Method and system for adaptive control of pressure at the outlet of the furnace of a circulating fluidized bed boiler

Abstract

This invention provides a method and system for adaptive control of furnace outlet pressure in circulating fluidized bed boilers, relating to the field of pressure control technology. The method includes: acquiring multiple historical operating samples by collecting historical operating data; grouping predefined fixed load interval sequences into multiple sets of sample points; solving for PID parameters using a least squares method to fit a linear model, obtaining multiple sets of load case points; associating and storing these cases to construct a load case library; performing case matching and retrieval, returning K cases with the same load value if they exist; performing arithmetic mean calculation to output real-time optimized parameters; and sending these parameters to the PID controller for furnace outlet pressure control. This invention solves the technical problem of existing boiler furnace outlet pressure control, which is typically based on preset static parameters and cannot update PID parameters in real time according to changes in different operating conditions, leading to reduced boiler efficiency and even the risk of equipment overpressure.

Description

Technical Field

[0001] This invention relates to the field of pressure control technology, specifically to an adaptive control method and system for furnace outlet pressure in a circulating fluidized bed boiler. Background Technology

[0002] Excessively high or low furnace outlet pressure in circulating fluidized bed boilers can have a significant impact on equipment lifespan, safety, and economic benefits. Therefore, maintaining stable furnace outlet pressure is of utmost importance.

[0003] The furnace outlet pressure control technology for circulating fluidized bed boilers is relatively mature. The time lag between the furnace outlet pressure and the forced draft and induced draft systems is small. The industry standard practice is to use a PID controller to adjust the induced draft fan speed frequency. When the furnace outlet pressure is high, the induced draft fan speed frequency is increased, thereby reducing the furnace outlet pressure; when the furnace outlet pressure is low, the induced draft fan speed frequency is decreased, thereby increasing the furnace outlet pressure.

[0004] However, in existing technologies, PID controllers typically operate under preset static parameters, which cannot be adjusted in a timely manner according to changes in boiler load, fan speed, and other operating conditions. Since different operating conditions may cause changes in the parameters of the PID controller, if the PID parameters are not optimized in time, the controller cannot respond quickly to changes in furnace outlet pressure when the PID parameters are not adapted to the new operating conditions. For example, when the boiler load increases rapidly, if the proportional coefficient is not appropriate, the furnace outlet pressure cannot reach the target value in a short time, which leads to a decrease in boiler efficiency and may even pose a risk of equipment overpressure. Summary of the Invention

[0005] This application provides a method and system for adaptive control of furnace outlet pressure in circulating fluidized bed boilers, aiming to solve the technical problem that existing boiler furnace outlet pressure control is usually based on preset static parameters, which cannot update PID parameters in real time according to changes in different operating conditions, resulting in reduced boiler efficiency and even the risk of equipment overpressure.

[0006] The first aspect disclosed in this application provides an adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler. The method includes: acquiring multiple historical operating samples by collecting historical operating data; grouping the multiple historical operating samples into load intervals according to a predefined fixed load interval sequence to obtain multiple sets of sample points; solving the PID parameters of the multiple sets of sample points using a least squares method to fit a linear model, resulting in multiple sets of load case points; associating and storing the fixed load interval sequence and the multiple sets of load case points to construct a load case library; inputting real-time boiler load values ​​into the load case library for case matching retrieval; if identical load values ​​exist, returning K identical load case points, where K ≥ 2 and K is a positive integer; performing an arithmetic mean calculation on the K sets of PID parameters of the K identical load case points to output real-time optimized parameters; and sending the real-time optimized parameters to a PID controller to execute furnace outlet pressure control.

[0007] The second aspect of this application discloses an adaptive control system for the furnace outlet pressure of a circulating fluidized bed boiler. This system is used in the aforementioned adaptive control method for the furnace outlet pressure of a circulating fluidized bed boiler. The system includes: a historical operation sample acquisition module for acquiring multiple historical operation samples by collecting historical operation data; a load interval grouping processing module for predefined fixed load interval sequences to group the multiple historical operation samples into multiple sets of sample points; and a PID parameter solving module for solving the PID parameters of the multiple sets of sample points using the least squares method to fit a linear model. The system includes: a load case point module; a load case library construction module, used to associate and store the fixed load interval sequence and multiple sets of load case points to complete the construction of the load case library; a case matching and retrieval module, used to input the real-time boiler load value into the load case library for case matching and retrieval, and if there are identical load values, return K identical load case points, where K≥2 and K is a positive integer; a real-time optimization parameter output module, used to perform arithmetic mean calculation on the K sets of PID parameters of the K identical load case points and output real-time optimization parameters; and a furnace outlet pressure control module, used to send the real-time optimization parameters to the PID controller to execute furnace outlet pressure control.

