description are considered to be essentially exemplary rather than limiting.

[0028] In the related technology, coal-fired power generation plays an irreplaceable basic role in ensuring the safe and stable operation of the power system. Although the installed capacity of new energy power generation represented by wind power and photovoltaic continues to grow rapidly, its output characteristics have significant instability and intermittency. This strong random power characteristic brings serious challenges to real-time balancing and accurate scheduling of the power grid. According to the 2023 annual power industry statistics report, coal-fired power generation still accounts for 65.2% of the absolute proportion of total power generation. Not only does it continue to bear the key role of basic load power in the power supply structure, but also provides more than 70% of the peak shaving capacity for the power grid due to its rapid response capability. It is worth noting that in the process of frequent participation in deep peak shaving, the mismatch of combustion kinetics parameters and the attenuation of heat transfer system efficiency caused by variable load conditions result in a loss of thermal efficiency of typical 350MW subcritical units of up to 12%-15%. The waste of standard coal caused by a single unit per year is as high as 18.25 million tons, equivalent to the emission of 47.66 million tons of carbon dioxide.

[0029] When the unit heat value of coal decreases, more coal needs to be consumed to maintain the same boiler load, and the primary air volume required for conveying the pulverized coal increases, resulting in increased power consumption of the coal mill and increased resistance of the pulverizing system. This in turn requires the fan to provide a higher pressure head, increasing the power consumption of the fan and ultimately increasing the plant power consumption rate. At the same time, the decrease in coal heat value will lower the theoretical combustion temperature and the level of furnace temperature, causing a delay in the ignition of the pulverized coal gas flow and a decrease in combustion stability, affecting the burnout rate; in addition, it will also lead to an increase in the exhaust gas temperature of the boiler, increasing the heat loss of the exhaust gas.

[0030] The boiler, as the initial link in the energy conversion chain of a coal-fired power plant, its thermal efficiency directly determines the overall system energy efficiency. According to statistics, the typical coal-fired boiler loses 6-12 percentage points of comprehensive thermal efficiency due to factors such as exhaust gas heat loss (about 5%-8%), solid unburned carbon loss (about 1%-3%), and heat loss (about 0.5%-1%). Through combustion optimization control technology (such as dynamic adjustment of air-coal ratio, control of oxygen content in flue gas of economizer, etc.), coal consumption can be reduced by 1%-5% (corresponding to about 30-40 thousand tons of coal per year for a 350MW unit), and the thermal efficiency of the boiler can be improved by 2.5-4.2 percentage points.

[0031] The mechanism model based on the law of thermodynamics needs to solve the mass / energy conservation equation, combustion reaction kinetics equation and heat transfer differential equation simultaneously, involving multi-physical field coupling solution, resulting in a long time of 23 minutes for a single efficiency evaluation, which is difficult to meet the real-time optimization demand of minutes. More seriously, the mechanism model has inherent defects in modeling dynamic processes: the deviation of dynamic heat transfer coefficient caused by the load regulation rate in the actual operation of coal-fired units can reach 12.18%, resulting in an efficiency prediction error of more than 23 percentage points.

[0032] Secondly, at the data-driven optimization level, the existing combustion optimization method faces a double data dilemma:

[0033] 1. Data quality defects: Due to the significant noise in the boiler operation parameters collected by the DCS system and the missing data of some measurement points, the output error of the data-driven combustion prediction model fluctuates greatly (±1.8%), which affects the stability of the optimization control;

[0034] 2. Insufficient working condition coverage: In the historical operation data, the proportion of deep peak shaving working conditions with load rate less than 40% is only 5.2%, and the lack of samples leads to the instability of the control command of the optimization strategy in extreme working conditions (such as the fluctuation amplitude of the coal supply amount command exceeding ±15%).

[0035] Therefore, the technical personnel of the present application are committed to developing a method system suitable for combustion optimization of different capacity coal-fired boilers, which uses the data of coal-fired power plant boiler operation to realize an intelligent combustion optimization system from predicting combustion efficiency to optimizing the combustion process.

[0036] Figure 1 A flowchart of a combustion optimization method for a coal-fired boiler according to an embodiment of the present application is shown. As shown in Figure 1 A combustion optimization method for a coal-fired boiler, the combustion optimization method for a coal-fired boiler comprises:

[0037] S110: obtaining historical operation data and operation condition data of a boiler;

[0038] S120: determining the combustion efficiency of the boiler according to the historical operation data of the boiler;

[0039] S130: dividing the working conditions according to the combustion efficiency of the boiler and the operation condition data, and obtaining each working condition interval information;

[0040] S140: constructing a time series data-driven combustion efficiency prediction model according to the each working condition interval information;

[0041] S150: embedding the combustion efficiency prediction model into a deep reinforcement learning method framework for training to generate an optimization model;

[0042] S160: inputting real-time operation data of the boiler into the optimization model to generate optimization decision data.

[0043] In the combustion optimization method of the embodiment, the state intervals of stable and unstable conditions in the condition are identified through the historical operation data and operation condition data of the boiler, and a combustion efficiency prediction model is constructed through the identified condition interval information. The deep reinforcement learning method framework is trained offline in the virtual environment of the combustion prediction model to generate an optimization model. The optimization model is deployed in the real-time data stream processing system of the boiler. The real-time operation data is input into the optimization model to generate optimization decision data. The boiler is optimized through the optimization decision data, which can effectively reduce the coal consumption and improve the combustion heat efficiency, realize online dynamic optimization control of the coal-fired boiler, significantly reduce the coal consumption while improving the unit heat efficiency, and comprehensively analyze each variable in the boiler operation process, effectively solve the problem that the boiler operation parameters have significant noise and some missing data points, resulting in large output error fluctuations of the data-driven combustion prediction model and affecting the stability of the optimization control.

[0044] In step S110, historical operation data and operation condition data of the boiler are obtained. In the embodiment of the application, temperature, pressure, gas composition and flow sensors are installed in the key combustion areas (such as burners, flues, air pipes, water pipes, etc.) of the boiler. The corresponding temperature, pressure, gas composition and flow data collected by the sensors are input into the distributed control system DCS for monitoring combustion state parameters (such as furnace temperature distribution, flue gas oxygen content, air-coal ratio, etc.), so as to obtain the historical operation data and operation condition data of the boiler.

[0045] According to actual needs and coal bunker fuel consumption, the data acquisition frequency can be set to 10 seconds on the distributed control system DCS to ensure the timeliness and accuracy of the data.

[0046] The historical operation data of the boiler monitored by the distributed control system DCS is used to calculate the boiler heat efficiency. The data not directly existing in the real data is obtained by recording and differentiating, decomposing, correlation analyzing and conventional arithmetic operations according to the heat efficiency and other characteristics.

[0047] The operation condition data includes combustion heat value, unit load and cooling medium temperature and other related parameter data, which can be directly obtained on the distributed control system DCS or calculated through the measured data on the distributed control system DCS.

[0048] In step S120, the combustion efficiency of the boiler is determined according to the historical operation data of the boiler.

[0049] In the embodiments of the present application, the historical operation data of the boiler monitored by the distributed control system DCS is used to calculate various heat losses (flue gas, incomplete combustion of gas, incomplete combustion of solid, heat dissipation) by using the counterbalance method, and then the combustion efficiency is calculated.

[0050] During the operation of the boiler, the higher the combustion efficiency, the higher the fuel utilization rate, and the smaller the energy waste.

[0051] The DCS (Distributed Control System) will collect, store the boiler operation parameters in real time, such as fuel consumption, flue gas composition, boiler steam production, temperature at each place, etc. Using these historical data, the combustion efficiency can be calculated by the counterbalance method (back calculation method).

