Steam turbine deep peak regulation main steam temperature control optimization method and device

Abstract

The invention provides a steam turbine deep peak regulation main steam temperature control optimization method and device, and relates to the technical field of energy power, and the method comprises the steps: carrying out the parametric modeling of a main steam temperature curve of a steam turbine deep peak regulation process according to an obtained temperature parameter, and generating a main steam temperature change curve; obtaining a sample based on the main steam temperature change curve, and carrying out finite element numerical simulation on the sample to obtain an optimized data set; performing model training by utilizing the optimized data set to obtain an agent model; and optimizing a main steam temperature control strategy based on the agent model and a rapid simulated annealing algorithm. According to the steam turbine deep peak regulation main steam temperature control optimization method and device provided by the invention, the optimal control strategy of the main steam temperature is quickly obtained on the premise of meeting the safety constraint of a steam turbine rotor, so that the response speed and the operation safety of the deep peak regulation process are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of energy and power technology, and in particular to a method and apparatus for optimizing the main steam temperature control during deep peak shaving of a steam turbine. Background Technology

[0002] In the energy and power sector, steam turbines play a crucial role in power system peak shaving. With the increasing proportion of renewable energy generation, the power grid's reliance on peak shaving capacity is growing stronger, leading to a significant increase in the frequency and intensity of deep peak shaving tasks performed by thermal power units. Deep peak shaving by steam turbines is of great significance for the power grid to absorb renewable energy and ensure the stable operation of the power system. However, during deep peak shaving, rapid fluctuations in main steam temperature can cause a sharp increase in thermal stress and fatigue losses in the turbine rotor, becoming a key factor restricting the safety and response speed of the peak shaving process.

[0003] Under deep peak shaving conditions, the turbine load changes frequently and significantly, and the unit's operating status exhibits rapid dynamic changes, which poses considerable challenges to the precise control of the main steam temperature. To achieve rapid and safe operation of the unit under deep peak shaving conditions, it is necessary to develop a better operation control strategy for the main steam temperature control process, balancing peak shaving speed and rotor structural safety.

[0004] Existing technologies for optimizing main steam temperature control primarily focus on optimizing the temperature rise curve during rotor startup, which is insufficient to meet the rapid load response requirements under deep peak shaving conditions. Furthermore, existing optimization methods often rely on numerical simulations or traditional shallow learning surrogate models combined with optimization algorithms, resulting in slow computation speed, low prediction accuracy, and low optimization efficiency. These shortcomings limit the rapid acquisition of the optimal control strategy during peak shaving, hindering the full realization of the flexibility and economy of deep peak shaving.

[0005] Therefore, there is an urgent need for a rapid optimization method for main steam temperature control that can balance safety and response speed. Summary of the Invention

[0006] To address the problems in the prior art, this application provides a method and apparatus for optimizing the main steam temperature control during deep peak shaving of steam turbines, which can solve the problems existing in the prior art.

[0007] In a first aspect, this application provides a method for optimizing the control of main steam temperature during deep peak shaving in a steam turbine, including:

[0008] Based on the obtained temperature parameters, the main steam temperature curve of the deep peak shaving process of the steam turbine is parametrically modeled to generate the main steam temperature change curve;

[0009] Based on the main steam temperature change curve, samples were obtained and finite element numerical simulations were performed on the samples to obtain an optimized dataset;

[0010] The agent model is obtained by training the model using the optimized dataset.

[0011] The main steam temperature control strategy is optimized based on the aforementioned proxy model and fast simulated annealing algorithm.

[0012] Furthermore, the temperature parameters include the initial temperature, the final temperature, the rate of temperature change for each segment, and the corresponding operating time; the step of parametrically modeling the main steam temperature curve of the deep peak shaving process of the steam turbine based on the acquired temperature parameters to generate the main steam temperature change curve includes:

[0013] A main steam temperature change model is constructed based on the initial temperature, the final temperature, the temperature change rate of each segment, and the corresponding running time.

[0014] The main steam temperature change curve is generated using the main steam temperature change model.

[0015] Furthermore, the step of obtaining an optimized dataset by acquiring samples based on the main steam temperature change curve and performing finite element numerical simulations on the samples includes:

[0016] A sampling space is constructed based on the main steam temperature change curve;

[0017] The sample is obtained by sampling using the Latin hypercube sampling method within the sampling space;

[0018] The sample is input into a pre-built finite element model for thermo-mechanical coupling numerical simulation to calculate the corresponding rotor temperature field and maximum equivalent stress.

[0019] The optimized dataset is constructed based on the main steam temperature variation curve, the rotor temperature field, and the maximum equivalent stress.

[0020] Furthermore, the step of training the model using the optimized dataset to obtain the surrogate model includes:

[0021] A first deep neural network model for predicting the temperature field of a steam turbine rotor and a second deep neural network model for predicting the maximum equivalent stress of the rotor are constructed.

[0022] The surrogate model is obtained by iteratively training the first deep neural network model and the second deep neural network model using the optimized dataset.

[0023] Furthermore, the optimization of the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm includes:

[0024] Using the temperature parameter as the design variable, the proxy model is invoked to predict the maximum equivalent stress of the turbine rotor under different temperature control strategies;

[0025] Under the constraint that the maximum equivalent stress does not exceed the preset allowable material stress, with the shortest main steam temperature adjustment time as the optimization objective, the optimal combination of temperature parameters is obtained by using a fast simulated annealing algorithm.

[0026] Secondly, this application provides a steam turbine deep peak-shaving main steam temperature control optimization device, comprising:

[0027] The curve generation unit is used to parametrically model the main steam temperature curve of the deep peak shaving process of the steam turbine based on the acquired temperature parameters, and generate the main steam temperature change curve.