[0008] One or more technical solutions provided in this application have at least the following beneficial effects:

[0009] By collecting historical data, a sample set under various operating conditions is established, capturing detailed information about boiler operation under different load conditions and ensuring that PID controller parameter adjustments have a realistic basis. By grouping data into load intervals, the complex historical dataset is simplified, divided into different intervals. The data within each interval is more consistent and representative, avoiding confusion and errors caused by significant differences in adjustment methods under different load conditions, thus making subsequent PID parameter optimization and adjustment more precise. The least squares method can effectively fit historical data, thereby deriving PID parameters related to furnace outlet pressure changes. Accurate fitting of historical data using mathematical optimization techniques yields PID parameters that more precisely reflect actual control requirements, making parameter solutions more scientific and stable. By constructing a load case library, historical data and corresponding PID parameters within different load intervals are stored, forming a structured data set. The case library provides a foundation for subsequent rapid retrieval and parameter reuse, ensuring that the system can obtain the optimal PID parameters in real time, reducing system response time and improving control efficiency. By matching real-time load values ​​with data in the load case library, the system can quickly locate the historical sample that best matches the current operating condition. This process reduces the problem of untimely PID parameter adjustment due to changes in operating conditions in traditional methods, ensuring that the system can match the corresponding PID parameters in real time and accurately. By averaging the PID parameters of multiple historical case points, the random errors of a single historical sample are avoided, resulting in more stable and reliable parameters. The optimized PID parameters directly control the regulation of the furnace outlet pressure, ensuring control accuracy and stability. By adjusting the PID parameters in real time, the problems of furnace outlet pressure fluctuation, slow regulation, or over-regulation caused by lag or inaccuracy of PID parameters in traditional methods are avoided.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of a method for adaptive control of furnace outlet pressure in a circulating fluidized bed boiler, provided in an embodiment of this application.

[0012] Figure 2 This is a schematic diagram of the adaptive control system for furnace outlet pressure of a circulating fluidized bed boiler, provided in an embodiment of this application.

[0013] Figure labeling: Historical operation sample acquisition module 10, load interval grouping processing module 20, PID parameter solving module 30, load case library construction module 40, case matching retrieval module 50, real-time optimization parameter output module 60, furnace outlet pressure control module 70. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1, as Figure 1 As shown in the embodiments of this application, an adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler is provided, the method comprising:

[0016] A100: By collecting historical operational data, multiple historical operational samples are obtained.

[0017] During the actual operation of the boiler, relevant data is collected, including but not limited to boiler load, induced draft fan speed and frequency, primary and secondary fan speed and frequency, and furnace outlet pressure changes. This data can be acquired through the boiler's DCS (Distributed Control System). Multiple sample data points are recorded over a certain period, including operating data under different conditions and covering a sufficient time frame to ensure the comprehensiveness and representativeness of the data.

[0018] A200: The predefined fixed load interval sequence is used to group the load intervals of the multiple historical operation samples to obtain multiple sets of sample points.

[0019] Based on the boiler's load range, a predefined sequence of fixed load intervals is established. For example, if the boiler load ranges from 100t to 300t, this interval can be divided into multiple sub-intervals, such as 100t-150t, 150t-200t, 200t-250t, and 250t-300t. Historical operating samples are then categorized according to load intervals. For instance, if a sample has a load of 200t, it will be classified into the 200t-250t interval. Within a specific load interval, with the primary and secondary air fan speeds remaining constant, the change in furnace outlet pressure corresponding to the adjustment of the induced draft fan speed frequency is used as a sample point. Based on this, data aggregation is performed using the constant primary and secondary air fan speeds as the aggregation condition to construct sample points.

[0020] A300: The least squares method is used to fit the linear model to solve the PID parameters of the multiple sets of sample points, and multiple sets of load case points are obtained.

[0021] For each load interval, a linear model is fitted using the least squares method. This linear model describes the relationship between the sample points and the PID controller parameters. The linear model includes the three main parameters of the PID controller: proportional gain, integral gain, and derivative gain. Through the least squares fitting process, the PID parameters for each load interval are obtained. The core idea of ​​the least squares method is to minimize the sum of squared errors between the model's predicted values ​​and the actual data. Using this method, the corresponding PID parameters for each set of sample points are calculated, generating multiple sets of load case points.

[0022] A400: Associates and stores the fixed load interval sequence and multiple sets of load case points to complete the construction of the load case library.

[0023] Each load interval is associated with its corresponding PID parameter to construct a load case library. This library is structured similarly to a dictionary or mapping table, where the key is the load interval and the value is the PID parameter for that interval. The load case library is stored in a way that facilitates quick retrieval and interpolation, improving the efficiency of subsequent operations. For example, a load case might contain the following entries: {Load 200, PID parameter [−0.0025, 0.02, 0.02]}; {Load 250, PID parameter [−0.0030, 0.025, 0.025]}.

