A method and device for winter antifreeze control of indirect air-cooled system of coal-fired unit

Abstract

This application provides a method and device for winter anti-freezing control of an indirect air-cooled system in a coal-fired power unit. The method includes obtaining a time-series feature matrix based on real-time data; an edge server obtaining predicted wall temperature data for each cooling triangle based on the time-series feature matrix and prediction model weight information obtained by a cloud server based on the time-series feature matrix and a deep neural network; and combining the physical freezing critical temperature and anti-freezing safety buffer compensation term to obtain the corresponding wall temperature safety margin for the cooling triangle; and obtaining an anti-freezing strategy for the fan-shaped area based on the wall temperature safety margin to control the anti-freezing structure. This method decouples the contradiction between "heavy learning" and "rapid inference," ensuring the accuracy and real-time performance of anti-freezing protection, as well as accurately predicting the delayed "overcooling" effect caused by sudden changes in the environmental wind field. Prediction accuracy and lead time are significantly improved, while achieving refined anti-freezing at the cooling triangle level, thus enhancing the safety and economy of coal-fired power unit operation.

Description

Technical Field

[0001] This application relates to the fields of industrial artificial intelligence control and winter operation safety technology of thermal power generation, specifically to a method and device for winter antifreeze control of an indirect air-cooled system for a coal-fired unit. Background Technology

[0002] Indirect air-cooling systems are widely used in coal-fired power plants in coal-rich but water-scarce regions due to their significant water-saving effects. Their core heat exchange element—the cooling triangle, or radiator tube bundle—is typically arranged in a ring at the bottom of the air-cooled tower. In extreme winter temperatures and strong winds, the heat exchange conditions vary greatly between different fan-shaped areas and different cooling triangle tube bundles. For example, the heat exchange conditions in the windward downstream section differ significantly from those in the leeward upstream section. This can easily lead to localized overcooling and icing in some tube bundles, causing them to freeze, crack, and leak.

[0003] Existing antifreeze monitoring and protection technologies suffer from the following technical bottlenecks: (1) The early warning model lacks the ability to extract time-series features: Traditional methods are mostly based on steady-state thermodynamic formulas or simple static neural networks, which cannot effectively capture the dynamic evolution of factors such as violent fluctuations in the environmental wind field and delays in hydraulic distribution in the time dimension, resulting in lagging predictions and low accuracy. (2) The control architecture is difficult to meet the stringent real-time requirements: Existing antifreeze interventions mostly rely on human experience or the conventional calculations of DCS (Distributed Control System), which have long calculation cycles and large communication delays, making it difficult to achieve millisecond-level rapid intervention at the critical point when the control bundle is about to freeze. (3) Lack of cloud-edge collaboration mechanism: Complex deep learning models require extremely high computing power and cannot be directly deployed in the underlying controllers of industrial sites, resulting in the model being unable to balance "high-precision prediction" and "extremely low-latency execution". Summary of the Invention

[0004] This application addresses the problems existing in the prior art by providing a method and device for winter antifreeze control of an indirect air-cooled system for coal-fired power units. It decouples the contradiction between "heavy learning" and "rapid inference," ensuring the accuracy and real-time performance of antifreeze protection. It accurately predicts the lag-induced "overcooling" effect caused by sudden changes in the environmental wind field, with prediction accuracy and lead time significantly superior to traditional calculations based solely on static thresholds obtained from return water temperature. Simultaneously, it achieves refined antifreeze protection at the cooling triangle level, avoiding the heat exchange efficiency loss caused by traditional fan-shaped area-level adjustments, thereby improving the safety and economy of coal-fired power unit operation.

[0005] To achieve the above objectives, the technical solution adopted in this application is as follows: In a first aspect, this application proposes a method for winter anti-freezing control of an indirect air-cooled system for a coal-fired unit. The indirect air-cooled system includes a fan-shaped region, which includes at least one cooling triangle. An anti-freezing structure is configured in the fan-shaped region. The control method includes: collecting real-time data; the real-time data includes the operating parameters of the coal-fired unit and the control parameters of the anti-freezing structure; the operating parameters include the ambient wind speed; and obtaining a time-series feature matrix based on the real-time data. The edge server obtains the wall temperature prediction data for each cooling triangle based on the time-series feature matrix and the prediction model weight information; the prediction model weight information is obtained by the cloud server based on the time-series feature matrix and deep neural network and transmitted to the edge server, and the time-series feature matrix is ​​transmitted from the edge server to the cloud server. The cumulative duration of the effect of the physical freezing critical temperature of circulating water and extreme ambient wind speed on the cooling triangle was obtained. The antifreeze safety buffer compensation item is obtained based on the ambient wind speed and cumulative duration; the wall temperature safety margin of the corresponding cooling triangle is obtained based on the predicted wall temperature data, the physical freezing critical temperature, and the antifreeze safety buffer compensation item. Based on the wall temperature safety margin, obtain the antifreeze strategy for the sector area; Control the antifreeze structure according to the antifreeze strategy.

[0006] In some embodiments, the antifreeze structure includes a cooling fan, at least one louver, and at least one bypass circulation pipe. The at least one louver and the at least one bypass circulation pipe correspond one-to-one with at least one cooling triangle. The louver is disposed around the periphery of the corresponding cooling triangle. The cooling fan is used to increase the airflow rate within the fan-shaped area. The bypass circulation pipe is disposed between the inlet pipe and the return pipe of the corresponding cooling triangle. The antifreeze structure is controlled according to the antifreeze strategy, including: controlling the opening degree of the louvers corresponding to the cooling triangle, and / or controlling the speed of the cooling fan, and / or controlling the on / off of the bypass circulation pipeline.

[0007] In some embodiments, the antifreeze strategy for the fan-shaped region includes a partitioned antifreeze strategy for each cooling triangle; The zoned anti-freeze strategy includes a normal monitoring strategy, a primary anti-freeze strategy, a secondary anti-freeze strategy, and a tertiary anti-freeze strategy. The normal monitoring strategy is characterized by the louver opening corresponding to the cooling triangle being within the normal operating range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating range. The primary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within the preset anti-freeze opening range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating range. The secondary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within the preset anti-freeze opening range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the preset anti-freeze speed range. The tertiary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within the preset anti-freeze opening range, the bypass circulation pipe being connected, and the cooling fan speed being within the preset anti-freeze speed range.