[0052] The counterbalance method is to calculate the various heat losses of the boiler, and then subtract the total proportion of these losses from 100% to get the combustion efficiency, so it is called "counterbalance" (not directly calculating the effective utilization, but inversely calculating from the loss).

[0053] At the same time, record and save the historical thermal efficiency value of each period and its corresponding main parameters, such as the calorific value of coal and the unit load, etc.

[0054] In step S130, the operating condition is divided in combination with the mill start-stop state, unit load and boiler combustion efficiency, etc. to accurately obtain the information of each operating condition interval;

[0055] In the present embodiment, the operating condition data covers key parameter data such as mill start-stop state, unit load, combustion calorific value, cooling medium temperature, etc. Based on the mill start-stop combination mode, unit load level, and in cooperation with factors such as coal calorific value and combustion efficiency, the operating condition is divided. Through this multi-parameter cooperative division method, the operating condition interval can be divided into stable operating condition interval and non-stable operating condition interval. The information of each operating condition interval can be represented and embodied by various historical data generated during the operation of the boiler in the corresponding operating condition interval.

[0056] Change the past extensive mode of relying on "load" alone to divide the boiler operating state, and realize the fine, dynamic and automatic identification of the boiler operating condition by integrating the coal characteristics and key operating parameters, so as to provide decision basis for the accurate control and optimized operation of the unit.

[0057] The offline modeling stage uses historical operation data to complete the construction of the model and the definition of the operating condition. In the online identification stage, the real-time collected data is used to automatically match the operating state of the current unit to the defined operating condition.

[0058] Establish a quantitative mapping model of coal industrial analysis parameters and combustion efficiency; create a mathematical model that can accurately predict the combustion efficiency of a boiler based on input coal quality parameters (calorific value, volatile matter, ash melting characteristics, etc.). Combustion efficiency is a core indicator, and it is one of the most critical performance indicators for evaluating the operation of a boiler. However, it is usually difficult to directly, real-time, and accurately measure online. Coal quality is a key disturbance factor, and the fuel coal in coal-fired power plants has varying characteristics (calorific value, flammability, and ease of coking, etc.), which is the most important external disturbance affecting combustion efficiency. By collecting a large amount of historical data of coal industrial analysis parameters, each data contains a complete set of coal quality test reports (calorific value, volatile matter, ash content, moisture content, ash melting point, etc.) and the corresponding actual boiler combustion efficiency (usually calculated by heat balance).

[0059] Various machine learning or statistical methods can be used to establish this mapping relationship, such as: Multiple Linear Regression, Support Vector Machines (SVM), Neural Networks, or Gradient Boosting Trees, etc. Use the above-mentioned large amount of historical data of coal industrial analysis parameters to verify the accuracy of the model using independent test data sets. The final model can be formalized as: Combustion Efficiency = f(Calorific Value, Volatile Matter, Ash Melting Characteristics,...); generate a trained combustion efficiency prediction model.

[0060] Construct a working condition feature vector; real-time coal calorific value is the core of energy input, directly determining the amount of fuel and air required to produce the same amount of heat, and is the fundamental factor affecting the combustion process. Unit load is the demand for energy output, determining the overall operation rhythm and parameter setting range of the boiler. Cooling medium temperature (such as circulating water inlet temperature) is an important boundary condition, which affects the vacuum degree of the steam turbine, and thus affects the efficiency of the entire thermal cycle. Under the same power generation load, different cooling water temperatures require different boiler evaporation capacities.

[0061] Vector construction: at any time t, the feature vector V(t) of this working condition can be represented as:

[0062] V(t) = [Real-time coal calorific value(t), Unit load(t), Cooling medium temperature(t)];

[0063] Real-time coal heat value is from online coal quality analyzer or estimated by weighted average of incoming coal. Load and temperature are directly obtained from DCS (Distributed Control System), which can generate a multi-dimensional dataset that comprehensively and quantitatively describes the real-time running state of the boiler. The real-time coal heat value is the primary basis for division, and the load and cooling medium temperature are used to construct the working condition feature vector.

[0064] The improved density clustering algorithm is used to realize fine segmentation of working conditions; and the unsupervised learning algorithm is used to automatically and objectively find the inherent clustering mode of the data in the above-constructed massive historical feature vector data, and identify these clustered "data cloud clusters" as a stable and repeatedly appearing typical working condition interval.

[0065] The traditional manual division of working conditions (such as high, medium and low load) is subjective and has a very coarse granularity. The clustering algorithm can be divided based on the distribution characteristics of the data itself, which is more scientific and fine. The improved density clustering algorithm can find some specific working condition combinations that are ignored by manual work but actually appear frequently (for example, "medium load + poor coal + high temperature in summer" may be an independent working condition that needs special attention).

[0066] The improved density clustering algorithm can be used as the clustering algorithm. The improved density clustering algorithm refers to DBSCAN (Density-Based Spatial Clustering of Applications with Noise) and its variants.

[0067] In this embodiment, the improved density clustering algorithm can be used without pre-setting the number of categories. Unlike algorithms such as K-Means, DBSCAN does not need to specify the number of working conditions to be divided in advance. The improved density clustering algorithm can automatically find the number of working conditions according to the data density. The improved density clustering algorithm can identify clusters of any shape. Real working conditions may not be circular or spherical in the feature space, and DBSCAN can well handle irregularly shaped working condition distributions. The improved density clustering algorithm can handle noise / abnormal points. Those transition states or abnormal data points that do not belong to any stable working condition will be identified as noise, which is very consistent with the actual industry. The improved density clustering algorithm optimizes the problem of DBSCAN being sensitive to parameters, such as adaptive parameter selection algorithm or HDBSCAN that can handle multi-density datasets.

[0068] The obtained large amount of historical feature vectors V(t) are input into the improved density clustering algorithm. The improved density clustering algorithm outputs the category (cluster) label to which each data point belongs.

[0069] Each identified cluster represents a data-driven defined "refined operating condition". For example: Condition A: [high calorific value coal, high load, winter low temperature]; Condition B: [high calorific value coal, high load, summer high temperature]; Condition C: [low calorific value coal, medium load,...]; and finally generate each operating condition interval information.

[0070] In step S140, based on the operating condition interval information, a time series modeling technology such as a long short-term memory network (LSTM) is used to construct a time series data-driven combustion efficiency prediction model.

[0071] In the embodiments of the present application, first, for each divided operating condition interval, the historical time series data of the boiler operation in the interval is extracted, covering the time series of key process parameters such as fuel supply amount, air distribution, furnace temperature, flue gas composition. Subsequently, these time series data are input into the LSTM network, and the model is trained to learn the dynamic change law of each operating parameter with time and the correlation between the operating parameters and the combustion efficiency under different operating conditions. The LSTM network can effectively mine the time series characteristics of parameter changes in the boiler combustion process due to its ability to capture long-time sequence dependencies, thereby achieving accurate prediction of the combustion efficiency and providing data support and prediction basis for subsequent combustion optimization control.

[0072] The boiler operation history data is divided into several intervals (such as high load, low load, different coal quality combinations, etc.) according to different operating conditions. Each operating condition interval has corresponding historical operation data, including: process variables (state variables): real-time collected temperature, pressure, concentration, coal fineness, primary air volume, secondary air volume, etc. (25-dimensional state variables); and decision variables (control variables): fuel distribution, damper opening, superheater water injection amount, online number of coal mills, etc. (11-dimensional decision variables). Based on the time series dependency between these process variables and decision variables, an initial model capable of predicting combustion efficiency is constructed.