[0024] A500: Input the real-time boiler load value into the load case database for case matching and retrieval. If there are identical load values, return K identical load case points, where K≥2 and K is a positive integer.

[0025] During boiler operation, the boiler load value is collected in real time. This real-time load value is a crucial parameter reflecting the boiler's current operating status and directly impacts the furnace outlet pressure control requirements. The real-time boiler load value is input into a load case library for matching and retrieval, checking for any matching case points. If a load value identical to the real-time boiler load value exists in the library, K corresponding case points with the same load are returned. The number of case points K is at least 2 (K≥2). The selection of K ensures sufficient historical data for optimization, guaranteeing the stability of the optimization results. If no completely matching case exists in the load case library, the PID parameters of the two load values ​​closest to the real-time load value are selected for further interpolation processing.

[0026] A600: Executes the arithmetic mean of the K sets of PID parameters for the K cases with the same load, and outputs the real-time optimization parameters.

[0027] K sets of PID parameters, including proportional coefficient, integral coefficient and derivative coefficient, are extracted from K cases with the same load. The arithmetic mean of these K sets of PID parameters is calculated, that is, the mean is processed for each parameter separately, to obtain a comprehensive real-time optimization parameter. This real-time optimization parameter represents the PID setting that is most suitable for controlling the furnace outlet pressure under the current load.

[0028] A700: The real-time optimized parameters are sent to the PID controller to execute furnace outlet pressure control.

[0029] The calculated real-time optimized parameters are sent to the boiler's PID controller. The PID controller is responsible for adjusting control variables such as the induced draft fan speed in real time based on these parameters, thereby regulating the furnace outlet pressure. The PID controller uses the real-time optimized parameters and adjusts the control output based on the deviation between the real-time furnace outlet pressure value and the target value to ensure that the furnace outlet pressure is kept as close as possible to the set target value.

[0030] Furthermore, the method also includes:

[0031] A710: Input the real-time boiler load value into the load case database for case matching retrieval. If no identical load value exists, calculate the two nearest load case points based on the cosine distance. A720: Perform linear interpolation on the two sets of PID parameters of the two nearest load case points to obtain interpolation parameters. A730: Use a smoothing factor to perform weighted fusion of the real-time parameters and the interpolation parameters, and output fused parameters, where the smoothing factor ∈ [0,1]. A740: Send the fused parameters to the PID controller to perform furnace outlet pressure control.

[0032] The real-time boiler load value is input into a load case database for matching and retrieval. If no case point in the database perfectly matches the real-time boiler load value, the load interval closest to the real-time boiler load value is found using cosine distance. Specifically, in the load case database, the cosine distance between the load value of each case point and the real-time boiler load value is calculated. Cosine distance is a measure of the difference between the directions of two vectors; the smaller the cosine distance, the closer the two load values ​​are, indicating that their characteristics are more similar. By calculating the cosine distance, case points of the two load intervals closest to the real-time boiler load value are selected. The PID parameters of these two case points are used for subsequent interpolation processing.

[0033] For the two nearest load case points, their corresponding PID parameters are extracted and processed. Using the linear interpolation method, new PID parameters are calculated based on these two sets of PID parameters. The linear interpolation method assumes that the PID parameters change linearly with the load. Therefore, the PID parameters under the current load are linearly extrapolated based on the PID parameters of the two nearest load case points. The new parameters are calculated by weighted averaging between the two parameters.

[0034] For example, if no completely identical cases exist, the two closest sets of cases need to be found. Assuming the two sets are 180t and 210t, the calculation logic for case reuse is as follows: d1 = a1 + (b1 - a1) * (200 - 180) / (210 - 180). The above expression uses linear interpolation. Alternatively, non-linear methods or constraints can be used, such as limiting the maximum and minimum values ​​of the parameters, to prevent the parameters from becoming too large or too small as the number of optimization iterations increases. Furthermore, the obtained parameters need to be smoothed to prevent excessively large changes in parameters in a single iteration.

[0035] The smoothing factor is a value between 0 and 1 that controls the weighting between real-time and interpolation parameters. Specifically, a larger smoothing factor results in a higher weight for the real-time parameters, while a smaller smoothing factor results in a higher weight for the interpolation parameters. Based on the smoothing factor, the real-time and interpolation parameters are weighted and fused. This ensures smoothness during adjustment and prevents control instability caused by sudden changes. The weighted fusion result yields a fused parameter, which is used for PID control in the system.

[0036] The calculated fusion parameters are sent to the PID controller, which uses the fusion parameters to adjust variables such as the induced draft fan speed and frequency, thereby achieving precise control of the furnace outlet pressure and ensuring that the furnace outlet pressure remains within the target range.