[0008] In some embodiments, the antifreeze state of each cooling triangle includes a normal antifreeze state, a first antifreeze state, a second antifreeze state, and a third antifreeze state. The normal antifreeze state represents the antifreeze state that executes the normal monitoring strategy, the first antifreeze state represents the antifreeze state that executes the first-level antifreeze strategy, the second antifreeze state represents the antifreeze state that executes the second-level antifreeze strategy, and the third antifreeze state represents the antifreeze state that executes the third-level antifreeze strategy. Based on the wall temperature safety margin, the antifreeze strategy for the sector area is obtained, including: Under normal antifreeze conditions, determine whether the wall temperature safety margin of the cooling triangle is not greater than the first preset safety margin; When the wall temperature safety margin is not greater than the first preset safety margin, the zone antifreeze strategy of the corresponding cooling triangle is determined to be the first-level antifreeze strategy. In the first antifreeze state, at the first evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time. The first evaluation time represents the time after the first preset time has elapsed since the first initial time. The first initial time represents the time when the antifreeze structure corresponding to the cooling triangle executes the first-level antifreeze strategy. When the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time period, the corresponding cooling triangle's zone antifreeze strategy is determined to be a secondary antifreeze strategy. In the second antifreeze state, at the second evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time. The second evaluation time represents the time after the second preset time has elapsed since the second initial time. The second initial time represents the time when the antifreeze structure corresponding to the cooling triangle executes the secondary antifreeze strategy. When the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time period, the corresponding cooling triangle's zone antifreeze strategy is determined to be a level three antifreeze strategy.

[0009] In some embodiments, obtaining the antifreeze strategy for the fan-shaped region based on the wall temperature safety margin further includes: in the third antifreeze state, determining whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, wherein the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, it is determined whether the first duration is greater than the third preset duration. The first duration represents the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the third antifreeze state. When the first duration exceeds the third preset duration, the corresponding cooling triangle's partition antifreeze strategy is determined to be a secondary antifreeze strategy.

[0010] In some embodiments, under the second antifreeze state, when the wall temperature safety margin of the cooling triangle does not continue to decrease within the second preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, and the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, it is determined whether the second duration is greater than the fourth preset duration. The second duration represents the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the second antifreeze condition. When the second duration is longer than the fourth preset duration, the corresponding cooling triangle's partition antifreeze strategy is determined to be a level one antifreeze strategy.

[0011] In some embodiments, under the first antifreeze state, when the wall temperature safety margin of the cooling triangle does not continue to decrease within a first preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, and the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, it is determined whether the third duration is greater than the fifth preset duration. The third duration represents the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the first antifreeze state. When the duration of the third phase exceeds the preset duration of the fifth phase, the corresponding cooling triangle's zone antifreeze strategy is determined to be a normal monitoring strategy.

[0012] In some embodiments, the edge server also obtains the icing probability of each cooling triangle within the prediction time window based on the temporal feature matrix and prediction model weight information; and obtains the anti-freezing strategy for the fan-shaped region based on the wall temperature safety margin, further including: Under normal antifreeze conditions, determine whether the probability of icing in the cooling triangle is greater than the first risk threshold; When the probability of freezing is greater than the first risk threshold, the corresponding cooling triangle's zone antifreeze strategy is determined to be a level one antifreeze strategy. Under the first antifreeze condition, at the first evaluation time, it is determined whether the probability of icing of the cooling triangle within the first preset time period is greater than the second risk threshold. When the probability of icing in the cooling triangle within the first preset time period is greater than the second risk threshold, the partition antifreeze strategy of the corresponding cooling triangle is determined to be a secondary antifreeze strategy. Under the second antifreeze condition, at the second evaluation time, it is determined whether the probability of icing of the cooling triangle within the second preset time period is greater than the third risk threshold. When the probability of icing in the cooling triangle within the second preset time period is greater than the third risk threshold, the partition antifreeze strategy of the corresponding cooling triangle is determined to be a level three antifreeze strategy. The first risk threshold is less than the second risk threshold, and the second risk threshold is less than the third risk threshold.

[0013] In some embodiments, the operating parameters of the coal-fired unit also include ambient air temperature, mass flow rate of circulating water, inlet temperature of cooling triangular circulating water, and outlet temperature of cooling triangular circulating water.

[0014] Secondly, this application provides a winter anti-freezing control device for an indirect air-cooled system of a coal-fired unit, which is used to implement a winter anti-freezing control method for the indirect air-cooled system of a coal-fired unit. The control device includes a data acquisition unit, an edge server, a cloud server, a controller, and an anti-freezing structure, which is configured in a fan-shaped area of ​​the indirect air-cooled system. The data acquisition unit is used to collect real-time data, obtain the physical freezing critical temperature of circulating water and the cumulative duration of the continuous effect of extreme environmental wind speed on the cooling triangle, and transmit the real-time data to the edge server. The edge server is used to obtain a time-series feature matrix based on real-time data, transmit the time-series feature matrix to the cloud server, obtain the wall temperature prediction data for each cooling triangle based on the time-series feature matrix and prediction model weight information, obtain the anti-freeze safety buffer compensation term based on the ambient wind speed and cumulative duration, obtain the wall temperature safety margin of the corresponding cooling triangle based on the wall temperature prediction data, physical freezing critical temperature and anti-freeze safety buffer compensation term, and obtain the anti-freeze strategy for the sector area based on the wall temperature safety margin. The cloud server is used to obtain the prediction model weight information based on the time-series feature matrix and deep neural network, and to transmit the prediction model weight information to the edge server; The controller is used to control the antifreeze structure according to the antifreeze strategy.

[0015] Compared with the prior art, this application has the following advantages: 1. This application employs a cloud server to handle computationally intensive self-attention mechanism calculations and weight iteration updates. The operation of building the prediction model based on a deep neural network is configured within the cloud server to absorb massive historical time-series features and form a global nonlinear cognition. Meanwhile, the edge server only receives the lightweight prediction model weight information. The task of the edge server is extremely compressed into pure forward propagation matrix operations, thus ensuring that complex deep learning inference is compressed to a preset, extremely low page computation time, decoupling the contradiction between "intensive learning" and "ultra-fast inference," and ensuring the accuracy and real-time performance of anti-freeze protection. Simultaneously, through cloud server-side computationally intensive training, edge server-side lightweight inference, and high-frequency execution at the underlying edge, extremely high standards of real-time computation and communication latency constraints are met.

[0016] 2. This application can accurately predict the hysteretic "supercooling" effect caused by sudden changes in the environmental wind field by using the wall temperature safety margin. The prediction accuracy and lead time are significantly better than the traditional calculation based solely on the static threshold obtained from the return water temperature.

[0017] 3. This application achieves refined antifreeze at the cooling triangle level, avoiding the heat exchange efficiency loss caused by traditional fan-shaped area level adjustment, and improving the safety and economy of coal-fired unit operation. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the winter antifreeze control method for the indirect air-cooled system of a coal-fired unit in an embodiment of this application. Figure 2 This is a flowchart illustrating the antifreeze strategy for obtaining a sector-shaped region based on the wall temperature safety margin in an embodiment of this application. Figure I ; Figure 3 This is a flowchart illustrating the antifreeze strategy for obtaining a sector-shaped region based on the wall temperature safety margin in an embodiment of this application. Figure II ; Figure 4 This is a flowchart illustrating the antifreeze strategy for obtaining a sector-shaped region based on the wall temperature safety margin in an embodiment of this application. Figure III ; Figure 5 This is a schematic diagram of the signal transmission relationship of the winter antifreeze control device of the indirect air-cooled system of the coal-fired unit in the embodiments of this application.