[0028] The dataset generation unit is used to obtain samples based on the main steam temperature change curve and perform finite element numerical simulation on the samples to obtain an optimized dataset;

[0029] The surrogate model generation unit is used to train the model using the optimized dataset to obtain the surrogate model.

[0030] The control optimization unit is used to optimize the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm.

[0031] Furthermore, the temperature parameters include the initial temperature, the final temperature, the rate of temperature change for each segment, and the corresponding running time; the curve generation unit includes:

[0032] The temperature change model construction module is used to construct a main steam temperature change model based on the initial temperature, the final temperature, the temperature change rate of each segment, and the corresponding running time.

[0033] The curve generation module is used to generate the main steam temperature change curve using the main steam temperature change model.

[0034] Furthermore, the dataset generation unit includes:

[0035] A sampling space construction module is used to construct a sampling space based on the main steam temperature change curve.

[0036] The sample acquisition module is used to collect samples within the sampling space using a Latin hypercube sampling device to obtain the samples.

[0037] The data calculation module is used to input the sample into a pre-built finite element model for thermo-mechanical coupling numerical simulation, and calculate the corresponding rotor temperature field and maximum equivalent stress.

[0038] The dataset construction module is used to construct the optimized dataset based on the main steam temperature change curve, the rotor temperature field, and the maximum equivalent stress.

[0039] Furthermore, the proxy model generation unit includes:

[0040] The neural network model building module is used to build a first deep neural network model for predicting the temperature field of the turbine rotor and a second deep neural network model for predicting the maximum equivalent stress of the rotor.

[0041] The surrogate model generation module is used to iteratively train the first deep neural network model and the second deep neural network model using the optimized dataset to obtain the surrogate model.

[0042] Furthermore, the control optimization unit includes:

[0043] The model prediction module is used to use the temperature parameter as a design variable and call the proxy model to predict the maximum equivalent stress of the turbine rotor under different temperature control strategies.

[0044] The control optimization module is used to obtain the optimal combination of temperature parameters by using a fast simulated annealing algorithm, under the constraint that the maximum equivalent stress does not exceed the preset allowable stress of the material, with the shortest main steam temperature adjustment time as the optimization objective.

[0045] Thirdly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in any of the above embodiments.

[0046] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described in any of the above embodiments.

[0047] Fifthly, this application provides a computer program product, which includes a computer program that, when executed by a processor, implements the method described in any of the above embodiments.

[0048] This application provides a method and apparatus for optimizing the main steam temperature control during deep peak shaving of a steam turbine. The method involves parametrically modeling the main steam temperature curve during the deep peak shaving process based on acquired temperature parameters to generate a main steam temperature variation curve. Samples are obtained from the main steam temperature variation curve, and finite element numerical simulations are performed on these samples to obtain an optimized dataset. The optimized dataset is then used for model training to obtain a surrogate model. Finally, the main steam temperature control strategy is optimized based on the surrogate model and a fast simulated annealing algorithm. This method achieves rapid acquisition of the optimal main steam temperature control strategy while meeting the turbine rotor safety constraints, thereby significantly improving the response speed and operational safety of the deep peak shaving process.

[0049] Specifically, by parametrically modeling the main steam temperature curve during the deep peak shaving process of the steam turbine based on the acquired temperature parameters, a main steam temperature change curve is generated, achieving an accurate description of the main steam temperature change law during the deep peak shaving process of the steam turbine, and providing a mathematical model basis for subsequent data sampling and optimization; by obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples, an optimized dataset is obtained, providing high-precision data on the influence of main steam temperature change on rotor stress and temperature field, providing reliable samples for model training; by using the optimized dataset for model training, a surrogate model is obtained, establishing a high-precision model that can quickly predict rotor thermal stress response, significantly reducing computation time; by optimizing the main steam temperature control strategy based on the surrogate model and fast simulated annealing algorithm, rapid optimization of the main steam temperature control strategy is achieved under the premise of ensuring rotor safety constraints, shortening the deep peak shaving response time.

[0050] This application achieves rapid optimization of main steam temperature control during deep peak shaving of steam turbines by combining deep learning with a fast simulated annealing algorithm. The surrogate model constructed using a deep neural network can perform rapid and accurate end-to-end prediction of the rotor's thermodynamic state, providing high-precision stress and temperature field assessments for the optimization process. Combined with the efficient fast simulated annealing algorithm for control strategy optimization, the optimal main steam temperature regulation scheme can be quickly obtained under the condition of satisfying the rotor's allowable stress constraints. Compared with traditional methods relying on numerical simulation or shallow optimization, this application significantly improves the speed and accuracy of main steam temperature control strategy optimization, achieving a simultaneous improvement in the safety and response performance of the steam turbine under deep peak shaving conditions. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic flowchart of a method for optimizing the main steam temperature control in deep peak shaving of a steam turbine, provided in an embodiment of this application.

[0053] Figure 2 This is a schematic flowchart of a method for optimizing the main steam temperature control in deep peak shaving of a steam turbine, provided in an embodiment of this application.

[0054] Figure 3 This is a schematic flowchart of a method for optimizing the main steam temperature control in deep peak shaving of a steam turbine, provided in an embodiment of this application.

[0055] Figure 4 This is a schematic flowchart of a method for optimizing the main steam temperature control in deep peak shaving of a steam turbine, provided in an embodiment of this application.

[0056] Figure 5 This is a schematic flowchart of a method for optimizing the main steam temperature control in deep peak shaving of a steam turbine, provided in an embodiment of this application.