[0037] Furthermore, the method also includes:

[0038] A750: During the process of the PID controller executing furnace outlet pressure control, the furnace outlet pressure is collected to obtain the real-time furnace outlet pressure; A760: The real-time pressure deviation between the target furnace outlet pressure and the real-time furnace outlet pressure is compared; A770: If the real-time pressure deviation exceeds a preset deviation threshold, the actual adjustment of the induced draft fan during the control process is obtained; A780: The real-time optimization parameters are synchronously corrected using the actual adjustment of the induced draft fan and the real-time pressure deviation to obtain an updated PID parameter set; A790: The updated PID parameter set is updated in a comprehensive manner in the load case library based on the real-time boiler load value.

[0039] When the PID controller adjusts the furnace outlet pressure, it continuously monitors the furnace outlet pressure. The real-time pressure value is an important basis for the control system to adjust equipment such as the induced draft fan. The change in furnace outlet pressure is obtained in real time through the pressure sensor installed in the boiler furnace, and this data serves as a real-time feedback signal.

[0040] The target furnace outlet pressure is a set value based on boiler operating conditions and requirements. This value is adjusted by the system operator according to boiler load and operating conditions. The real-time furnace outlet pressure is compared with the target furnace outlet pressure to obtain the real-time pressure deviation, which reflects the difference between the actual pressure and the target value.

[0041] Set an allowable deviation threshold, such as ±10Pa. When the real-time pressure deviation exceeds this threshold, the pressure regulation effect is considered unsatisfactory and further adjustments are needed. At this time, record the actual adjustment of the induced draft fan. The actual adjustment of the induced draft fan is achieved by changing the fan speed, and the change in speed will affect the furnace outlet pressure.

[0042] Based on the real-time pressure deviation and the actual adjustment of the induced draft fan, the current real-time optimization parameters are corrected. Specifically, the proportional, integral, and derivative coefficients are dynamically adjusted according to the magnitude of the real-time pressure deviation and the adjustment range of the induced draft fan. Using a gradient descent-like method, the values ​​of the real-time optimization parameters are optimized to more effectively adjust the pressure. During this process, various control parameters are optimized based on the current pressure deviation and adjustment amount. After synchronous correction, an updated PID parameter set is generated, which is used in subsequent control processes.

[0043] Based on the real-time boiler load value, the corresponding load point in the load case library is located. If the load value already exists in the case library, the new updated PID parameter set is written to that load point. Each updated PID parameter set is recorded with a timestamp to ensure that historical parameters are not forgotten. If multiple parameter sets are already stored for a load point, the oldest parameter set is deleted to keep the case library up-to-date. Whenever new optimized PID parameters are generated, the old parameters for that load point in the case library are overwritten to ensure that the case library always stores the latest optimization results.

[0044] Furthermore, if the real-time pressure deviation does not exceed the preset deviation threshold, the real-time boiler load value and real-time optimization parameters are written into the load case library.

[0045] When the real-time pressure deviation does not exceed the preset deviation threshold, meaning the pressure deviation is within the allowable range, this indicates that the current control strategy is effective and no further adjustments are needed. At this point, the current real-time boiler load value and the corresponding real-time optimized parameters are written into the load case library. This helps maintain the up-to-date state of the load case library, making it convenient to directly use these latest optimization results when encountering similar operating conditions in the future. If a similar load point already exists in the case library, the parameters for that load point are updated with a new timestamp, ensuring that the case library contains the optimal and latest control parameters.

[0046] Furthermore, the real-time optimization parameters are synchronously corrected using the actual adjustment amount of the induced draft fan and the real-time pressure deviation to obtain an updated PID parameter set. The method includes:

[0047] A781: Extract the initial proportional coefficient, initial integral coefficient, and initial derivative coefficient from the real-time optimization parameters; A782: Pre-construct the proportional coefficient update formula, which is as follows: ;in, The initial scaling factor is... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the scaling factor, The learning rate is [value], and the data volume is [value]. A783: Substitute the initial proportional coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the proportional coefficient update formula, and calculate and output the updated proportional coefficient using the gradient descent method; A784: Pre-construct the integral coefficient update formula, which is as follows: ;in, The initial integration coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the integral coefficients, The learning rate is [value], and the data volume is [value]. A785: Substitute the initial integral coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the integral coefficient update formula, and calculate and output the updated integral coefficient using the gradient descent method; A786: Pre-construct the differential coefficient update formula, which is as follows: ;in, The initial differential coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the differential coefficients, The learning rate is [value], and the data volume is [value]. A787: Substitute the initial differential coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the differential coefficient update formula, and use the gradient descent method to calculate and output the updated differential coefficient; wherein, the updated proportional coefficient, the updated integral coefficient, and the updated differential coefficient constitute the updated PID parameter group.