[0019] The attached diagram is labeled as follows: 1000, coal-fired unit; 1100, indirect air-cooled system; 1110, sector area; 1111, cooling triangle; 1112, anti-freeze structure; 2100, data acquisition unit; 2200, edge server; 2300, cloud server; 2400, controller. Detailed Implementation

[0020] Existing research has addressed the problems in the background technology. Chinese patent application CN114739200A, "A Sector-Level Anti-Freezing Control Method, System, and Surface-Type Indirect Cooling System Sector," uses sector-level state judgment but does not use a time-series prediction model. Chinese patent application CN120926772A, "An Air-Cooled Island Anti-Freezing Control Method and System Based on AI Prediction and Two-Stage Optimization," uses an automatic encoder for risk assessment but does not introduce physical mechanism constraints, and the anti-freezing measures are limited to fan control, lacking a heat source fallback mechanism. Chinese patent application CN115307370A, "A Cold Fan Defrosting Control Method and Device Based on Cloud-Edge Coordination," while involving cloud-edge collaboration, is applied to the cold storage defrosting field and does not consider the safety zoning compliance requirements of the power industry.

[0021] To clearly illustrate the technical features of this solution, the implementation methods of this application will be described in detail below with reference to the accompanying drawings and embodiments. This will allow for a full understanding and implementation of how this application uses technical means to solve technical problems and achieve corresponding technical effects. The embodiments of this application and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this application.

[0022] See Figure 1 In a first aspect, embodiments of this application propose a winter anti-freezing control method for an indirect air-cooled system of a coal-fired unit. The indirect air-cooled system includes a fan-shaped region, which includes at least one cooling triangle. An anti-freezing structure is configured within the fan-shaped region. The control method includes: acquiring real-time data; the real-time data includes the operating parameters of the coal-fired unit and the control parameters of the anti-freezing structure; optionally, distributed fiber optic temperature measurement technology is used to acquire the wall temperature data of each cooling triangle in real time, the wall temperature data of the cooling triangle being the wall temperature data of the cooling triangle tube bundle, and simultaneously acquiring the operating parameters of the coal-fired unit and the control parameters of the anti-freezing structure in the distributed control system; the operating parameters... The data includes ambient wind speed. In some embodiments, the operating parameters also include ambient air temperature, circulating water mass flow rate, cooling triangle circulating water inlet temperature, and cooling triangle circulating water outlet temperature. A time-series feature matrix is ​​obtained based on real-time data. The multi-source heterogeneous real-time data is denoised, aligned, and normalized according to a preset time sliding window to construct a multi-dimensional time-series feature matrix. Optionally, for the collected multi-source heterogeneous real-time data, a Z-score normalization method is used for dimensionless normalization to eliminate the influence of different physical dimensions on the convergence of the deep neural network. The normalization formula is: ; In the formula, For the first The feature values ​​after normalization of real-time data. The sequence number of the real-time data. For the first Real-time data, For the first The mean of real-time data within a sliding time window. For the first The standard deviation of real-time data within a time sliding window; The edge server obtains the predicted wall temperature data for each cooling triangle based on the temporal feature matrix and the prediction model weight information; The cumulative duration of the effect of the physical freezing critical temperature of circulating water and extreme ambient wind speed on the cooling triangle was obtained. The antifreeze safety buffer compensation term is obtained based on ambient wind speed and cumulative duration; the corresponding wall temperature safety margin of the cooling triangle is obtained based on predicted wall temperature data, physical freezing critical temperature, and the antifreeze safety buffer compensation term; the wall temperature safety margin is based on the predicted wall temperature data of the cooling triangle tube bundle at future times. Physical freezing critical temperature of circulating water and antifreeze safety buffer compensation item By considering the cumulative effect of ambient wind speed and time, the impact of ambient wind speed is assessed in real time, enabling the safety redundancy limit to be automatically tightened under strong wind conditions. The wall temperature safety margin is: ; In the formula, To ensure a safety margin for wall temperature, To obtain the predicted wall temperature data for the cooling triangular tube bundle at future moments, Predicted based on time-series feature matrix and prediction model weight information. This refers to the physical freezing critical temperature of circulating water. This is a freeze protection buffer compensation term dynamically calculated based on the cumulative duration of the continuous effect of ambient wind speed and extreme ambient wind speed on the cooling triangular tube bundle. For ambient wind speed, The cumulative duration of the continuous effect of extreme environmental wind speed on the cooling triangular tube bundle; Optionally, antifreeze safety buffer compensation item Used to characterize the impact of ambient wind speed on the continuous heat transfer enhancement and local supercooling risk of the cooling triangular tube bundle; extreme wind speed characterizes ambient wind speed. Greater than the preset strong wind threshold When the ambient wind speed Greater than the preset strong wind threshold At that time, the ambient wind speed First time exceeding the preset strong wind threshold From that moment onwards, the duration of the continuous effect of extreme environmental wind speed on the cooling triangular tube bundle is accumulated according to the sampling period. When the ambient wind speed Not greater than the preset strong wind threshold And when the preset recovery time is reached, the effect will continue for the specified duration. Reset to zero, or adjust the duration of action according to the preset attenuation coefficient. Attenuation; antifreeze safety buffer compensation item Depending on ambient wind speed Increased duration of action It increases with the increase of.

[0023] In one optional implementation, the antifreeze safety buffer compensation term satisfies: ; in, This is the wind speed compensation coefficient. For duration of action compensation coefficient; wind speed compensation coefficient. Duration of action compensation coefficient and preset strong wind threshold All parameters were pre-calibrated based on historical operating data of coal-fired units, cooling triangular structure parameters, ambient temperature ranges, and on-site anti-freezing test results. Specifically, the wind speed compensation coefficient can be determined through linear regression, piecewise fitting, or table lookup methods, based on the correspondence between ambient wind speed, duration of strong winds, cooling triangular tube bundle wall temperature change rate, wall temperature safety margin, and icing alarm records in historical operating data. Duration of action compensation coefficient and preset strong wind threshold .

[0024] By using the wall temperature safety margin, the delayed "supercooling" effect caused by sudden changes in the environmental wind field can be accurately predicted. The prediction accuracy and lead time are significantly better than the traditional calculation based solely on the static threshold obtained from the return water temperature.

[0025] The prediction model weights are obtained by the cloud server based on the temporal feature matrix and deep neural network and then transmitted to the edge server. Conversely, the temporal feature matrix is ​​transmitted from the edge server to the cloud server. Freeze protection is a typical "hard real-time" task. Deploying complex deep neural networks directly on the edge results in computational latency far exceeding the critical tolerance for freezing in a sector-shaped region. However, combining offline training on the cloud server with ultra-fast inference on the edge server allows for the formation of global nonlinear cognition on the cloud server while ensuring that complex deep learning inference is compressed to a preset, extremely low page computation time, thus guaranteeing the real-time performance of freeze protection.