[0057] Figure 6 This is a schematic diagram of the structure of a steam turbine deep peak shaving main steam temperature control optimization device provided in an embodiment of this application;

[0058] Figure 7 This is a schematic diagram of the structure of a steam turbine deep peak shaving main steam temperature control optimization device provided in an embodiment of this application;

[0059] Figure 8 This is a schematic diagram of the structure of a steam turbine deep peak shaving main steam temperature control optimization device provided in an embodiment of this application;

[0060] Figure 9 This is a schematic diagram of the structure of a steam turbine deep peak shaving main steam temperature control optimization device provided in an embodiment of this application;

[0061] Figure 10 This is a schematic diagram of the structure of a steam turbine deep peak shaving main steam temperature control optimization device provided in an embodiment of this application;

[0062] Figure 11 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application;

[0063] Figure 12 This is a parameterized schematic diagram of the main steam temperature control curve for the deep peak shaving process of a steam turbine provided in an embodiment of this application;

[0064] Figure 13 This is a flowchart of the process for establishing the main steam temperature control optimization dataset for the deep peak shaving process of a steam turbine, provided in one embodiment of this application.

[0065] Figure 14 This is a schematic diagram of a deep neural network (DNN) proxy model provided in an embodiment of this application. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments and their descriptions are used to explain this application, but are not intended to limit this application. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other.

[0067] The following describes the specific implementation process of the turbine deep peak shaving main steam temperature control optimization method provided in this application embodiment, taking the server as the execution subject as an example.

[0068] Figure 1 This is a flowchart illustrating an embodiment of the turbine deep peak-shaving main steam temperature control optimization method provided in this application, as shown below. Figure 1 As shown, the turbine deep peak-shaving main steam temperature control optimization method provided in this application includes:

[0069] S101: Based on the acquired temperature parameters, perform parameterized modeling of the main steam temperature curve during the deep peak shaving process of the steam turbine to generate the main steam temperature change curve;

[0070] S102: Based on the main steam temperature change curve, obtain samples and perform finite element numerical simulation on the samples to obtain an optimized dataset;

[0071] S103: Use the optimized dataset to train the model and obtain the surrogate model;

[0072] S104: Optimize the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm.

[0073] from Figure 1 As shown in the flowchart, this application provides an optimization method for main steam temperature control during deep peak shaving of a steam turbine. It generates a main steam temperature variation curve by parametrically modeling the main steam temperature curve during the deep peak shaving process based on the acquired temperature parameters. Samples are obtained from the main steam temperature variation curve, and finite element numerical simulations are performed on these samples to obtain an optimized dataset. The optimized dataset is then used for model training to obtain a surrogate model. Finally, the main steam temperature control strategy is optimized based on the surrogate model and a fast simulated annealing algorithm. This method achieves rapid acquisition of the optimal main steam temperature control strategy while meeting the turbine rotor safety constraints, thereby significantly improving the response speed and operational safety of the deep peak shaving process.

[0074] Each step is explained in detail below.

[0075] S101: Based on the acquired temperature parameters, perform parameterized modeling of the main steam temperature curve during the deep peak shaving process of the steam turbine to generate the main steam temperature change curve;

[0076] Specifically, the server performs parametric modeling of the main steam temperature curve during the deep peak shaving process of the steam turbine based on the acquired temperature parameters, generating a main steam temperature variation curve. The temperature parameters include key physical quantities describing the characteristics of steam temperature changes during peak shaving, such as the initial and final temperatures of the main steam, the rate of temperature change at each stage, and the corresponding operating time. By establishing a parametric model of the main steam temperature curve, a unified mathematical description of the main steam temperature variation law during deep peak shaving can be achieved, thus providing basic input conditions for subsequent simulation calculations and optimization analysis.

[0077] Figure 2 This is a flowchart illustrating an embodiment of the turbine deep peak-shaving main steam temperature control optimization method provided in this application. The temperature parameters include initial temperature, final temperature, temperature change rate for each segment, and corresponding running time. Figure 2 As shown, S101 includes:

[0078] S201: Construct a main steam temperature change model based on the initial temperature, the final temperature, the temperature change rate of each segment, and the corresponding running time;

[0079] Specifically, temperature parameters are used to characterize the basic laws governing the temperature change of the main steam during deep peak shaving, including the initial temperature, final temperature, rate of temperature change for each segment, and corresponding operating time of the main steam.

[0080] The initial temperature is the temperature of the main steam in the turbine at the start of deep peak shaving, and the ending temperature is the final temperature of the main steam after the peak shaving process is completed. The rate of temperature change in each segment describes the speed at which the main steam temperature decreases or increases at different stages. The corresponding running time defines the duration of temperature change in each stage. By determining the above temperature parameters, the dynamic characteristics of the turbine's main steam temperature changing over time can be fully characterized.

[0081] Based on the aforementioned temperature parameters, the server performs parametric modeling of the main steam temperature curve during the deep peak shaving process of the steam turbine. First, a main steam temperature change model is constructed based on the initial temperature, the final temperature, the rate of temperature change in each segment, and the corresponding operating time. This model divides the entire deep peak shaving process into several stages by establishing multiple linear or piecewise functional relationships. Each stage is determined by a specific rate of temperature change and its duration, thus achieving an accurate description of the main steam temperature change process.

[0082] In one embodiment, the main steam temperature curve of the deep peak shaving process of the steam turbine is parametrically modeled, as described below:

[0083] (1)

[0084] In the formula, T i The end temperature of the main steam in the i-th segment is... Let i be the peak-shaving operation time. and Let T = t0 and T = T0 be the temperature change rate and the initial temperature of the i-th time interval, respectively. At t = t0, T = T0 is the main steam temperature at the start of the peak-shaving process; at t = t3, T = T f This refers to the main steam temperature at the end of the peak shaving process.

[0085] The main steam temperature is parameterized as a multi-segment curve with different rates of change over operating time to simulate the varying demands and control strategies of the main steam temperature during the deep peak shaving process of the steam turbine. The schematic diagram shows that the temperature-time curve consists of three segments. Specifically, the main steam temperature curve is composed of... , , , , , , These seven variables are collectively defined, and based on this variable system, various main steam temperature-time curves can be generated, i.e., main steam temperature control strategies, providing input data for subsequent sample collection and calculation.