[0048] Each specific coefficient is extracted from the real-time optimization parameters. These extracted coefficients serve as the basis for subsequent parameter correction and optimization, especially when performing pressure deviation correction and case updates, where they will become key data for the next step of the operation.

[0049] The proportional gain update formula is designed based on the gradient descent method, an optimization algorithm used to update control parameters according to the error (deviation) and the control input (adjustment input). The learning rate is a constant that controls the speed at which parameters are updated. It determines the magnitude of the adjustment and is a very small value, on the order of the data size. To prevent overly drastic parameter updates, the proportional coefficient update formula updates the initial proportional coefficient by combining real-time pressure deviation and induced draft fan adjustment. The resulting updated proportional coefficient is used in the next control cycle to improve control performance.

[0050] Substitute the initial proportional coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the proportional coefficient update formula, and then apply the learning rate. The gradient descent method, an optimization algorithm, adjusts control parameters—specifically, the initial proportional gain—by ​​calculating the impact of errors on these parameters. In this example, gradient descent adjusts the initial proportional gain to improve the control system's response to pressure deviations and smooth the control process. The updated proportional gain, calculated by gradient descent, is then used for subsequent control decisions, ensuring the controller continuously optimizes its behavior to adapt to different operating conditions and loads.

[0051] The integral coefficient update formula is based on the gradient descent method. The goal is to adjust the initial integral coefficient by using real-time pressure deviation and induced draft fan adjustment. The initial integral coefficient is responsible for accumulating error in PID control and adjusting the output according to the accumulated error. By adjusting the initial integral coefficient, the PID controller can respond more accurately to long-term accumulated error, thereby optimizing the control effect.

[0052] By substituting the initial integral coefficients, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the integral coefficient update formula, the gradient descent method gradually updates the initial integral coefficients by calculating the impact of errors on them. This method allows the PID controller to better adjust its output, reducing accumulated errors. After each calculation, the initial integral coefficients are updated for use in the next control cycle. The updated integral coefficients, calculated using gradient descent, are then used in the next calculation of the control system. This ensures that the PID controller can progressively optimize its control performance and adapt to various real-time operating conditions.

[0053] In a PID controller, the derivative coefficient is used to handle the rate of change of error. It effectively predicts future trends in error and allows for appropriate adjustments. To ensure the PID controller can respond promptly to rapid pressure changes, the initial derivative coefficient is dynamically adjusted based on real-time pressure deviation and fan adjustment. By measuring the rate of change of error, the initial derivative coefficient helps the control system anticipate future pressure changes, avoiding overly fast or slow responses. The formula indicates that the initial derivative coefficient will be adjusted based on the current value according to real-time pressure deviation and fan adjustment. This update process helps the PID controller better cope with rapid system changes, avoiding slow response or over-adjustment.

[0054] By substituting the initial derivative coefficients, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the derivative coefficient update formula, the gradient descent method calculates the impact of the error on the initial derivative coefficients and gradually adjusts their values. This helps the PID controller better respond to error changes and avoids excessive fluctuations. After calculating the updated derivative coefficients using the gradient descent method, these coefficients are used in the next control operation. This allows the derivative control section to respond promptly to system changes, enhancing the overall control performance.

[0055] By continuously updating the proportional, integral, and derivative coefficients, the PID controller can gradually adjust its control behavior based on real-time data and changes in operating conditions, thereby achieving a more precise and stable control effect. The calculated updated proportional, integral, and derivative coefficients constitute the updated PID parameter set to achieve efficient control of the furnace outlet pressure.

[0056] Furthermore, a predefined fixed load interval sequence is used to group the multiple historical operating samples into load intervals to obtain multiple sets of sample points. The method includes:

[0057] A210: Predefined fixed load interval sequence is used to group the load intervals of the multiple historical operation samples to obtain multiple load interval sample sets, wherein each load interval sample set satisfies that the primary fan speed frequency and the secondary fan speed frequency remain unchanged within the same load interval; A220: Based on the induced draft fan adjustment action, the multiple load interval sample sets are aggregated within the group to obtain the multiple sample points.

[0058] Based on the actual load conditions of the boiler operation, a series of fixed load ranges are predefined, such as 150-170t, 170-190t, 190-210t, etc. These ranges are divided according to the stable range of boiler load under different operating conditions. Multiple historical operating samples, i.e., boiler load, fan speed, and furnace outlet pressure data at different time points, are classified according to load ranges. This means that data with load values ​​falling within the same range will be grouped together.

[0059] For each load range, ensure that the primary fan speed frequency and the secondary fan speed frequency remain constant within that range. This is because the speed of these two fans has a direct impact on the furnace outlet pressure. Therefore, maintaining their stability can reduce the fluctuations caused by fan adjustments within the same load range and ensure data comparability.