[0026] The prediction model weights are obtained by the cloud server based on the time-series feature matrix and deep neural network and transmitted to the edge server, while the time-series feature matrix is ​​transmitted from the edge server to the cloud server. Traditional pure data-driven deep learning models are prone to overfitting when facing complex and variable industrial thermal systems. Once the environmental wind field or load conditions exceed the data distribution patterns of the historical training set, the prediction accuracy of the pure data-driven deep learning model will plummet. In some embodiments, when the cloud server constructs and trains the prediction model based on historical datasets and deep neural networks, it hard-codes and embeds the aerodynamic and heat transfer physical mechanisms into the loss function of the deep neural network. This not only requires the predicted wall temperature output by the prediction model obtained based on the deep neural network to approximate the historical true value of distributed fiber optic temperature measurement, but also mandates that the output of the prediction model must strictly adhere to the law of conservation of energy. Specifically, the joint loss function is constructed as follows: ; ; In the formula, For the joint loss function, i.e. Let be the loss function of the deep neural network. The convective heat transfer coefficient can be calculated in real time using empirical formulas or obtained by looking up tables; details will not be elaborated here. The loss function responsible for learning the evolution of explicit temporal features. The weight coefficients for the physical constraint regularization term are typically obtained using a cross-validation grid search method; optionally, The range is 0.1~10. For physical constraint regularization terms, To measure the convective heat transfer coefficient of air, This parameter varies with wind speed and / or geometry; when applied to a cooling triangle, optionally, natural convection, i.e., when there is no wind, is also considered. The value range is usually 1 to 10. Forced convection, i.e., when there is wind, The value range is usually 10~120 , The effective heat dissipation coefficient of the cooling triangle, The overall heat transfer coefficient varies with operating conditions, typically The value range is 25~45 , For the wall temperature data of the cooling triangular tube bundle, The ambient air temperature, The mass flow rate of the circulating water. The specific heat capacity at constant pressure of circulating water, To cool the inlet water temperature of the triangular circulating water system, To cool the outlet water temperature of the triangular circulating water, For air-side convective heat transfer in an indirect air-cooling system, For water-side convective heat exchange in indirect air-cooled systems; By constraining the joint loss function, the prediction model built on the trial neural network acquires the physical common-sense boundaries of a "white box" during the "black box" mapping process. Even under extreme and abnormal gusts of wind, the prediction model can still output a predicted wall temperature safety margin value that conforms to physical logic, fundamentally improving the generalization ability and robustness of the prediction model. Compared with traditional methods, the prediction lead time can be increased from ≤30 seconds to ≥5 minutes, and the false alarm rate is expected to be reduced by about 60%.

[0027] Optionally, a computing acceleration card is configured in the cloud server, and the edge server transmits the historical dataset to the cloud server. The historical dataset represents a set of temporal feature matrices for historical periods. The cloud server constructs and trains a prediction model based on the historical dataset and a deep neural network. The prediction model includes a spatial feature extraction module and a temporal attention module. Optionally, the temporal attention module is constructed based on a multi-head attention mechanism. To capture the long-distance temporal dependency between environmental wind field evolution and local supercooling, a multi-head attention mechanism is used to calculate the attention weights of the temporal features. The calculation formula for the multi-head attention mechanism is as follows: ; In the formula, This represents the function for the multi-head attention mechanism. The query matrix is ​​obtained by mapping the input feature matrix obtained from the time-series feature matrix. The key matrix is ​​obtained by mapping the input feature matrix based on the temporal feature matrix. This is the value matrix obtained by mapping the input feature matrix based on the time-series feature matrix. Represents the normalized exponential function, with superscript indicating the function. Indicates transpose. The dimension of the key vector; During training, the temporal feature matrix is ​​used as input, and the wall temperature safety margin, or the icing probability of the predicted time window, or the wall temperature safety margin and the icing probability of the predicted time window are used as output. This allows the prediction model to implicitly learn the nonlinear mapping relationship between the evolution of the environment and the wall temperature or local supercooling of the fan-shaped region. The prediction model is forced to converge to the energy conservation plane through the joint loss function. The prediction model weight information is extracted from the prediction model after training. The connection between the cloud server and the external network is through a dedicated power line or VPN encrypted channel, and meets the requirements of the third level of network security protection.

[0028] The cloud server transmits the prediction model weight information to the edge server. The edge server then lightweights the prediction model weight file and deploys it to the intelligent computing engine in the industrial site. Based on the time-series feature matrix and under constraints below a preset page computation time threshold, the intelligent computing engine rapidly outputs the wall temperature safety margin, or the icing probability within the predicted time window, or both the wall temperature safety margin and the icing probability within the predicted time window, through forward propagation. The prediction model weight information is obtained through offline training and is updated periodically according to actual needs during use. The update cycle can be preset according to actual needs or triggered when there are significant changes in data distribution or a decline in prediction performance.

[0029] The anti-freezing strategy for the sector area is obtained based on the wall temperature safety margin, or based on the icing probability of the predicted time window, or based on the wall temperature safety margin and the icing probability of the predicted time window; optionally, when the wall temperature safety margin is not greater than the first preset safety margin and / or the icing probability of the predicted time window is greater than the preset risk threshold, the edge server generates the corresponding anti-freezing strategy. The antifreeze structure is controlled according to the antifreeze strategy, specifically: The control parameter value set is obtained based on the antifreeze strategy. The control parameter value set represents the set of values ​​of the control parameters. The anti-freezing structure is controlled based on a set of control parameter values. Optionally, the anti-freezing structure is controlled using a hardware interrupt triggering method. Upon receiving a high-risk signal, the traditional DCS periodic polling scan is replaced by a hardware interrupt triggering method. After the edge server infers the high risk, it sends an interrupt signal directly to the underlying controller via a dedicated high-speed bus, compressing the control command response time from seconds to milliseconds. By utilizing the underlying hardware interrupt and high-frequency communication bus, the anti-freezing structure is controlled within a preset, extremely short control cycle. Optionally, the controller includes a high-level application controller, a low-level process controller, and actuators. The high-level application controller meets low-latency communication requirements. It receives control strategies, parses and optimizes these strategies to generate high-level control commands. The low-level process controller receives and parses these high-level control commands to convert them into executable control parameter values. The actuators control the anti-freezing structure to perform corresponding anti-freezing intervention actions based on the control parameter values.

[0030] Beneficially, the cloud server handles the computationally intensive self-attention mechanism calculations and weight iteration updates. The operation of building the prediction model based on deep neural networks is configured within the cloud server to absorb massive historical time-series features and form a global non-linear understanding. Meanwhile, the edge server only receives the lightweight prediction model weight information. The edge server's task is extremely compressed into pure forward propagation matrix operations, ensuring that complex deep learning inference is compressed to a preset, extremely low page computation time. This decouples the contradiction between "intensive learning" and "ultra-fast inference," ensuring the accuracy and real-time performance of anti-freeze protection. Simultaneously, through the cloud server's heavy computational training, the edge server's lightweight inference, and the high-frequency execution at the underlying edge, extremely high standards of real-time computing and communication latency constraints are met.