[0086] S202: Generate the main steam temperature change curve using the main steam temperature change model.

[0087] Specifically, after the model is built, the main steam temperature change curve is generated using the main steam temperature change model. This curve reflects the temperature change trend of the steam turbine throughout the entire deep peak shaving process, including cooling, steady-state, and recovery stages, and intuitively demonstrates the dynamic change process of the main steam temperature under different peak shaving strategies. Through this parametric modeling method, representative temperature change curves can be generated, providing a unified input basis for subsequent sample collection, finite element analysis, and optimization calculations.

[0088] In one embodiment, a parameterized schematic diagram of the main steam temperature profile is shown below. Figure 12 As shown.

[0089] Through the above parametric modeling steps, this application realizes a standardized description of the main steam temperature change process, enabling the temperature control strategy to flexibly express the temperature regulation characteristics of different peak-shaving conditions with a small number of parameters, laying the foundation for the rapid optimization of main steam temperature control.

[0090] S102: Based on the main steam temperature change curve, obtain samples and perform finite element numerical simulation on the samples to obtain an optimized dataset;

[0091] Specifically, the server acquires samples based on the main steam temperature variation curve and performs finite element numerical simulations on these samples to obtain an optimized dataset. The main steam temperature curves under different operating conditions are then input into the finite element analysis model of the turbine rotor to conduct thermo-mechanical coupling numerical simulations, calculating the rotor temperature field distribution and maximum equivalent stress value corresponding to each sample curve. Through simulation calculations on a large number of samples, a dataset covering various operating states is formed, providing a high-quality data foundation for training the surrogate model.

[0092] Figure 3 This is a flowchart illustrating an embodiment of the turbine deep peak-shaving main steam temperature control optimization method provided in this application, as shown below. Figure 3 As shown, S102 includes:

[0093] S301: Construct a sampling space based on the main steam temperature change curve;

[0094] Specifically, the process of obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples to obtain an optimized dataset is mainly used to establish a high-precision data correlation between the main steam temperature and the rotor stress response during the deep peak shaving process of the steam turbine, providing data support for subsequent surrogate model training.

[0095] The server constructs a sampling space based on the aforementioned main steam temperature change curve. The sampling space is defined by the parameterized variables of the main steam temperature curve, including initial temperature values, final temperatures, the rate of temperature change at each stage, and the corresponding operating time. By determining the range of these parameters, a multi-dimensional parameter space covering the characteristics of different peak-shaving conditions can be formed, thereby ensuring the representativeness and diversity of the samples.

[0096] S302: The sample is obtained by sampling using the Latin hypercube sampling method within the sampling space;

[0097] Specifically, after constructing the sampling space, the server selects samples within that space using the Latin hypercube sampling method. This method achieves a uniform and efficient sample distribution in a multidimensional variable space, avoiding the concentration bias problem that occurs in traditional random sampling. The multiple samples generated by this sampling method each correspond to a different main steam temperature change curve, representing different peak-shaving operation strategies.

[0098] S303: Input the sample into the pre-built finite element model to perform thermo-mechanical coupling numerical simulation, and calculate the corresponding rotor temperature field and maximum equivalent stress;

[0099] Specifically, the server inputs the samples into a pre-established finite element model of the turbine rotor and conducts a thermo-mechanical coupled numerical simulation analysis. By applying the main steam temperature curve as a thermal boundary condition to the model, the temperature field changes and stress distribution of the turbine rotor during deep peak shaving are calculated. The simulation yields the rotor temperature field distribution and maximum equivalent stress value for each sample.

[0100] S304: Construct the optimized dataset based on the main steam temperature change curve, the rotor temperature field, and the maximum equivalent stress.

[0101] Specifically, the server constructs an optimized dataset based on the main steam temperature variation curve, rotor temperature field, and maximum equivalent stress results. This dataset uses the main steam temperature curve parameters as input data and the calculated rotor thermal stress and temperature field response as output data, forming a complete sample set reflecting the relationship between the main steam temperature control process and the rotor response. This optimized dataset effectively captures the impact of different temperature control strategies on the rotor's thermodynamic behavior, providing high-precision and comprehensive sample support for subsequent surrogate model training.

[0102] Through the above steps, this application can establish a high-quality dataset covering various deep peak-shaving operation states, ensuring that the subsequent prediction model has good generalization ability and accuracy, and laying a solid data foundation for realizing the intelligent optimization of the turbine main steam temperature control strategy.

[0103] In one embodiment, firstly define , , , , , , The range of values ​​for seven variables constitutes the sampling space for the parameterized curve of the main steam temperature during the deep peak shaving process. Based on the combination of these seven variables, multiple sets of main steam temperature curves for the deep peak shaving process of the steam turbine can be generated by changing the values ​​of the variables. Then, the main steam temperature curve is sampled using the Latin hypercube sampling method within the sampling space. The sampling space of the main steam temperature curve is shown in Table 1. The corresponding temperature parameters and convective heat transfer coefficient are calculated using the temperature curve data and empirical formulas for calculating the heat transfer coefficient of the steam turbine rotor. Then, for each curve, the finite element method is used to conduct thermo-mechanical coupling calculation and analysis of the steam turbine rotor in the corresponding deep peak shaving process. Specifically, the temperature curve is input as the thermal load boundary condition into the finite element model to calculate the stress field distribution of the rotor, and the maximum equivalent stress value and the corresponding temperature field are extracted as the output data corresponding to that temperature curve. The input data of the sample set is defined as the main steam temperature curve defined by the seven variables, and the output data are the maximum equivalent stress of the rotor and the corresponding temperature field under the corresponding temperature curve. This led to the construction of a deep peak-shaving process main steam temperature control optimization dataset, which includes inputs (temperature curves defined by 7 variables) and outputs (rotor maximum equivalent stress and temperature field at the moment of maximum equivalent stress). This dataset was then divided into a training set and a test set in a 7:3 ratio. The training set was used for training the subsequent surrogate model, while the test set was used to verify the model's accuracy.