[0073] The current combustion efficiency is not only affected by the current state and operation, but also affected by the control actions and state evolution of the previous few minutes (or tens of seconds). Simply training the model with static samples will ignore the process inertia and dynamic response lag, and the prediction effect will be poor.

[0074] Establishing a time series data set not only records the state and control amount at a single time point, but also records the change sequence within a time window (for example, the past 60 seconds, 120 seconds).

[0075] LSTM (Long Short-Term Memory) is a specially designed recurrent neural network (RNN) architecture that can remember important information over a long time span and ignore irrelevant noise: its long-term memory can perceive the cumulative impact of coal quality changes or changes in combustion state on the next few minutes; its short-term response to the efficiency changes of instantaneous adjustments of boiler operating parameters; LSTM selectively updates information through a gating mechanism (input gate, forget gate, output gate), which is very suitable for industrial time series data such as boilers, which are nonlinear, strongly lagging, and have many interference factors.

[0076] Specifically, within each divided operating condition interval, the historical boiler operating data in each operating condition interval information is used to construct a time series data set containing state variable and control variable sequences, the long short-term memory network (LSTM) learning process variable and decision variable time series dependence relationship is used to fit the nonlinear dynamic mapping relationship between input and combustion efficiency, and the preliminary combustion efficiency prediction value of each operating condition interval is obtained, which provides a mean value benchmark for subsequent uncertainty modeling and optimization control strategy.

[0077] The preliminary combustion efficiency prediction value does not indicate uncertainty, i.e., how much fluctuation range and risk distribution the prediction may have; in industrial practice, there are unpredictable factors such as coal quality fluctuations, measurement noise, and environmental changes, which will affect efficiency; if the subsequent single prediction value is directly used to guide the reinforcement learning optimization strategy, once the prediction error is large, it may lead to unsafe or suboptimal decisions for the strategy.

[0078] The diffusion model is essentially a probabilistic generative model, which has the advantages of: it can learn the complete distribution of data (not just the mean), and can generate any quantile or sample set to characterize the uncertainty of the model output. Compared with the Bayesian method with Gaussian assumption, the diffusion model is more flexible and robust when modeling complex multi-modal error distribution. When training the preliminary LSTM model, the difference between the preliminary combustion efficiency prediction value and the historical actual combustion efficiency value, i.e., the prediction residual. These residuals have different distribution characteristics under different operating conditions, with some intervals having small fluctuations and others having large fluctuations.

[0079] The historical residual data is classified according to the working condition interval and sent to the diffusion model for training. The diffusion model gradually learns the change pattern and distribution form of the residual, such as the upper and lower boundaries of the residual and the conditions that are prone to produce large deviations. When the real-time boiler state and control decision are input, the preliminary combustion efficiency prediction value is obtained by using the LSTM. The preliminary combustion efficiency prediction value is input into the diffusion model corresponding to the working condition interval, and the diffusion model is used to sample the possible residual multiple times to generate a batch of potential true values with slight differences. By counting the batch of potential true values, the mean update prediction center, maximum and minimum value, and specific quantile can be obtained. For example, the middle 90% of the data can be taken as the confidence interval of the prediction, and different quantile points can represent the range of different confidence levels. Each working condition interval has its own diffusion model, so the corresponding residual distribution characteristics can be selected according to the working condition during prediction to obtain a more accurate confidence interval. Finally, a working condition-based combustion efficiency prediction model is constructed, and each combustion efficiency prediction model outputs not only a number but also a reliable range with upper and lower boundaries.

[0080] Based on the preliminary prediction value of the LSTM, a diffusion model is connected to each working condition interval to generate a large number of efficiency prediction samples, extract quantiles, mean and variance, and then establish a probability prediction model that fully reflects the dynamic characteristics of the boiler, providing a solid mathematical foundation for intelligent optimization and risk control. By introducing a diffusion generation model based on the preliminary efficiency prediction model in each working condition interval to model and sample the prediction residual, the prediction value confidence interval containing the mean, variance and each quantile is output, thereby expanding the original deterministic prediction into a working condition-based probability prediction model that can reflect the prediction uncertainty, providing a more robust reward evaluation basis for subsequent reinforcement learning optimization control.

[0081] To maintain the prediction accuracy, the combustion efficiency prediction model is updated based on the newly added historical boiler operation data every 6 hours, so that the combustion efficiency prediction model can be optimized and updated, improving the accuracy of the combustion efficiency prediction model.

[0082] In step S150, the combustion efficiency prediction model is embedded into the deep reinforcement learning method framework, and a virtual simulation environment is constructed to carry out offline training and generate an optimization model.

[0083] In the embodiment, firstly, a state transition model and a composite reward function are acquired. The state transition model is used to depict the law of evolution of a subsequent state under the action of a control action (such as coal supply adjustment, damper opening adjustment, etc.) on the basis of a current state (such as fuel quantity, air quantity and the like) in the process of boiler operation; the composite reward function comprehensively considers multiple targets such as combustion efficiency improvement and operation stability, and provides a feedback basis for reinforcement learning training. Subsequently, in a virtual environment driven by a combustion efficiency prediction model, a deep reinforcement learning method framework algorithm is trained offline according to the state transition model and the composite reward function. The offline training process does not need to rely on online operation of a real boiler, a large number of boiler operation scenarios are simulated in the virtual environment, the reinforcement learning intelligent agent continuously explores and learns an optimal combustion control strategy, and finally an optimization model capable of guiding actual combustion optimization is generated.

[0084] A combustion efficiency is modeled by historical boiler operation data to obtain a working condition probability prediction model (LSTM model + diffusion model). Based on each working condition probability prediction model, a virtual environment is constructed as a simulation environment, which can predict a future state according to a given control decision and give a combustion efficiency and a confidence interval. A deep reinforcement learning (DRL) algorithm (such as DDPG) is used to train an optimal control strategy offline in the virtual environment, and finally an online optimization model is obtained to guide the real boiler operation. The composite reward function is a numerical score used to evaluate the good or bad of an operation in reinforcement learning. There is not only one reward index, but multiple target indexes are integrated together: the core target is to improve the combustion efficiency; the constraint / auxiliary targets are to control pollutant emission (NOx,

[0085] According to the historical operation data of the boiler, 25-dimensional process variables are extracted as state observation values and 11-dimensional decision variables are defined as an action space; according to the state observation values, the action space and a long short-term memory network, a state transition model is determined;

[0086] An action change penalty term, a first weight coefficient and a second weight coefficient are acquired; the preliminary efficiency prediction value, the combustion efficiency prediction model, the action change penalty term, the first weight coefficient and the second weight coefficient generate a composite reward function.

[0087] The boiler combustion process is modeled as a Markov decision process (MDP), specifically including extracting 25-dimensional process variables as state observation values in historical operation data of the boiler operation , defining 11-dimensional decision variables as an action space in the historical operation data , using an LSTM neural network to construct a state transition model: . ​

[0088] MDP (Markov Decision Process) is the standard modeling framework for reinforcement learning, which describes the relationship between "state-decision-state transition-reward". Abstracting the boiler combustion process into MDP means that we regard the boiler operation as a decision-making process that evolves at all times, where the current state determines the next action, and the action affects the subsequent state and reward. The state definition (State, S) is first analyzed for correlation and features are selected from the historical operation data of the boiler. Finally, 25-dimensional process variables (such as furnace temperature distribution, coal supply, primary air / secondary air flow, steam pressure, oxygen content, etc.) are selected as the observed state of the boiler at a certain time. The state represents both the "snapshot" of the current operation of the boiler and the basis for decision-making. The action definition (Action, A) defines 11-dimensional decision variables representing adjustable control means such as coal mill on / off, secondary air damper opening, air distribution ratio, coal conveying speed, etc.