[0031] In some embodiments, the antifreeze structure includes a cooling fan, at least one louver, and at least one bypass circulation pipe. The at least one louver and the at least one bypass circulation pipe correspond one-to-one with at least one cooling triangle. The louver is arranged around the corresponding cooling triangle. When the louver opening is within a preset antifreeze opening range, it can block the intrusion of local extremely cold air to reduce the impact of ambient wind speed on the wall temperature of the corresponding cooling triangle. The cooling fan is used to increase the airflow velocity within the fan-shaped area. The bypass circulation pipe is arranged between the inlet and outlet pipes of the corresponding cooling triangle. When the bypass circulation pipe is connected, it is used to introduce high-temperature circulating water. Optionally, the fan-shaped area includes at least two cooling triangles arranged in a ring, or the fan-shaped area includes multiple cooling triangles arranged in a ring. When the number of cooling triangles is at least two or more, the corresponding at least two or more louvers are arranged in a ring and connected end to end around the at least two or more cooling triangles.

[0032] The antifreeze structure is controlled according to the antifreeze strategy, including: controlling the opening degree of the louvers corresponding to the cooling triangle, and / or controlling the speed of the cooling fan, and / or controlling the on / off of the bypass circulation pipeline.

[0033] In some embodiments, the antifreeze strategy for the fan-shaped region includes a partitioned antifreeze strategy for each cooling triangle; The zoned anti-freeze strategy includes a normal monitoring strategy, a primary anti-freeze strategy, a secondary anti-freeze strategy, and a tertiary anti-freeze strategy. The normal monitoring strategy is characterized by the louver opening corresponding to the cooling triangle being within the normal operating range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating speed range. The primary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within the preset anti-freeze range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating speed range. That is, when the primary anti-freeze strategy is implemented at a lower anti-freeze level, it ensures the local safety of the corresponding cooling triangle without affecting the overall unit back pressure. The secondary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within the preset anti-freeze range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating speed range. In other words, when the primary anti-freeze strategy is implemented at a lower anti-freeze level, it ensures the local safety of the corresponding cooling triangle without affecting the overall unit back pressure. The first-level anti-freeze strategy, where the louver opening corresponding to the cooling triangle is within the preset anti-freeze range, the bypass circulation pipe is disconnected, and the cooling fan speed is within the preset anti-freeze speed range, offers a higher level of anti-freeze protection compared to the first-level and second-level strategies. It further adjusts the cooling fan speed in addition to the first-level anti-freeze strategy. The third-level anti-freeze strategy, where the louver opening corresponding to the cooling triangle is within the preset anti-freeze range, the bypass circulation pipe is connected, and the cooling fan speed is within the preset anti-freeze speed range, offers the highest level of anti-freeze protection. It only activates the bypass circulation pipe of the corresponding cooling triangle under extreme conditions approaching freezing. This three-level flexible anti-freeze strategy avoids the heat exchange efficiency losses caused by traditional sector-level regulation, improving the safety and economy of coal-fired unit operation.

[0034] Each zone's anti-freeze strategy includes a normal monitoring strategy, a primary anti-freeze strategy, a secondary anti-freeze strategy, and a tertiary anti-freeze strategy. The normal monitoring strategy indicates that the louver opening of the corresponding zone's anti-freeze structure is within the normal operating range, the fan speed is within the normal operating speed range, and the bypass circulation pipe is disconnected. The primary anti-freeze strategy indicates that the louver opening of the corresponding zone's anti-freeze structure is within the preset anti-freeze range, the fan speed is within the normal operating speed range, and the bypass circulation pipe is disconnected. In other words, when the primary anti-freeze strategy is executed at a lower anti-freeze level, it ensures the corresponding cooling triangle... The first-level anti-freezing strategy ensures local safety without affecting the overall unit back pressure. The second-level anti-freezing strategy indicates that the louver opening of the corresponding zone's anti-freezing structure is within the preset anti-freezing opening range, the fan speed is within the preset anti-freezing speed range, and the bypass circulation pipeline is disconnected. Compared with the first-level anti-freezing strategy, the second-level anti-freezing strategy has a higher level of anti-freezing protection. In addition to the first-level anti-freezing strategy, the fan speed is also adjusted. The third-level anti-freezing strategy indicates that the louver opening of the corresponding zone's anti-freezing structure is within the preset anti-freezing opening range, the fan speed is within the preset anti-freezing speed range, and the bypass circulation pipeline is connected.

[0035] In some embodiments, the antifreeze state of each cooling triangle includes a normal antifreeze state, a first antifreeze state, a second antifreeze state, and a third antifreeze state. The normal antifreeze state represents the antifreeze state implementing a normal monitoring strategy. Under the normal antifreeze state, the wall temperature safety margin is greater than a first preset safety margin, and the probability of icing in the predicted time window is less than a first risk threshold. , The first preset safety margin is defined as follows: the first antifreeze state represents the antifreeze state implementing the first-level antifreeze strategy; the second antifreeze state represents the antifreeze state implementing the second-level antifreeze strategy; and the third antifreeze state represents the antifreeze state implementing the third-level antifreeze strategy. See Figure 2 Based on the wall temperature safety margin, the antifreeze strategy for the sector area is obtained, including: Under normal antifreeze conditions, determine whether the wall temperature safety margin of the cooling triangle is not greater than the first preset safety margin; When the wall temperature safety margin is greater than the first preset safety margin, i.e. When this occurs, it indicates that the current indirect air-cooling system is in a safe range, and the zone antifreeze strategy can maintain the status quo, that is, the zone antifreeze strategy of the corresponding cooling triangle is a normal monitoring strategy. When the wall temperature safety margin is not greater than the first preset safety margin, i.e. When the corresponding cooling triangle's zoned anti-freezing strategy is determined to be a Level 1 anti-freezing strategy, a sudden gust of wind may cause a sudden drop in the wall temperature safety margin of the corresponding cooling triangle, triggering the corresponding anti-freezing intervention. The opening of the louvers corresponding to the cooling triangle is adjusted to the preset anti-freezing opening range. The bottom-level process controller bypasses the traditional large closed-loop polling and independently adjusts the corresponding louver opening for the target cooling triangle with high freezing risk to block the intrusion of localized extremely cold winds. This anti-freezing intervention by adjusting the louver opening is a millimeter-level precision intervention. Because it only limits the air intake of a single high-freezing-risk cooling triangle, other cooling triangles in the same sector area that are in a safe state still maintain efficient heat exchange, thereby ensuring that the overall vacuum level of the coal-fired unit is not affected to the greatest extent.

[0036] In the first antifreeze state, at the first evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time. The first evaluation time represents the time after the first preset time has elapsed since the first initial time. The first initial time represents the time when the antifreeze structure corresponding to the cooling triangle executes the first-level antifreeze strategy. If the wall temperature safety margin of the cooling triangle does not continue to decrease within the first preset time period, it means that the freezing risk of the cooling triangle has been successfully resolved after adjusting the opening of the corresponding louvers. When the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time period, the corresponding cooling triangle's zone anti-freeze strategy is determined to be a level two anti-freeze strategy. This indicates that the current cold wave or gust intensity has exceeded the physical defense limit of the local louvers. Based on the level one anti-freeze strategy, the speed of the cooling fan corresponding to the entire fan-shaped area where the cooling triangle is located is adjusted to the preset anti-freeze speed range, i.e., switching to the preset low-frequency anti-freeze operation mode. By reducing the convective heat transfer intensity of the fan-shaped area corresponding to the cooling triangle at the global level, the circulating water temperature in the entire fan-shaped area is rapidly increased, providing heat support for the vulnerable windward tube bundle.