[0104] Table 1

[0105]

[0106] in, This is the initial main steam temperature; The main steam temperature at the end of peak-shaving operation; The rate of temperature change within a single segment of the curve; The corresponding peak-shaving operating time; n is the number of segments in the main steam temperature curve; during sampling, the parameters in the table must meet the following requirements:

[0107] (2)

[0108] After sampling, the calculation results need to be extracted to establish a dataset for model training and testing. A single sample consists of input data comprised of the main steam temperature curve for a peak-shaving operation and output data comprised of the rotor temperature field and maximum equivalent stress. Since the input and output data have different dimensions and significant numerical differences, a min-max normalization method is used to map the input and output data to the interval [0,1] in order to remove the dimensionality and accelerate the model's convergence during training.

[0109] (3)

[0110] In one embodiment, the dataset creation flowchart is as follows: Figure 13 As shown.

[0111] S103: Use the optimized dataset to train the model and obtain the surrogate model;

[0112] Specifically, the server uses an optimized dataset to train a surrogate model. This surrogate model establishes a mapping between the main steam temperature variation curve and the rotor stress response using deep learning algorithms. It can quickly predict the thermodynamic response of the turbine rotor under different temperature control strategies without relying on complex finite element calculations, achieving end-to-end prediction from temperature parameters to rotor stress indices. This model features high accuracy and high computational efficiency, providing data support for rapid optimization.

[0113] Figure 4 This is a flowchart illustrating an embodiment of the turbine deep peak-shaving main steam temperature control optimization method provided in this application, as shown below. Figure 4 As shown, S103 includes:

[0114] S401: Construct a first deep neural network model for predicting the temperature field of a steam turbine rotor and a second deep neural network model for predicting the maximum equivalent stress of the rotor;

[0115] Specifically, the process of training a model using an optimized dataset to obtain a surrogate model aims to establish a high-precision computational model that can quickly predict the thermal response of a steam turbine rotor through deep learning technology, thereby replacing complex finite element numerical simulation and significantly improving the efficiency of optimization calculations.

[0116] The server first constructs a first deep neural network model to predict the turbine rotor temperature field, and a second deep neural network model to predict the rotor's maximum equivalent stress. The first deep neural network model uses the parameters of the main steam temperature curve as input data and predicts the corresponding rotor temperature field distribution through a nonlinear mapping relationship, achieving rapid calculation from external temperature control signals to internal thermal response. The second deep neural network model uses the rotor temperature field output by the first model as input data to further predict the rotor's maximum equivalent stress, achieving an accurate mapping from temperature field to stress response. This two-level neural network structure enables end-to-end prediction from temperature curves to stress indicators.

[0117] S402: The first deep neural network model and the second deep neural network model are iteratively trained using the optimized dataset to obtain the surrogate model.

[0118] Specifically, after the model is built, the first and second deep neural network models are iteratively trained using the optimized dataset. During training, the optimized dataset is divided into a training set and a test set according to a certain ratio. The training set is used for parameter learning, and the test set is used for model performance verification. By continuously iteratively optimizing the neural network parameters, the error between the prediction results and the finite element simulation results gradually converges. During training, mainstream optimization algorithms (such as the Adam optimizer) and appropriate loss functions (such as the SmoothL1 loss function) can be used to update the model parameters and evaluate accuracy, thereby improving the accuracy and stability of the model in stress and temperature field prediction.

[0119] Once training is complete, the resulting surrogate model can replace the finite element method (FEM) model in subsequent optimization processes, enabling rapid prediction of turbine rotor stress states under different main steam temperature control strategies. This surrogate model possesses high accuracy, low computational cost, and good generalization performance, significantly shortening optimization computation time and providing a reliable prediction tool for rapid optimization of main steam temperature control strategies under deep peak-shaving conditions.

[0120] Through the above steps, this application introduces a deep learning model into the turbine temperature control optimization process, realizing the intelligent replacement of complex physical models. This not only improves the model's computational efficiency but also enhances real-time response capabilities while ensuring prediction accuracy, providing technical support for achieving safe and rapid optimization control of deep peak shaving in turbines.

[0121] In one embodiment, such as Figure 14 As shown, a computational proxy model for optimizing the main steam temperature control in the deep peak shaving process of a steam turbine is established based on a deep neural network (DNN). First, a turbine rotor temperature field mapping network DNN1 (the first deep neural network model) is established for the deep peak shaving process: the network input consists of the main steam temperature curve parameters for the deep peak shaving process, namely seven variables: the initial temperature of deep peak shaving... and the end temperature Temperature change rate Time parameters ( (This can be derived from other variables and is not used as input). Through a deep neural network DNN1, after linear transformation, ReLU activation function, and other operations, the turbine rotor temperature field under the corresponding main steam temperature curve is output. The mapping relationship is expressed as:

[0122] (4)

[0123] In the formula, For the predicted rotor temperature field, These are the learnable parameters of DNN1. This indicates the mapping of the DNN1 network.