[0089] The state transition model (LSTM implementation) uses LSTM (Long Short-Term Memory Network) to fit the state transition law of the boiler, i.e., to predict the next state given the current state and the action taken.

[0090] LSTM is suitable for handling the dependence of process variables on time, and can capture the dynamic characteristics and time series correlation of boiler combustion. The resulting model is the state transition model, which is one of the core components of the subsequent virtual simulation environment.

[0091] Compound reward function: where is the predicted value of the combustion thermal efficiency at time t, , is the predicted model of the combustion efficiency of each operating condition, is the action change penalty term, , is the weight coefficient .

[0092] Core indicators: predicted value of combustion thermal efficiency at time t, the main positive goal of the reward function is to improve the combustion thermal efficiency of the boiler. Call the combustion efficiency prediction model (based on LSTM+diffusion model probability prediction) for each operating condition to reduce uncertainty risk. Select the corresponding prediction model according to the current operating condition interval to give the efficiency prediction value under this state. The penalty term is the action change penalty; in actual operation, large adjustments of control variables will cause fluctuations in the boiler and even safety risks, so the reward function must include a penalty term for the change in control amplitude. The penalty term is used to suppress frequent large adjustments or excessive operations and encourage smooth optimization.

[0093] The weight coefficient can be calibrated by historical data or run tuning to balance efficiency, stability, environmental protection and other multiple objectives. The generated composite reward function can provide optimization direction and constraint for reinforcement learning.

[0094] The virtual environment is constructed by taking into account efficiency, safety and environmental protection through the combustion efficiency prediction model for calculating the efficiency change after the control action, the state transition model for calculating the boiler state evolution after the control action, and the composite reward function for judging the operation. The state transition model and the combustion efficiency prediction model are combined to form an interactive simulation environment. The reward function is embedded in the environment feedback system. The deep reinforcement learning method framework is trained offline in the virtual environment. The optimal control strategy is learned through the continuous "trial → state change → reward feedback" process. After the training is completed, an optimization model is output. The optimization model can be deployed in the real boiler operation to give optimal control suggestions in real time according to the state.

[0095] The virtual environment of the boiler operation is constructed based on the combustion efficiency prediction model and the state transition model. The deep reinforcement learning (such as DDPG) method is trained offline in the environment combined with the composite reward function considering efficiency, emission and stability. Finally, an optimization model capable of outputting optimal control instructions is obtained to realize energy-saving, stable efficiency and controllable risk of the boiler operation.

[0096] In step S160, the real-time operation data of the boiler is input into the optimization model to generate optimization decision data.

[0097] By deploying the trained optimization model to the real-time control system and inputting the operation data of the boiler into the optimization model, optimal decisions (optimization decision data) are obtained to realize online dynamic optimization control of the coal-fired boiler, which significantly reduces the coal consumption while improving the thermal efficiency of the unit.

[0098] The real-time operation data of the boiler can be obtained from the DCS system of the boiler in real time, which can be process variables (i.e. state variables) of the boiler, such as temperature of each point in the furnace, coal supply amount, primary air / secondary air volume, main steam pressure, temperature, emission concentration and the like. The dimensions of these variables can be consistent with the state space defined when training the MDP (for example, 25 dimensions). The current collected real-time state data is packaged and input into the optimization model. The optimization model will calculate the action value of obtaining the maximum long-term comprehensive income under the current state based on the previously trained strategy, i.e. optimization decision data.

[0099] The output is a control suggestion defined by the matching action space (for example, values of 11 continuous control variables): control variables such as damper opening, coal feeding belt speed, burner switching signal, air preheater adjustment, etc. The control variables can be directly issued as boiler control instructions to the actuator, or can be transmitted to the human operator for review and confirmation.

[0100] The optimization model can automatically adjust the control strategy according to the operating state. The optimization model can adjust the decision according to different coal quality, load change, and environmental conditions, rather than a fixed strategy. The control strategy takes into account the stable combustion (safety, low fluctuation), improved thermal efficiency (energy saving), and reduced emissions (environmental protection), and is more secure and reliable, with higher combustion efficiency, better energy saving and environmental protection effect. At the same time, it reduces the dependence on manual operation, and the optimization decision data is continuously output with low delay, which can update the control measures at the level of seconds or minutes.

[0101] The optimization model trained by historical data can quickly output the corresponding optimal control variables by taking the real-time collected boiler operating state as input after deployment, realizing real-time, dynamic, and closed-loop combustion process optimization control.

[0102] The combustion optimization method of the embodiment can effectively solve the problem that the boiler operating parameters collected by the DCS system have significant noise and some missing data, resulting in large output error fluctuations (±1.8%) of the data-driven combustion prediction model, affecting the stability of the optimization control. It can also solve the technical problem that the proportion of deep peak shaving conditions with a load rate of less than 40% in the historical operating data is only 5.2%, and the lack of samples leads to control instruction instability (such as coal feeding amount instruction fluctuation amplitude exceeding ±15%) of the optimization strategy in extreme conditions.

[0103] In an embodiment of the present application, the determination of the boiler combustion efficiency according to the historical operating data of the boiler comprises:

[0104] The historical operating data of the boiler is calculated by using the counterbalance method to determine the heat loss of each operating item;

[0105] The boiler combustion efficiency is determined according to the heat loss of each operating item.

[0106] In the embodiment of the present application, the historical operating data of the boiler monitored by the distributed control system DCS is used to calculate the heat loss (flue gas, incomplete combustion of gas, incomplete combustion of solid, heat dissipation) by using the counterbalance method, and then the combustion efficiency is calculated.

[0107] In the process of boiler operation, the higher the combustion efficiency, the higher the fuel utilization rate, and the smaller the energy waste.

[0108] The distributed control system will collect and store the boiler operating parameters in real time, such as fuel consumption, flue gas composition, boiler steam output, and temperature at various locations. Using these historical data, the combustion efficiency can be calculated by the counterbalance method (back calculation method).

[0109] The counterbalance method calculates the combustion efficiency by calculating the various heat losses of the boiler and then subtracting the total proportion of these losses from 100%. Therefore, it is called counterbalance (not directly calculating the effective utilization, but inversely calculating from the losses). Total heat input (fuel heat value) = useful heat (heat that converts water into steam) + various heat losses. Combustion efficiency = 1 - sum of various heat loss rates.

[0110] Common heat loss items include: flue gas heat loss The temperature of the flue gas discharged is higher than the ambient temperature, carrying away a large amount of heat; chemical incomplete combustion heat loss The flammable gas (such as CO) in the flue gas is not completely combusted, carrying away chemical energy; mechanical incomplete combustion heat loss The residual combustible carbon in fly ash and slag is not completely combusted; heat loss through the furnace wall Heat loss through conduction / radiation to the environment through furnace walls, pipes, etc.; other special losses such as unutilized air preheating loss due to air leakage, heat loss during soot blowing, etc.

[0111] According to historical data and boiler thermodynamic working condition formulas, the proportion of each loss to fuel input heat is calculated. For example: flue gas heat loss rate = (flue gas volume x specific heat capacity x (flue gas temperature - ambient temperature)) / fuel input heat; chemical incomplete combustion heat loss rate = (CO volume x combustion heat value) / fuel input heat; mechanical incomplete combustion heat loss rate = (unburned carbon in fly ash x carbon combustion heat value + unburned carbon in slag x carbon combustion heat value) / fuel input heat; heat loss rate = (boiler heat dissipation area x (furnace wall average temperature - ambient temperature) x heat dissipation coefficient) / fuel input heat; other heat loss rate (such as air leakage heat loss, soot blowing heat loss, etc.) = (cold air heat carried by air leakage - heat absorbed by corresponding air heating + soot blowing medium consumption heat) / fuel input heat. Counterbalance method calculation: combustion efficiency That is, efficiency = 100% - sum of all heat loss rates.