[0037] In the second antifreeze state, at the second evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time. The second evaluation time represents the time after the second preset time since the second initial time, and the second initial time represents the time when the antifreeze structure corresponding to the cooling triangle executes the secondary antifreeze strategy. Optionally, the second preset time is less than the first preset time. In the second antifreeze state, the freezing risk of the corresponding cooling triangle is further increased, and the monitoring frequency is increased to reduce the freezing risk.

[0038] When the wall temperature safety margin of the cooling triangle does not continue to decrease within the second preset time period, it indicates that the freezing risk of the cooling triangle has been successfully resolved by adjusting the speed of the cooling fan in the entire fan-shaped area where the cooling triangle is located after this local fine adjustment. When the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time period, the corresponding cooling triangle's zone antifreeze strategy is determined to be a level 3 antifreeze strategy. This indicates that the cooling triangle is facing an abnormal operating condition such as a rare extreme cold storm or blockage in the pipes within the cooling triangle. In addition to implementing the level 2 antifreeze strategy, the bypass circulation pipe is further connected, allowing the high-temperature circulating water that has not been sufficiently cooled to directly mix into the cooling triangle. The powerful water-side heat source is used for forced "bottom heating," completely eliminating the possibility of physical freezing and preventing pipe burst accidents. Optionally, an electric bypass valve is installed on the bypass circulation pipe for convenient control.

[0039] The above process, from routine monitoring to Level 1 antifreeze strategy, then to Level 2 antifreeze strategy, and finally to Level 3 antifreeze strategy, constitutes the freezing defense process. After the freezing risk is eliminated, i.e., when environmental conditions improve such as reduced wind or rising temperatures, the antifreeze strategy for the fan-shaped area can be withdrawn in an economically oriented manner. Specifically: See Figure 3 In some embodiments, obtaining the antifreeze strategy for the fan-shaped region based on the wall temperature safety margin further includes: in the third antifreeze state, determining whether the wall temperature safety margin of the corresponding cooling triangle is greater than a second preset safety margin. , ; When the wall temperature safety margin of the corresponding cooling triangle is not greater than the second preset safety margin, it means that the freezing risk has not been eliminated and the zonal antifreeze strategy can maintain the status quo, that is, the zonal antifreeze strategy of the corresponding cooling triangle is a level three antifreeze strategy. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, the first duration is determined. Is it greater than the third preset duration? The first duration characterizes the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the third antifreeze state. When the first duration is not greater than the third preset duration, i.e. If the freezing risk has not been eliminated, the zoned antifreeze strategy can maintain the status quo, that is, the zoned antifreeze strategy of the corresponding cooling triangle is a level three antifreeze strategy. When the first duration is longer than the third preset duration, i.e. When the corresponding cooling triangle is determined to be a secondary antifreeze strategy, the bypass circulation pipeline corresponding to that cooling triangle is disconnected.

[0040] In some embodiments, under the second antifreeze state, when the wall temperature safety margin of the cooling triangle does not continue to decrease within the second preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin. When the wall temperature safety margin of the corresponding cooling triangle is not greater than the second preset safety margin, it means that the freezing risk has not been eliminated and the zonal antifreeze strategy can maintain the status quo, that is, the zonal antifreeze strategy of the corresponding cooling triangle is a secondary antifreeze strategy. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, the second duration is determined. Is it greater than the fourth preset duration? The second duration characterizes the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the second antifreeze condition. When the second duration is not greater than the fourth preset duration, that is If the freezing risk has not been eliminated, the zoned antifreeze strategy can maintain the status quo, that is, the zoned antifreeze strategy of the corresponding cooling triangle is a secondary antifreeze strategy. When the second duration is greater than the fourth preset duration, that is If the antifreeze strategy of the corresponding cooling triangle is determined to be a level one antifreeze strategy, then the speed of the cooling fan in the corresponding fan-shaped area of ​​the cooling triangle is adjusted to the normal operating speed range.

[0041] In some embodiments, under the first antifreeze state, when the wall temperature safety margin of the cooling triangle does not continuously decrease within a first preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than a second preset safety margin. ; When the wall temperature safety margin of the corresponding cooling triangle is not greater than the second preset safety margin, it means that the freezing risk has not been eliminated and the zone antifreeze strategy can maintain the status quo, that is, the zone antifreeze strategy of the corresponding cooling triangle is a first-level antifreeze strategy. When the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, it is determined whether the third duration is greater than the fifth preset duration. The third duration represents the duration during which the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin under the first antifreeze state. When the duration of the third period is not greater than the preset duration of the fifth period, that is If the freezing risk has not been eliminated, the zoned antifreeze strategy can maintain the status quo, that is, the zoned antifreeze strategy of the corresponding cooling triangle is judged to be the first-level antifreeze strategy. When the duration of the third event is greater than the preset duration of the fifth event, that is... When the corresponding cooling triangle's zone antifreeze strategy is determined to be a normal monitoring strategy, the opening degree of the louvers corresponding to the cooling triangle is adjusted to the normal operating opening degree range.

[0042] In addition to assessing the freezing risk of the cooling triangle through the wall temperature safety margin, the probability of icing can also be used. Or by the probability of freezing and wall temperature safety margin Assessing the freezing risk of the cooling triangle. In some embodiments, the anti-freezing strategy for the sector region is obtained based on the wall temperature safety margin, and further includes: See Figure 4 Under normal antifreeze conditions, determine whether the probability of icing in the cooling triangle exceeds the first risk threshold. Alternatively, it can determine whether the probability of icing in the cooling triangle is greater than the first risk threshold and whether the wall temperature safety margin is not greater than the first preset safety margin. When the probability of icing is greater than the first risk threshold, i.e. When, or when the probability of icing is greater than the first risk threshold and the wall temperature safety margin is not greater than the first preset safety margin, and When the corresponding cooling triangle's zone antifreeze strategy is determined to be a Level 1 antifreeze strategy, optionally, when the probability of icing is greater than the first risk threshold, i.e. At that time, the indirect air-cooling system will be controlled to perform an icing alarm under normal antifreeze conditions.

[0043] Under the first antifreeze condition, at the first assessment time, it is determined whether the probability of icing of the cooling triangle within a first preset time period is greater than the second risk threshold. Or, at the first assessment moment, determine whether the probability of icing of the cooling triangle within the first preset time period is greater than the second risk threshold and whether the wall temperature safety margin continues to decrease. When the probability of the cooling triangle freezing within the first preset time period is greater than the second risk threshold, that is... When, or when the probability of icing in the cooling triangle within the first preset time period is greater than the second risk threshold and the wall temperature safety margin continues to decrease, the corresponding cooling triangle's zone antifreeze strategy is determined to be a secondary antifreeze strategy; optionally, when the probability of icing in the cooling triangle within the first preset time period is greater than the second risk threshold, i.e. When the indirect air-cooling system is activated, an icing alarm will be triggered under the first antifreeze condition.