[0060] Within the same load range, the adjustment of the induced draft fan, i.e., the change in the induced draft fan speed, has a significant impact on the change in the furnace outlet pressure. At the same time, it should be understood that within a certain load range, when the speed and frequency of the primary fan and the secondary fan remain unchanged, the amount of adjustment of the induced draft fan speed and frequency corresponds to the amount of change in the furnace outlet pressure. This is used as a sample point constructed from a single induced draft fan adjustment event.

[0061] Based on this, for a set of load interval sample sets within the same load interval, the sample data is aggregated using the aggregation rule of fixed primary fan speed frequency and secondary fan speed frequency, but induced draft fan speed frequency adjustment. The data within the same interval are aggregated into a set of sample points. Each sample point represents the impact of induced draft fan adjustment under a certain combination of fixed primary fan speed frequency and secondary fan speed frequency within the load interval. Each induced draft fan adjustment event generates an independent sample point.

[0062] For example, within a load range of 170t ± 5t, under the condition that the primary fan speed frequency and secondary fan speed frequency remain constant, each induced draft fan adjustment event (including the induced draft fan speed adjustment amount and the corresponding furnace outlet pressure change) generates an independent sample point. This sample point includes: the sample induced draft fan speed frequency adjustment amount, the sample furnace outlet pressure change amount, the sample pressure change cumulative value (the historical pressure change accumulation within the same load range), and the sample adjacent pressure change difference. All sample points are used to construct a linear model, and the PID is solved using the least squares method. Further, each sample point includes the sample induced draft fan speed frequency adjustment amount, the sample furnace outlet pressure change amount, the sample pressure change cumulative value, and the sample adjacent pressure change difference.

[0063] The sample induced draft fan speed and frequency adjustment records the change in induced draft fan speed and frequency for each adjustment, reflecting the fan's adjustment action. The sample furnace outlet pressure change records the change in furnace outlet pressure after the induced draft fan adjustment, indicating the direct impact of the fan adjustment on the furnace outlet pressure. The sample pressure change cumulative value accumulates the pressure changes within the same load range, showing the overall trend over a period of time and helping the control system predict whether the system is in a stable state. The sample adjacent pressure change difference records the difference between the current event and the previous event; here, "event" refers to a single induced draft fan adjustment event, reflecting the rate and fluctuation of furnace outlet pressure change. These features help the model better understand how to adjust PID parameters based on different induced draft fan adjustment actions and historical pressure changes.

[0064] Furthermore, each historical operating sample includes the sample boiler load, sample primary air fan speed and frequency, sample secondary air fan speed and frequency, sample induced draft fan speed and frequency adjustment, and sample furnace outlet pressure change.

[0065] The sample boiler load record records the boiler load value for each sample, representing the boiler's load status. Load is one of the important factors affecting various parameters of the boiler system. The sample primary air fan speed frequency record records the speed frequency of the primary air fan, serving as an important parameter for furnace airflow regulation. The sample secondary air fan speed frequency record records the speed frequency of the secondary air fan, which also affects airflow and pressure within the furnace. The sample induced draft fan speed frequency adjustment record records the change in induced draft fan speed, directly affecting changes in furnace outlet pressure. The sample furnace outlet pressure change record records the change in furnace outlet pressure after each adjustment, reflecting the adjustment effect. By collecting the above historical operating data, the operating status of the boiler system can be analyzed from multiple dimensions.

[0066] Example 2, based on the same inventive concept as the adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler in the foregoing examples, such as... Figure 2As shown in the embodiment of this application, an adaptive control system for the furnace outlet pressure of a circulating fluidized bed boiler is provided, the system comprising:

[0067] The system comprises the following modules: a historical operation sample acquisition module 10, which acquires multiple historical operation samples by collecting historical operation data; a load interval grouping processing module 20, which predefines a fixed load interval sequence and performs load interval grouping processing on the multiple historical operation samples to obtain multiple sets of sample points; a PID parameter solving module 30, which uses the least squares method to fit a linear model and solve for the PID parameters of the multiple sets of sample points to obtain multiple sets of load case points; a load case library construction module 40, which associates and stores the fixed load interval sequence and multiple sets of load case points to complete the construction of the load case library; a case matching and retrieval module 50, which inputs the real-time boiler load value into the load case library for case matching and retrieval. If there are identical load values, it returns K identical load case points, where K ≥ 2 and K is a positive integer; a real-time optimization parameter output module 60, which performs arithmetic mean calculation on the K sets of PID parameters of the K identical load case points and outputs real-time optimization parameters; and a furnace outlet pressure control module 70, which sends the real-time optimization parameters to the PID controller to perform furnace outlet pressure control.