[0112] At the same time, record and save the historical thermal efficiency values of each period and their corresponding main parameters, such as coal heat value and unit load, etc.

[0113] In an embodiment of the present application, the operating condition data includes a combustion heat value, a unit load, and a cooling medium temperature, and the operating condition division according to the boiler combustion efficiency and the operating condition data includes:

[0114] A quantitative mapping model of the coal industry analysis parameters and the boiler combustion efficiency is established.

[0115] A historical operating condition feature vector set is generated according to the quantitative mapping model, the combustion heat value, the unit load, and the cooling medium temperature.

[0116] The operating condition interval information is obtained according to the historical operating condition feature vector set and an improved density clustering algorithm.

[0117] In the embodiment, the operating condition data includes related parameter data such as a combustion heat value, a unit load, and a cooling medium temperature, and the operating condition division is performed according to the unit load, the cooling medium temperature, the combustion heat value, and the combustion efficiency. The various operating condition intervals can be divided into stable operating condition intervals and unstable operating condition intervals. The operating condition interval information can be obtained through various historical operating data and operating public data of the boiler in various operating condition intervals.

[0118] The coarse mode of relying on only the “load” to divide the boiler operating state in the past is changed, and the fine, dynamic, and automatic identification of the boiler operating condition is realized by fusing the coal characteristics and the key operating parameters, thereby providing a decision basis for the precise control and optimized operation of the unit.

[0119] The offline modeling stage utilizes historical operating data to complete the construction of the model and the definition of the operating condition.

[0120] The online identification stage utilizes the real-time collected data to automatically match the operating state of the current unit to the defined operating condition.

[0121] A quantitative mapping model of the coal industry analysis parameters and the combustion efficiency is established. A mathematical model is created, which can accurately predict the combustion efficiency of the boiler according to the input coal quality parameters (heat value, volatile matter, ash melting characteristics, etc.). The combustion efficiency is a core index, and the combustion efficiency is one of the most critical performance indicators for evaluating the operation of the boiler. However, it is usually difficult to directly, real-time, and accurately measure online. Coal quality is a key disturbance factor. The fuel coal of the coal-fired power plant has various characteristics (heat value, flammability, and easy coking, etc.), which is the most important external disturbance affecting the combustion efficiency. By collecting a large amount of historical data of the coal industry analysis parameters, each piece of data contains a complete set of coal quality test reports (heat value, volatile matter, ash content, moisture content, ash melting point, etc.) and the corresponding actual boiler combustion efficiency (usually calculated by heat balance).

[0122] A variety of machine learning or statistical methods can be used to establish this mapping relationship, such as: multiple linear regression (Multiple Linear Regression), support vector machine (Support Vector Machines, SVM), neural network (Neural Networks) or gradient boosting tree (Gradient Boosting Trees) and the like. A large number of historical data of the above-mentioned collected coal industry analysis parameters are used, and an independent test data set is used to verify the accuracy of the model. The final obtained model can be formalized as: combustion efficiency = f (calorific value, volatile matter, ash melting characteristics,...); generate a trained combustion efficiency prediction model.

[0123] The working condition feature vector is constructed, and the real-time coal calorific value is the core of energy input, which directly determines the required amount of fuel and air supply for generating the same amount of heat, and is the fundamental factor affecting the combustion process. The unit load is the demand of energy output, which determines the overall operation rhythm and parameter setting range of the boiler. The cooling medium temperature (such as the circulating water inlet temperature) is an important boundary condition, which affects the vacuum degree of the steam turbine, and in turn affects the efficiency of the entire thermal cycle. Under the same power generation load, different cooling water temperatures require different boiler evaporation capacities.

[0124] The vector construction is at any time t, the feature vector V(t) of the working condition can be represented as:

[0125] V(t) = [real-time coal calorific value (t), unit load (t), cooling medium temperature (t)].

[0126] The real-time coal calorific value comes from the online coal quality analyzer, or is estimated by the weighted average of the coal entering the furnace. The load and temperature are directly obtained from the DCS (Distributed Control System), which can generate a multi-dimensional data set that comprehensively and quantitatively describes the real-time operating state of the boiler. With the real-time coal calorific value as the primary division basis, the working condition feature vector is constructed in cooperation with the unit load and the cooling medium temperature.

[0127] An improved density clustering algorithm is used to realize fine segmentation of the working condition, and an unsupervised learning algorithm is used to automatically and objectively find the inherent clustering pattern of the data in the constructed massive historical feature vector data, and identify these clustered "data cloud clusters" as a stable and repeatedly appearing typical working condition interval.

[0128] In this embodiment, the improved density clustering algorithm can be used without pre-setting the number of categories. Unlike algorithms such as K-Means, DBSCAN does not need to specify in advance how many working conditions to divide into, and the improved density clustering algorithm can automatically find the number of working conditions according to the data density.

[0129] The obtained large number of historical feature vectors V(t) are input into the improved density clustering algorithm. The improved density clustering algorithm outputs the class (cluster) label to which each data point belongs. Each identified cluster represents a "refined operating condition" defined by data. For example: operating condition A: [high calorific value coal, high load, winter low temperature]; operating condition B: [high calorific value coal, high load, summer high temperature]; operating condition C: [low calorific value coal, medium load,...]; and finally, the operating condition interval information is generated.

[0130] In an embodiment of the present application, the construction of the time series data-driven combustion efficiency prediction model according to the operating condition interval information comprises:

[0131] establishing a time series data set based on the operating condition interval information;

[0132] generating a preliminary efficiency prediction value according to the time series association between the process variables and the decision variables in the time series data set by a long short-term memory network processing;

[0133] obtaining historical actual combustion efficiency values;

[0134] determining a prediction residual according to the preliminary efficiency prediction value and the historical actual combustion efficiency value;

[0135] diffusion model generates prediction residual samples based on the distribution of the prediction residual;

[0136] extracting quantiles, mean and variance from the prediction residual samples to obtain a confidence interval of the prediction value;

[0137] generating a time series data-driven combustion efficiency prediction model according to the confidence interval of the prediction value.

[0138] In this embodiment, within the divided operating condition intervals, the historical boiler operation data in the operating condition interval information is used to construct a time series data set containing sequences of state variables and control variables. A long short-term memory network (LSTM) is used to learn the time series dependence relationship between process variables and decision variables, fit the nonlinear dynamic mapping relationship between input and combustion efficiency, and obtain preliminary combustion efficiency prediction values for each operating condition interval, providing a mean value benchmark for subsequent uncertainty modeling and optimization control strategies.

[0139] In training the preliminary LSTM model, the difference between the preliminary combustion efficiency prediction value and the historical actual combustion efficiency value, i.e. the prediction residual, is used. These residuals have different distribution characteristics under different operating conditions, with some intervals having small fluctuations and some intervals having large fluctuations.

[0140] The historical residual data is classified according to the working condition interval and sent to the diffusion model for training. The diffusion model learns the overall probability distribution of the residual data through the iterative process of forward noise addition and reverse noise removal. The working condition characteristics can be additionally input as a condition to obtain the residual distribution of the working condition. Under the condition of the current working condition and the preliminary prediction value, the diffusion model starts the "reverse sampling" process from pure noise, generates a new sample data set that conforms to the residual distribution learned before through multi-step denoising iteration. The set of values obtained is the prediction residual sample set (possibly hundreds or thousands), which reflects the possible error range of the model under this working condition.