[0044] Under the second antifreeze condition, at the second assessment time, it is determined whether the probability of icing of the cooling triangle within the second preset time period is greater than the third risk threshold. Or, at the second assessment time, determine whether the probability of icing of the cooling triangle within the second preset time period is greater than the third risk threshold and whether the wall temperature safety margin continues to decrease. When the probability of the cooling triangle freezing within the second preset time period is greater than the third risk threshold, that is... When, or when the probability of icing in the cooling triangle within the second preset time period is greater than the third risk threshold and the wall temperature safety margin continues to decrease, the corresponding cooling triangle's zone antifreeze strategy is determined to be a level three antifreeze strategy; optionally, when the probability of icing in the cooling triangle within the second preset time period is greater than the third risk threshold, i.e. When the indirect air-cooling system is activated, an icing alarm will be triggered under the first antifreeze condition.

[0045] The first risk threshold is less than the second risk threshold, and the second risk threshold is less than the third risk threshold.

[0046] In some embodiments, the antifreeze strategy for the sector region, based on the wall temperature safety margin, further includes: Under normal antifreeze conditions, determine whether the wall temperature safety margin is not greater than the first preset safety margin; When the wall temperature safety margin is not greater than the first preset safety margin, i.e. When the corresponding cooling triangle's zone antifreeze strategy is determined to be a level one antifreeze strategy; Under the first antifreeze condition, at the first assessment time, it is determined whether the probability of icing of the cooling triangle within a first preset time period is greater than the second risk threshold. Using specific numerical values ​​for judgment facilitates control. When the probability of the cooling triangle freezing within the first preset time period is greater than the second risk threshold, that is... When the corresponding cooling triangle's zone antifreeze strategy is determined to be a secondary antifreeze strategy; When the probability of the cooling triangle freezing within the first preset time period is not greater than the second risk threshold, that is... When the corresponding cooling triangle's zone antifreeze strategy is determined to be a level one antifreeze strategy; Under the second antifreeze condition, at the second assessment time, it is determined whether the probability of icing of the cooling triangle within the second preset time period is greater than the third risk threshold. Using specific numerical values ​​for judgment facilitates control. When the probability of the cooling triangle freezing within the second preset time period is greater than the third risk threshold, that is... When the corresponding cooling triangle is selected, the antifreeze strategy for the partition is determined to be a level three antifreeze strategy. When the probability of icing in the cooling triangle within the second preset time period is not greater than the third risk threshold, that is... When the corresponding cooling triangle is selected, the antifreeze strategy for the partition is determined to be a secondary antifreeze strategy.

[0047] See Figure 5 Secondly, this application provides a winter anti-freezing control device for an indirect air-cooled system of a coal-fired unit, which is used to implement a winter anti-freezing control method for the indirect air-cooled system of a coal-fired unit. The control device includes a data acquisition unit 2100, an edge server 2200, a cloud server 2300, a controller 2400, and an anti-freezing structure 1112. The anti-freezing structure 1112 is disposed in the fan-shaped area 1110 of the indirect air-cooled system 1100. The data acquisition unit 2100 is used to collect real-time data, obtain the physical freezing critical temperature of circulating water and the cumulative duration of the continuous effect of extreme environmental wind speed on the cooling triangle, and transmit the real-time data to the edge server 2200; the real-time data includes the operating data of the coal-fired unit 1000 and the control parameters of the antifreeze structure 1112. Edge server 2200 is used to obtain a time-series feature matrix based on real-time data, transmit the time-series feature matrix to cloud server 2300, obtain wall temperature prediction data of the sector area based on the time-series feature matrix and prediction model weight information, obtain anti-freezing safety buffer compensation term based on ambient wind speed and cumulative duration, obtain the wall temperature safety margin of the corresponding cooling triangle 1111 based on the wall temperature prediction data, physical freezing critical temperature and anti-freezing safety buffer compensation term, and obtain the anti-freezing strategy of sector area 1110 based on the wall temperature safety margin. The cloud server 2300 is used to obtain the prediction model weight information based on the time series feature matrix and the deep neural network, and to transmit the prediction model weight information to the edge server 2200. The controller 2400 is used to control the antifreeze structure 1112 according to the antifreeze strategy.

[0048] In some embodiments, controller 2400 includes a high-level application controller, a low-level process controller, and an actuator; The advanced application controller is used to receive control policies, parse and optimize the control policies to generate high-level control instructions; The lower-level process controller is used to receive control commands from higher levels and parse them into control commands; the control commands are the values ​​of executable control parameters. The actuator is used to control the antifreeze structure 1112 according to the control command, i.e., the value of the control parameter.

[0049] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus (systems), or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0050] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce an instruction that executes via the processor of the computer or other programmable data processing apparatus to create an instruction for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0051] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0052] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0053] Finally, it should be noted that the above content is only used to illustrate the technical solution of this application, and is not intended to limit the scope of protection of this application. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of this application shall not depart from the substance and scope of the technical solution of this application.

Claims

1. A method for winter antifreeze control of an indirect air-cooled system for a coal-fired power unit, wherein the indirect air-cooled system includes a fan-shaped region, and the fan-shaped region includes at least one cooling triangle, characterized in that, The fan-shaped area is equipped with an anti-freezing structure, and the control method includes: collecting real-time data; the real-time data includes the operating parameters of the coal-fired unit and the control parameters of the anti-freezing structure; the operating parameters include ambient wind speed; and obtaining a time-series feature matrix based on the real-time data. The edge server obtains the predicted wall temperature data for each cooling triangle based on the time-series feature matrix and the prediction model weight information; the prediction model weight information is obtained by the cloud server based on the time-series feature matrix and the deep neural network and transmitted to the edge server, and the time-series feature matrix is ​​transmitted from the edge server to the cloud server. The cumulative duration of the effect of the physical freezing critical temperature of circulating water and extreme ambient wind speed on the cooling triangle was obtained. The antifreeze safety buffer compensation term is obtained based on the ambient wind speed and the cumulative duration; the wall temperature safety margin of the corresponding cooling triangle is obtained based on the predicted wall temperature data, the physical freezing critical temperature, and the antifreeze safety buffer compensation term. The antifreeze strategy for the sector area is obtained based on the wall temperature safety margin. The antifreeze structure is controlled according to the antifreeze strategy.

2. The method for winter antifreeze control of indirect air-cooled system of coal-fired unit according to claim 1, characterized in that, The antifreeze structure includes a cooling fan, at least one louver, and at least one bypass circulation pipe. The at least one louver and the at least one bypass circulation pipe correspond one-to-one with the at least one cooling triangle. The louver is arranged around the corresponding cooling triangle. The cooling fan is used to increase the airflow rate in the fan-shaped area. The bypass circulation pipe is arranged between the inlet pipe and the return pipe of the corresponding cooling triangle. Controlling the antifreeze structure according to the antifreeze strategy includes: controlling the opening degree of the louvers corresponding to the cooling triangle, and / or controlling the cooling fan speed of the cooling fan, and / or controlling the on / off of the bypass circulation pipeline.