[0124] Then, a rotor maximum equivalent stress prediction network DNN2 (the second deep neural network model) was established: the rotor temperature field output by DNN1 was used as the basis for prediction. As input, the output, through a downsampling layer and a fully connected layer, is the predicted value of the rotor's maximum equivalent stress. The corresponding mapping relationship is:

[0125] (5)

[0126] In the formula: For the learnable parameters of DNN2, This represents the mapping of the DNN2 network. Then, SmoothL1 Loss is used as the loss function, based on the predicted rotor temperature field. Numerical calculation of rotor temperature field Train the DNN1 network and predict the maximum equivalent stress. Numerical calculation of maximum equivalent stress The DNN2 network was trained. During training, 70% of the dataset was randomly selected as the training set, and the remaining 30% was used as the test set. The Adam optimizer was used uniformly during training, with an initial learning rate of 0.001. Every 100 training steps, the learning rate was reduced to 1 / 5 of its original value, iteratively optimizing the network parameters. , The accuracy of the trained DNN1 and DNN2 networks was validated on the test set. To evaluate the overall level of network accuracy, the goodness-of-fit R (R² score) and the dimensionless mean absolute error were selected. As an indicator of accuracy:

[0127] (6)

[0128] (7)

[0129] In the formula, For the dataset sample space, The number of samples in the dataset. , and These are the actual value, the predicted value, and the average value, respectively.

[0130] S104: Optimize the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm.

[0131] Specifically, the server optimizes the main steam temperature control strategy based on the aforementioned surrogate model and fast simulated annealing algorithm. By using the main steam temperature curve parameters as design variables, the surrogate model is invoked to predict the rotor stress response in real time. Then, using the fast simulated annealing algorithm, under the constraint that the rotor stress does not exceed the allowable stress of the material, the optimal temperature control strategy is quickly searched with the shortest main steam temperature adjustment time as the optimization objective. This optimization process can provide the optimal adjustment scheme that meets safety constraints in a short time, thereby significantly improving the temperature control efficiency and safety of the unit under deep peak-shaving conditions.

[0132] In summary, this application achieves highly efficient optimization of turbine main steam temperature control by organically combining parametric modeling, finite element simulation, deep learning proxy modeling, and heuristic optimization algorithms. This method can quickly determine the optimal main steam temperature change path while ensuring that rotor thermal stress does not exceed limits, thereby improving the unit's deep peak-shaving response speed and operational safety level.

[0133] Figure 5 This is a flowchart illustrating an embodiment of the turbine deep peak-shaving main steam temperature control optimization method provided in this application, as shown below. Figure 5 As shown, S104 includes:

[0134] S501: Using the temperature parameter as the design variable, call the surrogate model to predict the maximum equivalent stress of the turbine rotor under different temperature control strategies;

[0135] Specifically, the process of optimizing the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm aims to quickly determine the optimal main steam temperature control strategy that meets the allowable stress constraints while ensuring the safe operation of the turbine rotor, thereby improving the temperature regulation response speed and system safety during deep peak shaving.

[0136] The server uses the temperature parameters of the main steam as design variables. These temperature parameters include the initial temperature, final temperature, rate of temperature change at each stage, and corresponding running time of the main steam, which characterize different temperature control strategies. By adjusting these parameters, various possible temperature change curves of the main steam can be generated.

[0137] Subsequently, a pre-trained surrogate model is invoked to rapidly evaluate different temperature control strategies. The surrogate model can predict the maximum equivalent stress value of the turbine rotor under the corresponding main steam temperature curve without relying on complex finite element calculations. This process enables real-time assessment of the safety of the control strategy, significantly reduces computational costs, and substantially improves the efficiency of the optimization search.

[0138] S502: Under the constraint that the maximum equivalent stress does not exceed the preset allowable stress of the material, with the shortest main steam temperature adjustment time as the optimization objective, the optimal combination of temperature parameters is obtained by using a fast simulated annealing algorithm.

[0139] Specifically, the server uses the constraint that the maximum equivalent stress of the rotor does not exceed the allowable stress of the material, and the optimization objective is to minimize the main steam temperature regulation time. It employs a fast simulated annealing algorithm to search for the optimal solution. The fast simulated annealing algorithm is a heuristic algorithm based on the principles of random search and global optimization. By continuously generating new solutions within the parameter space and selecting the optimal parameter combination based on the acceptance probability function, it effectively avoids getting trapped in local optima. This algorithm gradually reduces the system's "temperature" parameter, causing the search process to gradually converge from global exploration to the optimal region, thereby obtaining the optimal temperature parameter combination that satisfies the constraints.

[0140] During the optimization process, the algorithm updates the design variables through multiple iterations, with each iteration calling a surrogate model for stress prediction and target calculation. When the objective function (i.e., the main steam temperature regulation time) stabilizes or reaches the preset maximum number of iterations, the algorithm terminates and outputs the optimal main steam temperature control strategy for the turbine's deep peak shaving process. This strategy enables the unit to operate safely and stably within the shortest temperature regulation time, effectively improving the response performance of deep peak shaving.

[0141] Through the aforementioned optimization steps, this application, supported by the rapid prediction capabilities of the surrogate model and combined with an efficient fast simulated annealing algorithm, achieves high-precision and rapid optimization of the main steam temperature control strategy. This method not only significantly shortens the optimization calculation time but also obtains the optimal temperature regulation scheme that meets safety constraints under complex operating conditions, thereby improving the regulation efficiency and operational reliability of the steam turbine during deep peak shaving.

[0142] In one embodiment, the control strategy for the main steam temperature curve during deep peak shaving is optimized by combining a fast simulated annealing algorithm and a high-precision surrogate model. The optimization objective is to quickly provide the shortest time for main steam temperature regulation under different deep peak shaving demand conditions, while meeting the turbine rotor safety operation constraints (i.e., the maximum equivalent stress value is less than the allowable stress of the rotor material), thereby improving the unit's peak shaving response speed.

[0143] When using the fast simulated annealing algorithm for optimization, the stress constraint condition is given as the acceptance probability function of the new solution. The algorithm parameters are set according to the default optimal values. The optimization algorithm can be represented as follows:

[0144] (8)

[0145] In the formula, For the algorithm optimization objective, This represents the allowable stress of the rotor material.