[0141] The diffusion model gradually learns the change pattern and distribution form of the residual, such as the upper and lower boundaries of the residual and the conditions that are prone to produce large deviations. When the real-time boiler state and control decision are input, the LSTM is used to obtain the preliminary combustion efficiency prediction value. The preliminary combustion efficiency prediction value is input into the diffusion model corresponding to the working condition interval, and the diffusion model is used to sample the possible residual multiple times to generate a batch of potential true values with slight differences. By counting this batch of potential true values, the mean update prediction center, maximum and minimum value, and specific quantile can be obtained. For example, the middle 90% of the data can be taken as the confidence interval of the prediction, and different quantile points can represent different confidence levels.

[0142] Each working condition interval has its own diffusion model, so the corresponding residual distribution characteristics can be selected according to the working condition during prediction to obtain a more accurate confidence interval. Finally, a working condition-specific combustion efficiency prediction model is constructed, which outputs not only a number but also a reliable range with upper and lower boundaries. Based on the preliminary prediction value of the LSTM, the diffusion model is connected to each working condition interval to generate a large number of efficiency prediction samples, extract quantiles, mean and variance, and establish a probability prediction model that fully reflects the dynamic characteristics of the boiler, providing a solid mathematical foundation for intelligent optimization and risk management.

[0143] By introducing the diffusion generation model to model and sample the prediction residual based on the preliminary efficiency prediction model in each working condition interval, the prediction value confidence interval containing the mean, variance and each quantile is output, thereby expanding the original deterministic prediction into a working condition-specific probability prediction model that can reflect the prediction uncertainty, providing a more robust reward evaluation basis for subsequent reinforcement learning optimization control.

[0144] To maintain the prediction accuracy, the combustion efficiency prediction model is updated based on the newly added historical boiler operation data every 6 hours, so that the combustion efficiency prediction model can be optimized and updated, improving the accuracy of the combustion efficiency prediction model.

[0145] In an embodiment of the present application, the establishment of the time series data set based on the working condition interval information comprises:

[0146] Obtaining a boiler operating process variable in a working condition interval;

[0147] Performing correlation analysis on the boiler operating process variable in the working condition interval to determine that the variable most related to the combustion efficiency is the core decision variable;

[0148] Taking the coal supply amount, the air-coal ratio, and the flue gas oxygen content as the core decision variable feature set, a time series data set is established.

[0149] In this embodiment, the boiler operating history data is divided into several intervals (such as high load, low load, different coal quality combinations, etc.) according to different working conditions. Each working condition interval has corresponding historical operating data, including: process variables (state variables): real-time collected temperature, pressure, concentration, coal fineness, primary air volume, secondary air volume, etc. (that is, 25-dimensional state variables); and decision variables (control variables): fuel distribution, damper opening, superheater water injection amount, online number of coal mills, etc. (11-dimensional decision variables) Based on the time series dependence relationship between the process variables and the decision variables, an initial model capable of predicting the combustion efficiency is constructed. Establishing a time series data set not only records the state and control variables at a single time point, but also records the change sequence within a time window (for example, 60 seconds, 120 seconds).

[0150] In the divided working condition intervals, the historical boiler operating data in the working condition interval information is used to construct a time series data set containing state variable and control variable sequences. The long short-term memory network (LSTM) is used to learn the time series dependence relationship between the process variables and the decision variables, fit the nonlinear dynamic mapping relationship between the input and the combustion efficiency, and obtain the preliminary combustion efficiency prediction value of each working condition interval, providing a mean value benchmark for subsequent uncertainty modeling and optimization control strategy.

[0151] Calculate the correlation index of all variables and the combustion efficiency (target variable), such as the Pearson correlation coefficient, the Spearman rank correlation, or the feature importance ranking. Remove variables with a correlation coefficient below a certain threshold. The coal supply amount (directly affecting heat input), the air-coal ratio (affecting the degree of complete combustion and flue gas loss), and the flue gas oxygen content (at the economizer outlet) (reflecting the air excess or deficiency of combustion), these variables are not only highly correlated features, but also adjustable variables in future reinforcement learning optimization.

[0152] For each working condition interval, the feature set data corresponding to the time period is extracted. Arrange each sample in time sequence, maintaining the dynamic dependence relationship between variables. Including: historical lag values (such as variable values of the past 5 minutes or 10 sampling points), possible derived features (moving average, difference value, etc.), each working condition interval can generate an independent time series data set.

[0153] In an embodiment of the present application, the combustion efficiency prediction model is embedded into a deep reinforcement learning method framework for training to generate an optimization model, comprising:

[0154] obtaining a state transition model and a composite reward function;

[0155] In a combustion efficiency prediction model driven virtual environment, the deep reinforcement learning method framework algorithm is trained offline according to the state transition model and the composite reward function to generate an optimization model.

[0156] The combustion efficiency is modeled by the historical operation data of the boiler to obtain a sub-working condition probability prediction model (LSTM model + diffusion model).

[0157] Based on each sub-working condition probability prediction model, a virtual environment is constructed as a simulation environment, which can predict future states according to a given control decision and give a combustion efficiency and a confidence interval.

[0158] An optimization control strategy is trained offline in the virtual environment by a deep reinforcement learning (DRL) algorithm (such as DDPG) to finally obtain an online optimization model to guide the operation of the real boiler.

[0159] In an embodiment of the present application, the obtaining of the boiler combustion variable data and the composite reward function comprises:

[0160] According to the historical operation data of the boiler, 25-dimensional process variables are extracted as state observation values and 11-dimensional decision variables are defined as action spaces;

[0161] According to the state observation values, the action spaces and a long short-term memory network, a state transition model is determined;

[0162] An action change penalty term, a first weight coefficient and a second weight coefficient are obtained;

[0163] The preliminary efficiency prediction value, the combustion efficiency prediction model, the action change penalty term, the first weight coefficient and the second weight coefficient generate a composite reward function.

[0164] The composite reward function is a numerical score used to evaluate the goodness or badness of an operation in reinforcement learning. There is not only one reward indicator, but multiple target indicators are integrated together: the core target is to improve the combustion efficiency; the constraint / auxiliary targets are to control the pollutant emission (NOx, ), to keep the steam temperature stable, to reduce the boiler wear, etc.

[0165] According to historical operation data of the boiler, 25-dimensional process variables are extracted as state observations, and 11-dimensional decision variables are defined as action space; according to the state observations, the action space and a long short-term memory network, a state transition model is determined.

[0166] An action change penalty term, a first weight coefficient and a second weight coefficient are obtained; the preliminary efficiency prediction value, the combustion efficiency prediction model, the action change penalty term, the first weight coefficient and the second weight coefficient generate a composite reward function.

[0167] The boiler combustion process is modeled as a Markov decision process (MDP), specifically including extracting 25-dimensional process variables as state observations from historical operation data of the boiler , defining 11-dimensional decision variables as action space from historical operation data , using an LSTM neural network to construct a state transition model: .

[0168] MDP (Markov Decision Process) is a standard modeling framework for reinforcement learning. State definition (State, S) is from the historical operation data of the boiler, and correlation analysis and feature selection are performed first. Finally, 25-dimensional process variables (such as furnace temperature distribution, coal supply, primary air / secondary air flow, steam pressure, oxygen content, etc.) are selected as the observed state of the boiler at a certain time. The state represents both the "snapshot" of the current operation of the boiler and the basis for decision-making. Action definition (Action, A) defines 11-dimensional decision variables, representing adjustable control means such as coal mill switch, secondary air damper opening, air distribution ratio, coal conveying speed, etc.