3. The method for winter antifreeze control of the indirect air-cooled system of a coal-fired unit according to claim 2, characterized in that, The antifreeze strategy for the sector area includes a zoned antifreeze strategy for each of the cooling triangles; The zoned anti-freeze strategy includes a normal monitoring strategy, a primary anti-freeze strategy, a secondary anti-freeze strategy, and a tertiary anti-freeze strategy. The normal monitoring strategy is characterized by the louver opening corresponding to the cooling triangle being within the normal operating range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating range. The primary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within a preset anti-freeze opening range, the bypass circulation pipe being disconnected, and the cooling fan speed being within the normal operating range. The secondary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within a preset anti-freeze opening range, the bypass circulation pipe being disconnected, and the cooling fan speed being within a preset anti-freeze speed range. The tertiary anti-freeze strategy is characterized by the louver opening corresponding to the cooling triangle being within a preset anti-freeze opening range, the bypass circulation pipe being connected, and the cooling fan speed being within a preset anti-freeze speed range.

4. The method for winter antifreeze control of the indirect air-cooled system of a coal-fired unit according to claim 3, characterized in that, The antifreeze state of each cooling triangle includes a normal antifreeze state, a first antifreeze state, a second antifreeze state, and a third antifreeze state. The normal antifreeze state represents the antifreeze state of executing the normal monitoring strategy. The first antifreeze state represents the antifreeze state of executing the first-level antifreeze strategy. The second antifreeze state represents the antifreeze state of executing the second-level antifreeze strategy. The third antifreeze state represents the antifreeze state of executing the third-level antifreeze strategy. The antifreeze strategy for the sector region is obtained based on the wall temperature safety margin, including: Under the normal antifreeze condition, determine whether the wall temperature safety margin of the cooling triangle is not greater than the first preset safety margin; When the wall temperature safety margin is not greater than the first preset safety margin, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the first-level antifreeze strategy. In the first antifreeze state, at the first evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time. The first evaluation time represents the time after the first preset time since the first initial time. The first initial time represents the time when the antifreeze structure corresponding to the cooling triangle is controlled to execute the first-level antifreeze strategy. When the wall temperature safety margin of the cooling triangle continues to decrease within the first preset time period, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the secondary antifreeze strategy. In the second antifreeze state, at the second evaluation time, it is determined whether the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time. The second evaluation time represents the time after the second preset time has elapsed since the second initial time. The second initial time represents the time when the antifreeze structure corresponding to the cooling triangle is controlled to execute the secondary antifreeze strategy. When the wall temperature safety margin of the cooling triangle continues to decrease within the second preset time period, the partitioned antifreeze strategy corresponding to the cooling triangle is determined to be the three-level antifreeze strategy.

5. The method for winter antifreeze control of the indirect air-cooled system of a coal-fired unit according to claim 4, characterized in that, The method for obtaining the antifreeze strategy for the fan-shaped region based on the wall temperature safety margin also includes: in the third antifreeze state, determining whether the wall temperature safety margin corresponding to the cooling triangle is greater than a second preset safety margin, wherein the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin, it is determined whether the first duration is greater than the third preset duration. The first duration represents the duration during which the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin under the third antifreeze state. When the first duration is longer than the third preset duration, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the secondary antifreeze strategy.

6. The method for winter antifreeze control of an indirect air-cooled system for a coal-fired unit according to claim 4 or 5, characterized in that, In the second antifreeze state, when the wall temperature safety margin of the cooling triangle does not continuously decrease within the second preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, and the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin, it is determined whether the second duration is greater than the fourth preset duration. The second duration represents the duration during which the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin under the second antifreeze state. When the second duration is longer than the fourth preset duration, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the first-level antifreeze strategy.

7. The method for winter antifreeze control of the indirect air-cooled system of a coal-fired unit according to claim 4, characterized in that, In the first antifreeze state, when the wall temperature safety margin of the cooling triangle does not continuously decrease within the first preset time period, it is determined whether the wall temperature safety margin of the corresponding cooling triangle is greater than the second preset safety margin, and the second preset safety margin is greater than the first preset safety margin. When the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin, it is determined whether the third duration is greater than the fifth preset duration. The third duration represents the duration during which the wall temperature safety margin corresponding to the cooling triangle is greater than the second preset safety margin under the first antifreeze state. When the third duration is longer than the fifth preset duration, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the normal monitoring strategy.

8. The method for winter antifreeze control of the indirect air-cooled system of a coal-fired unit according to claim 4, characterized in that, The edge server also obtains the freezing probability of each cooling triangle in the prediction time window based on the time-series feature matrix and the prediction model weight information; The antifreeze strategy for the sector region, obtained based on the wall temperature safety margin, further includes: Under the normal antifreeze condition, it is determined whether the icing probability of the cooling triangle is greater than the first risk threshold. When the freezing probability is greater than the first risk threshold, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the first-level antifreeze strategy. Under the first antifreeze state, at the first evaluation time, it is determined whether the icing probability of the cooling triangle within the first preset time period is greater than the second risk threshold. When the icing probability of the cooling triangle within the first preset time period is greater than the second risk threshold, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the secondary antifreeze strategy. Under the second antifreeze state, at the second evaluation time, it is determined whether the icing probability of the cooling triangle within the second preset time period is greater than the third risk threshold. When the icing probability of the cooling triangle within the second preset time period is greater than the third risk threshold, the partition antifreeze strategy corresponding to the cooling triangle is determined to be the level three antifreeze strategy. The first risk threshold is less than the second risk threshold, and the second risk threshold is less than the third risk threshold.

9. The method for winter antifreeze control of an indirect air-cooled system for a coal-fired unit according to claim 1, characterized in that, The operating parameters of the coal-fired unit also include ambient air temperature, mass flow rate of circulating water, inlet temperature of cooling triangular circulating water, and outlet temperature of cooling triangular circulating water.

10. A winter antifreeze control device for an indirect air-cooled system of a coal-fired unit, characterized in that, The control device is used to implement the winter anti-freezing control method for the indirect air-cooled system of a coal-fired unit as described in any one of claims 1-9. The control device includes a data acquisition unit, an edge server, a cloud server, a controller, and an anti-freezing structure, wherein the anti-freezing structure is configured in a fan-shaped area of ​​the indirect air-cooled system. The data acquisition device is used to collect real-time data, obtain the physical freezing critical temperature of circulating water and the cumulative duration of the continuous effect of extreme environmental wind speed on the cooling triangle, and transmit the real-time data to the edge server. The edge server is used to obtain a time-series feature matrix based on the real-time data, transmit the time-series feature matrix to the cloud server, obtain wall temperature prediction data for each cooling triangle based on the time-series feature matrix and prediction model weight information, obtain an anti-freeze safety buffer compensation term based on the ambient wind speed and the cumulative duration, obtain the corresponding wall temperature safety margin for the cooling triangle based on the wall temperature prediction data, the physical freezing critical temperature and the anti-freeze safety buffer compensation term, and obtain the anti-freeze strategy for the fan-shaped area based on the wall temperature safety margin. The cloud server is used to obtain the prediction model weight information based on the temporal feature matrix and the deep neural network, and to transmit the prediction model weight information to the edge server; The controller is used to control the antifreeze structure according to the antifreeze strategy.

Images