[0068] Furthermore, the furnace outlet pressure control module 70 is used to perform the following operational steps:

[0069] The real-time boiler load value is input into the load case database for case matching retrieval. If no identical load value is found, the two nearest load case points are returned based on cosine distance calculation. Linear interpolation is performed on the two sets of PID parameters of the two nearest load case points to obtain interpolation parameters. A smoothing factor is used to perform weighted fusion of the real-time parameters and the interpolation parameters to output fused parameters, where the smoothing factor ∈ [0,1]. The fused parameters are sent to the PID controller to perform furnace outlet pressure control.

[0070] Furthermore, the furnace outlet pressure control module 70 is used to perform the following operational steps:

[0071] During the process of the PID controller executing furnace outlet pressure control, the furnace outlet pressure is acquired to obtain the real-time furnace outlet pressure; the real-time pressure deviation between the target furnace outlet pressure and the real-time furnace outlet pressure is compared; if the real-time pressure deviation exceeds a preset deviation threshold, the actual adjustment of the induced draft fan during the control process is obtained; the real-time optimization parameters are synchronously corrected using the actual adjustment of the induced draft fan and the real-time pressure deviation to obtain an updated PID parameter set; and the updated PID parameter set is updated in a comprehensive manner in the load case library based on the real-time boiler load value.

[0072] Furthermore, if the real-time pressure deviation does not exceed the preset deviation threshold, the real-time boiler load value and real-time optimization parameters are written into the load case library.

[0073] Furthermore, the furnace outlet pressure control module 70 is used to perform the following operational steps:

[0074] The initial proportional coefficient, initial integral coefficient, and initial derivative coefficient are extracted from the real-time optimization parameters; a proportional coefficient update formula is pre-constructed, which is as follows: ;in, The initial scaling factor is... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the scaling factor, The learning rate is [value], and the data volume is [value]. Substitute the initial proportional coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the proportional coefficient update formula, and calculate and output the updated proportional coefficient using the gradient descent method; pre-construct the integral coefficient update formula, which is as follows: ;in, The initial integration coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the integral coefficients, The learning rate is [value], and the data volume is [value]. Substitute the initial integral coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the integral coefficient update formula, and calculate and output the updated integral coefficient using the gradient descent method; pre-construct the differential coefficient update formula, which is as follows: ;in, The initial differential coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the differential coefficients, The learning rate is [value], and the data volume is [value]. Substitute the initial differential coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the differential coefficient update formula, and use the gradient descent method to calculate and output the updated differential coefficient; wherein, the updated proportional coefficient, the updated integral coefficient, and the updated differential coefficient constitute the updated PID parameter group.

[0075] Furthermore, the load interval grouping processing module 20 is used to perform the following operation steps:

[0076] A predefined fixed load interval sequence is used to group the load intervals of the multiple historical operating samples to obtain multiple load interval sample sets. Each load interval sample set satisfies that the primary fan speed frequency and the secondary fan speed frequency remain unchanged within the same load interval. Based on the induced draft fan adjustment action, the multiple load interval sample sets are aggregated within the group to obtain the multiple sample points.

[0077] Furthermore, each sample point includes the sample induced draft fan speed frequency adjustment, the sample furnace outlet pressure change, the cumulative value of sample pressure change, and the pressure change difference between adjacent samples.

[0078] Furthermore, each historical operating sample includes the sample boiler load, sample primary air fan speed and frequency, sample secondary air fan speed and frequency, sample induced draft fan speed and frequency adjustment, and sample furnace outlet pressure change.

[0079] Through the foregoing detailed description of the adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler, those skilled in the art can clearly understand the adaptive control system for furnace outlet pressure of a circulating fluidized bed boiler in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to in the method section.

[0080] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for adaptive control of furnace outlet pressure in a circulating fluidized bed boiler, characterized in that, The method includes: By collecting historical operational data, multiple historical operational samples were obtained; A predefined fixed load interval sequence is used to group the load intervals of the multiple historical operation samples to obtain multiple sets of sample points; The least squares method is used to fit a linear model to solve the PID parameters of the multiple sets of sample points, thus obtaining multiple sets of load case points; The fixed load interval sequence and multiple sets of load case points are associated and stored to complete the construction of the load case library; Input the real-time boiler load value into the load case database for case matching and retrieval. If there are identical load values, return K identical load case points, where K≥2 and K is a positive integer. Calculate the arithmetic mean of the K sets of PID parameters for the K cases with the same load, and output the real-time optimization parameters. The real-time optimized parameters are sent to the PID controller to execute furnace outlet pressure control; The method further includes: The real-time boiler load value is input into the load case database for case matching and retrieval. If no identical load value is found, the two nearest load case points are returned based on the cosine distance calculation. Linear interpolation is performed on the two sets of PID parameters for the two nearest load case points to obtain the interpolation parameters; A smoothing factor is used to perform weighted fusion of the real-time parameters and the interpolation parameters, and the fused parameters are output, wherein the smoothing factor ∈ [0,1]. The fusion parameters are sent to the PID controller to perform furnace outlet pressure control.

2. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 1, characterized in that, The method further includes: During the process of the PID controller performing furnace outlet pressure control, the furnace outlet pressure is collected to obtain the real-time furnace outlet pressure. Compare the real-time pressure deviation between the target furnace outlet pressure and the real-time furnace outlet pressure; If the real-time pressure deviation exceeds the preset deviation threshold, the actual adjustment amount of the induced draft fan in the control process is obtained; The real-time optimization parameters are synchronously corrected using the actual adjustment amount of the induced draft fan and the real-time pressure deviation to obtain an updated PID parameter set; Based on the real-time boiler load value, the updated PID parameter group is updated in a comprehensive manner in the load case library.

3. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 2, characterized in that, If the real-time pressure deviation does not exceed the preset deviation threshold, the real-time boiler load value and real-time optimization parameters are written into the load case library.

4. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 2, characterized in that, The method involves synchronously correcting the real-time optimization parameters using the actual adjustment of the induced draft fan and the real-time pressure deviation to obtain an updated PID parameter set. The initial proportional coefficient, initial integral coefficient, and initial derivative coefficient are extracted from the real-time optimization parameters; A pre-constructed scaling factor update formula is provided, which is as follows: ； in, The initial scaling factor is... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the scaling factor, The learning rate is [value], and the data volume is [value]. ; Substitute the initial proportional coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the proportional coefficient update formula, and use the gradient descent method to calculate and output the updated proportional coefficient. A pre-constructed integral coefficient update formula is provided, which is as follows: ； in, The initial integration coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the integral coefficients, The learning rate is [value], and the data volume is [value]. ; Substitute the initial integral coefficient, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the integral coefficient update formula, and use the gradient descent method to calculate and output the updated integral coefficient. A pre-constructed formula for updating differential coefficients is provided, as follows: ； in, The initial differential coefficients are... The real-time pressure deviation is referred to as [the real-time pressure deviation]. This refers to the actual adjustment amount of the induced draft fan. To update the differential coefficients, The learning rate is [value], and the data volume is [value]. ; Substitute the initial differential coefficients, the actual adjustment of the induced draft fan, and the real-time pressure deviation into the differential coefficient update formula, and use the gradient descent method to calculate and output the updated differential coefficients; The updated proportional coefficient, updated integral coefficient, and updated derivative coefficient constitute the updated PID parameter group.

5. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 1, characterized in that, The method involves grouping multiple historical operating samples into load intervals using a predefined fixed load interval sequence to obtain multiple sets of sample points. A predefined fixed load interval sequence is used to group the load intervals of the multiple historical operation samples to obtain multiple load interval sample sets. Each load interval sample set satisfies that the primary fan speed frequency and the secondary fan speed frequency remain unchanged within the same load interval. Based on the adjustment action of the induced draft fan, the sample sets of the multiple load intervals are aggregated within each group to obtain the multiple sample points.

6. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 5, characterized in that, Each sample point includes the sample induced draft fan speed and frequency adjustment, the sample furnace outlet pressure change, the cumulative value of sample pressure change, and the pressure change difference between adjacent samples.

7. The adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler as described in claim 1, characterized in that, Each historical operating sample includes the sample boiler load, sample primary air fan speed and frequency, sample secondary air fan speed and frequency, sample induced draft fan speed and frequency adjustment, and sample furnace outlet pressure change.

8. An adaptive control system for furnace outlet pressure in a circulating fluidized bed boiler, characterized in that, For implementing the adaptive control method for furnace outlet pressure of a circulating fluidized bed boiler according to any one of claims 1-7, the system comprises: The historical operation sample acquisition module is used to obtain multiple historical operation samples by collecting historical operation data; The load interval grouping processing module is used to predefine a fixed load interval sequence and perform load interval grouping processing on the multiple historical operation samples to obtain multiple groups of sample points; The PID parameter solving module is used to solve the PID parameters of the multiple sets of sample points by fitting a linear model using the least squares method, and to obtain multiple sets of load case points. The load case library construction module is used to associate and store the fixed load interval sequence and multiple sets of load case points to complete the construction of the load case library; The case matching and retrieval module is used to input the real-time boiler load value into the load case database for case matching and retrieval. If there are identical load values, it returns K identical load case points, where K≥2 and K is a positive integer. The real-time optimization parameter output module is used to perform an arithmetic mean on the K sets of PID parameters of the K cases with the same load, and output the real-time optimization parameters. The furnace outlet pressure control module is used to send the real-time optimized parameters to the PID controller to execute furnace outlet pressure control.

Images