[0169] The state transition model (LSTM implementation) uses LSTM (Long Short-Term Memory Network) to fit the state transition law of the boiler, i.e. to predict the next state given the current state and the action taken. LSTM is suitable for handling the dependence of process variables on time, and can capture the dynamic characteristics and time series correlation of boiler combustion. The resulting model is a state transition model, which is one of the core components of the subsequent virtual simulation environment.

[0170] Composite reward function: where is the predicted value of the combustion thermal efficiency at time t, , is the corresponding combustion efficiency prediction model for each working condition, is the action change penalty term, , is the weight coefficient .

[0171] The core index is a predicted value of the combustion heat efficiency at time t, and the main positive target of the reward function is to improve the combustion heat efficiency of the boiler. The combustion efficiency prediction model (probability prediction based on LSTM+diffusion model) of each working condition is called to reduce the uncertainty risk. The corresponding prediction model is selected according to the current working condition interval, and the efficiency prediction value in the state is given. The penalty term is the action change penalty; in actual operation, a large adjustment of the control quantity will cause the boiler to fluctuate or even pose a safety risk, so the penalty term of the control change amplitude must be added to the reward function. The penalty term is used to suppress frequent large adjustments or excessive operation and encourage smooth optimization.

[0172] The weight coefficient can be calibrated or optimized by historical data to balance multiple objectives such as efficiency, stability and environmental protection, and the generated composite reward function can provide an optimization direction and constraint for reinforcement learning.

[0173] The combustion efficiency prediction model is used to calculate the efficiency change after the control action, the state transition model is used to calculate the boiler state evolution after the control action, and the composite reward function is used to judge the operation. The virtual environment is constructed by taking into account efficiency, safety and environmental protection: the state transition model and the combustion efficiency prediction model are combined to form an interactive simulation environment; the reward function is embedded in the environment feedback system, and the deep reinforcement learning method framework is trained offline: in the virtual environment, the agent of the deep reinforcement learning method framework is repeatedly tested; through the continuous “trial→state change→reward feedback” process, the optimal control strategy is learned, and after training, a strategy network (optimization model) is output. The optimization model can be deployed in real boiler operation to give optimal control suggestions in real time according to the state.

[0174] Based on the combustion efficiency prediction model and the state transition model, a virtual environment for boiler operation is constructed, a composite reward function considering efficiency, emission and stability is combined, and a deep reinforcement learning (such as DDPG) method is used for offline training in the environment to finally obtain an optimization model capable of outputting optimal control instructions, realizing energy-saving, stable efficiency and controllable risk of boiler operation.

[0175] In a second aspect, the embodiments of the present application provide a combustion optimization system of a coal-fired boiler, comprising:

[0176] An acquisition module is configured to acquire historical operation data and operation condition data of the boiler;

[0177] A determination module is configured to determine the combustion efficiency of the boiler according to the historical operation data of the boiler;

[0178] A obtaining module is configured to divide working conditions according to the combustion efficiency of the boiler and the operation condition data, and obtain working condition interval information.

[0179] a construction module configured to construct a time-series data-driven combustion efficiency prediction model according to the working condition interval information;

[0180] a generation module configured to embed the combustion efficiency prediction model into a deep reinforcement learning method framework to train the combustion efficiency prediction model and generate an optimization model;

[0181] an input module configured to input real-time operation data of the boiler into the optimization model to generate optimization decision data.

[0182] In the combustion optimization system, the historical operation data and the operation condition data of the boiler are used to obtain the working condition interval information, the state intervals of stable working conditions and unstable working conditions in the working conditions are identified, the combustion efficiency prediction model is constructed according to the identified working condition interval information, the deep reinforcement learning method framework is trained offline in the virtual environment of the combustion prediction model to generate the optimization model, the optimization model is deployed in the real-time data stream processing system of the boiler, the real-time operation data is input into the optimization model to generate the optimization decision data, and the boiler is optimized according to the optimization decision data. The coal consumption can be effectively reduced and the combustion thermal efficiency can be improved, the online dynamic optimization control of the coal-fired boiler is realized, the coal consumption is significantly reduced while the thermal efficiency of the unit is improved, and all variables in the operation process of the boiler are analyzed, thereby effectively solving the problem that the output error of the data-driven combustion prediction model fluctuates greatly due to the significant noise of the operation parameters of the boiler and the missing data of some measurement points, and the stability of the optimization control is affected.

[0183] In an embodiment of the present application, the combustion efficiency of the boiler is determined according to the historical operation data of the boiler, including:

[0184] The historical operation data of the boiler is calculated by using the counterbalance method to determine the heat loss of each operation item;

[0185] The combustion efficiency of the boiler is determined according to the heat loss of each operation item.

[0186] In an embodiment of the present application, the operation condition data includes the combustion heat value, the unit load and the cooling medium temperature, and the working condition interval information is obtained according to the combustion efficiency of the boiler and the operation condition data, including:

[0187] A quantitative mapping model of the coal industry analysis parameters and the combustion efficiency of the boiler is established;

[0188] A historical working condition feature vector set is generated according to the quantitative mapping model, the combustion heat value, the unit load and the cooling medium temperature;

[0189] Information for each operating condition interval is obtained based on the set of historical operating condition feature vectors and the improved density clustering algorithm.

[0190] In one embodiment of this application, constructing a time-series data-driven combustion efficiency prediction model based on the information of each operating condition interval includes:

[0191] A time-series dataset is established based on the information of each operating condition interval;

[0192] Based on the time-series dataset, the temporal correlation between process variables and decision variables is processed using a long short-term memory network to generate preliminary efficiency prediction values;

[0193] Obtain historical actual combustion efficiency values;

[0194] The prediction residual is determined based on the preliminary efficiency prediction and the historical actual combustion efficiency.

[0195] The diffusion model generates predicted residual samples based on the distribution of the predicted residuals;

[0196] The quantiles, mean, and variance are extracted from the predicted residual samples to obtain the confidence interval of the predicted values.

[0197] Based on the confidence interval of the predicted values, a time-series data-driven combustion efficiency prediction model is generated.

[0198] In one embodiment of this application, establishing a time-series dataset based on the information of each operating condition interval includes:

[0199] Obtain the boiler operation process variables within the operating condition range;

[0200] A correlation analysis was conducted on the boiler operation process variables within the aforementioned operating condition range to identify the variables most relevant to combustion efficiency as the core decision variables.

[0201] A time-series dataset is established using a feature set with coal feed rate, air-to-coal ratio, and flue gas oxygen content as the core decision variables.

[0202] In one embodiment of this application, the step of embedding the combustion efficiency prediction model into a deep reinforcement learning framework for training to generate an optimized model includes:

[0203] Obtain the state transition model and the compound reward function;

[0204] In a virtual environment driven by the combustion efficiency prediction model, an optimized model is generated by offline training of the deep reinforcement learning framework algorithm based on the state transition model and the composite reward function.

[0205] In one embodiment of this application, obtaining boiler combustion variable data and the composite reward function includes:

[0206] Based on the historical operating data of the boiler, 25-dimensional process variables are extracted as state observations and 11-dimensional decision variables are defined as the action space.

[0207] The state transition model is determined based on the state observations, the action space, and the long short-term memory network.

[0208] Obtain the action change penalty, the first weight coefficient, and the second weight coefficient;

[0209] The preliminary efficiency prediction value, the combustion efficiency prediction model, the action change penalty term, the first weight coefficient, and the second weight coefficient are used to generate a composite reward function.

[0210] The functions of each module in each device in the embodiments of this application can be found in the corresponding descriptions in the above methods, and will not be repeated here.