[0146] The formula for calculating the algorithm to update the design variable values ​​is as follows:

[0147] (9)

[0148] In the formula, xc is the updated value of the design variable:

[0149] (10)

[0150] In the formula, y is the parameter update coefficient:

[0151] (11)

[0152] In the formula, u is a random array in the range [-1, 1], and the array size dimension is consistent with the design variable dimension; sign is the sign function:

[0153] (12)

[0154] The updated temperature value is calculated as follows:

[0155] (13)

[0156] In the formula, it represents the number of iterations at a certain temperature, ranging from [1, L]; q represents the cooling exponent, which is set to 1; and c represents the cooling coefficient.

[0157] (14)

[0158] In the formula, m is the cooling parameter, which is set to 1; n is the exponential parameter, which is set to 1.

[0159] To rapidly optimize the main steam temperature curve control during the deep peak shaving process, the fast simulated annealing algorithm described in this application is used to solve the main steam temperature control strategy optimization. During this process, the surrogate models DNN1 and DNN2 established in this application provide the maximum equivalent rotor stress under different main steam temperature control strategies, thereby accelerating the optimization process and achieving rapid optimization. When the optimization objective value no longer changes or the maximum number of iterations is reached, the optimized main steam temperature control strategy for the deep peak shaving process of the turbine is obtained.

[0160] This application provides an optimization method for main steam temperature control during deep peak shaving of a steam turbine. The method involves parametrically modeling the main steam temperature curve during the deep peak shaving process based on acquired temperature parameters to generate a main steam temperature variation curve. Samples are obtained from the main steam temperature variation curve, and finite element numerical simulations are performed on these samples to obtain an optimized dataset. The optimized dataset is then used for model training to obtain a surrogate model. Finally, the main steam temperature control strategy is optimized based on the surrogate model and a fast simulated annealing algorithm. This method achieves rapid acquisition of the optimal main steam temperature control strategy while meeting the turbine rotor safety constraints, thereby significantly improving the response speed and operational safety of the deep peak shaving process.

[0161] Specifically, by parametrically modeling the main steam temperature curve during the deep peak shaving process of the steam turbine based on the acquired temperature parameters, a main steam temperature change curve is generated, achieving an accurate description of the main steam temperature change law during the deep peak shaving process of the steam turbine, and providing a mathematical model basis for subsequent data sampling and optimization; by obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples, an optimized dataset is obtained, providing high-precision data on the influence of main steam temperature change on rotor stress and temperature field, providing reliable samples for model training; by using the optimized dataset for model training, a surrogate model is obtained, establishing a high-precision model that can quickly predict rotor thermal stress response, significantly reducing computation time; by optimizing the main steam temperature control strategy based on the surrogate model and fast simulated annealing algorithm, rapid optimization of the main steam temperature control strategy is achieved under the premise of ensuring rotor safety constraints, shortening the deep peak shaving response time.

[0162] This application achieves rapid optimization of main steam temperature control during deep peak shaving of steam turbines by combining deep learning with a fast simulated annealing algorithm. The surrogate model constructed using a deep neural network can perform rapid and accurate end-to-end prediction of the rotor's thermodynamic state, providing high-precision stress and temperature field assessments for the optimization process. Combined with the efficient fast simulated annealing algorithm for control strategy optimization, the optimal main steam temperature regulation scheme can be quickly obtained under the condition of satisfying the rotor's allowable stress constraints. Compared with traditional methods relying on numerical simulation or shallow optimization, this application significantly improves the speed and accuracy of main steam temperature control strategy optimization, achieving a simultaneous improvement in the safety and response performance of the steam turbine under deep peak shaving conditions.

[0163] Based on the same inventive concept, this application also provides a turbine deep peak shaving main steam temperature control optimization device, which can be used to implement the method described in the above embodiments, as described in the following embodiments. Since the principle of the turbine deep peak shaving main steam temperature control optimization device in solving the problem is similar to that of the turbine deep peak shaving main steam temperature control optimization method, the implementation of the turbine deep peak shaving main steam temperature control optimization device can refer to the implementation of the software performance benchmark determination method, and repeated details will not be elaborated further. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the system described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0164] Figure 6 This is a schematic diagram of the structure of a turbine deep peak-shaving main steam temperature control optimization device provided in an embodiment of this application, as shown below. Figure 6 As shown, the turbine deep peak-shaving main steam temperature control optimization device provided in this application includes:

[0165] The curve generation unit 601 is used to parametrically model the main steam temperature curve of the deep peak shaving process of the steam turbine based on the acquired temperature parameters, and generate the main steam temperature change curve.

[0166] Data set generation unit 602 is used to obtain samples based on the main steam temperature change curve and perform finite element numerical simulation on the samples to obtain an optimized data set;

[0167] The surrogate model generation unit 603 is used to train the model using the optimized dataset to obtain the surrogate model.

[0168] The control optimization unit 604 is used to optimize the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm.

[0169] Figure 7 This is a schematic diagram of the structure of a turbine deep peak-shaving main steam temperature control optimization device provided in an embodiment of this application. Figure 6 Based on the previous embodiment, the temperature parameters further include the initial temperature, the final temperature, the rate of temperature change for each segment, and the corresponding running time; for example... Figure 7 As shown, the curve generation unit 601 includes:

[0170] Temperature change model construction module 701 is used to construct a main steam temperature change model based on the initial temperature, the end temperature, the temperature change rate of each segment and the corresponding running time.

[0171] The curve generation module 702 is used to generate the main steam temperature change curve using the main steam temperature change model.

[0172] Figure 8 This is a schematic diagram of the structure of a turbine deep peak-shaving main steam temperature control optimization device provided in an embodiment of this application. Figure 6 Based on the embodiments, further, such as Figure 8 As shown, the dataset generation unit 602 includes:

[0173] The sampling space construction module 801 is used to construct a sampling space based on the main steam temperature change curve.