[0211] Figure 2 A structural block diagram of an electronic device according to an embodiment of this application is shown. Figure 2 As shown, the electronic device includes a memory 410 and a processor 420. The memory 410 stores instructions that can be executed on the processor 420. When the processor 420 executes the instructions, it implements the combustion optimization method for the coal-fired boiler in the above embodiments. The number of memories 410 and processors 420 can be one or more. This electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0212] The electronic device can also include a communication interface 430 for communicating with external devices to exchange data. The various devices are interconnected by different buses, and can be mounted on a common motherboard or otherwise, as desired. The processor 420 can process instructions for execution within the electronic device, including instructions stored in the memory or on the memory to display graphical information for a GUI on an external input / output device, such as a display device coupled to the interface. In other implementations, multiple processors and / or multiple buses can be employed as desired to implement these functions, and multiple memories and types of memory can be used. Also, various devices can be made and operated in conjunction with the present application, e.g., by virtualizing the components, as a server array, a group of blade servers, or a multi-processor system. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of presentation, Figure 2 Only one bus is shown in the figure, but there can be more buses than the one shown, and the bus that is shown can represent several buses or several types of buses.

[0213] Optionally, if the memory 410, the processor 420 and the communication interface 430 are integrated on a chip, the memory 410, the processor 420 and the communication interface 430 can communicate with each other through an internal interface.

[0214] It should be understood that the processor described above can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It is worth noting that the processor can be a processor supporting an advanced RISC machine (ARM) architecture.

[0215] The embodiments of the present application provide a computer readable storage medium (such as the memory 410 described above), which stores computer instructions, and the program is executed by the processor to implement the method provided in the embodiments of the present application.

[0216] Optionally, the memory 410 can include a program storage area and a data storage area. The program storage area can store the operating system, application programs required by at least one function, and the like. The data storage area can store data created by the electronic device while it is being used. Additionally, the memory 410 can include a volatile memory, and also include a non-volatile memory such as at least one of a magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 410 can optionally include a memory that is remotely arranged with respect to the processor 420, and these remote memories can be connected to the electronic device through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0217] Any processes or methods described in the flowcharts or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions (or steps) of the process. Moreover, in some embodiments, the various functions or processes described in the flowcharts can be implemented as software code that is executed by a processor, such as the processor 420. Additionally, in some embodiments, the various functions or processes described in the flowcharts can be implemented as software code that is executed by a processor, such as the processor 420.

[0218] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be embodied in computer-readable storage media, which can be executed by a processing system, apparatus, or device, such as a computer-based system, a processor-based system, or other system that can fetch the instructions from the instruction-executing system, apparatus, or device and execute the instructions.

[0219] It should be understood that various aspects of the application can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by suitable instruction-executing systems. All or part of the embodiments of the methods described above can be implemented by program instructions that are stored in a computer-readable storage medium. The program instructions, when executed, implement one or more of the steps of the embodiments of the methods.

[0220] In addition, each of the function units in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium. The storage medium can be a read-only memory, a magnetic disk or an optical disk, etc.

[0221] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for combustion optimization of a coal-fired boiler, characterized in that, The method comprises the following steps: acquiring historical operation data and operation condition data of a boiler; determining a boiler combustion efficiency according to the historical operation data of the boiler; dividing conditions according to the boiler combustion efficiency and the operation condition data to obtain each condition interval information; constructing a time series data driven combustion efficiency prediction model according to the each condition interval information; embedding the combustion efficiency prediction model into a deep reinforcement learning method framework to train an optimization model; inputting real-time operation data of the boiler into the optimization model to generate optimization decision data; the operation condition data comprises a combustion heat value, a unit load and a cooling medium temperature, and the dividing conditions according to the boiler combustion efficiency and the operation condition data to obtain each condition interval information comprises: establishing a quantitative mapping model of coal industry analysis parameters and the boiler combustion efficiency; generating a historical condition feature vector set according to the quantitative mapping model, the combustion heat value, the unit load and the cooling medium temperature; obtaining each condition interval information according to the historical condition feature vector set and an improved density clustering algorithm.

2. The method of claim 1, wherein, the determining the boiler combustion efficiency according to the historical operation data of the boiler comprises: calculating each operation item heat loss by using a counterbalance method on the historical operation data of the boiler; determining the boiler combustion efficiency according to the heat loss of each operation item.

3. The method of claim 2, wherein, the constructing a time series data driven combustion efficiency prediction model according to the each condition interval information comprises: establishing a time series data set based on the each condition interval information; generating a preliminary efficiency prediction value by processing time series correlation of process variables and decision variables by using a long short-term memory network according to the time series data set; acquiring a historical actual combustion efficiency value; determining a prediction residual according to the preliminary efficiency prediction value and the historical actual combustion efficiency value; diffusing a model based on distribution of the prediction residual to generate a prediction residual sample; extracting quantiles, mean and variance from the prediction residual sample to obtain a confidence interval of the prediction value; generating a time series data driven combustion efficiency prediction model according to the confidence interval of the prediction value.

4. The method of claim 3, wherein, the establishing a time series data set based on the each condition interval information comprises: acquiring boiler operation process variables in a condition interval; determining a core decision variable which is most relevant to the combustion efficiency by performing correlation analysis on the boiler operation process variables in the condition interval; constructing a time series data set based on a feature set of the core decision variable which is the coal supply amount, the air-coal ratio and the oxygen content in flue gas.

5. The method of claim 4, wherein, the embedding the combustion efficiency prediction model into a deep reinforcement learning method framework to train an optimization model comprises: acquiring a state transition model and a compound reward function; training an optimization model by performing offline training on a deep reinforcement learning method framework algorithm in a virtual environment driven by the combustion efficiency prediction model according to the state transition model and the compound reward function.

6. The method of claim 5, wherein, the acquiring a state transition model and a compound reward function comprises: extracting 25-dimensional process variables as state observation values and defining 11-dimensional decision variables as action spaces according to the historical operation data of the boiler; determining a state transition model according to the state observation values, the action spaces and a long short-term memory network; an action change penalty term, a first weight coefficient, and a second weight coefficient are obtained; the preliminary efficiency prediction value, the combustion efficiency prediction model, the action change penalty term, the first weight coefficient, and the second weight coefficient generate a composite reward function.

7. A combustion optimization system for a coal-fired boiler, characterized by, comprise: an acquisition module, configured to acquire historical operation data and operation condition data of a boiler; a determination module, configured to determine a boiler combustion efficiency according to the historical operation data of the boiler; an obtaining module, configured to divide a condition according to the boiler combustion efficiency and the operation condition data to obtain each condition interval information; a construction module, configured to construct a time series data driven combustion efficiency prediction model according to the each condition interval information; a generation module, configured to embed the combustion efficiency prediction model into a deep reinforcement learning method framework to train an optimization model; an input module, configured to input real-time operation data of the boiler into the optimization model to generate optimization decision data; the operation condition data comprises a combustion heat value, a unit load, and a cooling medium temperature, and the dividing of the condition according to the boiler combustion efficiency and the operation condition data to obtain the each condition interval information comprises: establishing a quantitative mapping model of a coal industry analysis parameter and the boiler combustion efficiency; generating a historical condition feature vector set according to the quantitative mapping model, the combustion heat value, the unit load, and the cooling medium temperature; obtaining the each condition interval information according to the historical condition feature vector set and an improved density clustering algorithm.

8. An electronic device, comprising: comprise: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method in any one of claims 1-6. 9.A computer readable storage medium, the computer readable storage medium storing computer instructions, the computer instructions being executed by a processor to implement the method in any one of claims 1-6.

Images