[0174] The sample acquisition module 802 is used to collect samples in the sampling space using a Latin hypercube sampling device to obtain the sample;

[0175] The data calculation module 803 is used to input the sample into a pre-built finite element model for thermo-solid coupling numerical simulation, and calculate the corresponding rotor temperature field and maximum equivalent stress.

[0176] The dataset construction module 804 is used to construct the optimized dataset based on the main steam temperature change curve, the rotor temperature field, and the maximum equivalent stress.

[0177] Figure 9 This is a schematic diagram of the structure of a turbine deep peak-shaving main steam temperature control optimization device provided in an embodiment of this application. Figure 6 Based on the embodiments, further, such as Figure 9 As shown, the proxy model generation unit 603 includes:

[0178] Neural network model building module 901 is used to build a first deep neural network model for predicting the temperature field of the turbine rotor and a second deep neural network model for predicting the maximum equivalent stress of the rotor.

[0179] The surrogate model generation module 902 is used to iteratively train the first deep neural network model and the second deep neural network model using the optimized dataset to obtain the surrogate model.

[0180] Figure 10 This is a schematic diagram of the structure of a turbine deep peak-shaving main steam temperature control optimization device provided in an embodiment of this application. Figure 6 Based on the embodiments, further, such as Figure 10 As shown, the control optimization unit 604 includes:

[0181] The model prediction module 1001 is used to use the temperature parameter as a design variable and call the proxy model to predict the maximum equivalent stress of the turbine rotor under different temperature control strategies.

[0182] The control optimization module 1002 is used to obtain the optimal combination of temperature parameters by using a fast simulated annealing algorithm, under the constraint that the maximum equivalent stress does not exceed the preset allowable stress of the material, with the shortest main steam temperature adjustment time as the optimization objective.

[0183] This application provides a method and apparatus for optimizing the main steam temperature control during deep peak shaving of a steam turbine. The method involves parametrically modeling the main steam temperature curve during the deep peak shaving process based on acquired temperature parameters to generate a main steam temperature variation curve. Samples are obtained from the main steam temperature variation curve, and finite element numerical simulations are performed on these samples to obtain an optimized dataset. The optimized dataset is then used for model training to obtain a surrogate model. Finally, the main steam temperature control strategy is optimized based on the surrogate model and a fast simulated annealing algorithm. This method achieves rapid acquisition of the optimal main steam temperature control strategy while meeting the turbine rotor safety constraints, thereby significantly improving the response speed and operational safety of the deep peak shaving process.

[0184] Specifically, by parametrically modeling the main steam temperature curve during the deep peak shaving process of the steam turbine based on the acquired temperature parameters, a main steam temperature change curve is generated, achieving an accurate description of the main steam temperature change law during the deep peak shaving process of the steam turbine, and providing a mathematical model basis for subsequent data sampling and optimization; by obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples, an optimized dataset is obtained, providing high-precision data on the influence of main steam temperature change on rotor stress and temperature field, providing reliable samples for model training; by using the optimized dataset for model training, a surrogate model is obtained, establishing a high-precision model that can quickly predict rotor thermal stress response, significantly reducing computation time; by optimizing the main steam temperature control strategy based on the surrogate model and fast simulated annealing algorithm, rapid optimization of the main steam temperature control strategy is achieved under the premise of ensuring rotor safety constraints, shortening the deep peak shaving response time.

[0185] This application achieves rapid optimization of main steam temperature control during deep peak shaving of steam turbines by combining deep learning with a fast simulated annealing algorithm. The surrogate model constructed using a deep neural network can perform rapid and accurate end-to-end prediction of the rotor's thermodynamic state, providing high-precision stress and temperature field assessments for the optimization process. Combined with the efficient fast simulated annealing algorithm for control strategy optimization, the optimal main steam temperature regulation scheme can be quickly obtained under the condition of satisfying the rotor's allowable stress constraints. Compared with traditional methods relying on numerical simulation or shallow optimization, this application significantly improves the speed and accuracy of main steam temperature control strategy optimization, achieving a simultaneous improvement in the safety and response performance of the steam turbine under deep peak shaving conditions.

[0186] Figure 11This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of this application, as shown below. Figure 11 As shown, the electronic device may include: a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104, wherein the processor 1101, the communication interface 1102, and the memory 1103 communicate with each other through the communication bus 1104. The processor 1101 can call the logic instructions in the memory 1103 to execute the following methods: parametrically modeling the main steam temperature curve of the deep peak shaving process of the steam turbine according to the acquired temperature parameters to generate a main steam temperature change curve; obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples to obtain an optimized dataset; using the optimized dataset to train the model to obtain a surrogate model; and optimizing the main steam temperature control strategy based on the surrogate model and the fast simulated annealing algorithm.

[0187] Furthermore, the logical instructions in the aforementioned memory 1103 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a top-drive control center server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0188] This embodiment discloses a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by the computer, the computer can execute the methods provided in the above-described method embodiments, such as: parametrically modeling the main steam temperature curve of the deep peak shaving process of the steam turbine based on the acquired temperature parameters to generate a main steam temperature change curve; obtaining samples based on the main steam temperature change curve and performing finite element numerical simulation on the samples to obtain an optimized dataset; using the optimized dataset to train a model to obtain a surrogate model; and optimizing the main steam temperature control strategy based on the surrogate model and a fast simulated annealing algorithm.

[0189] This embodiment provides a computer-readable storage medium storing a computer program that causes a computer to execute the methods provided in the above-described method embodiments. For example, the methods include: parametrically modeling the main steam temperature curve of the deep peak-shaving process of a steam turbine based on acquired temperature parameters to generate a main steam temperature variation curve; acquiring samples based on the main steam temperature variation curve and performing finite element numerical simulation on the samples to obtain an optimized dataset; training a model using the optimized dataset to obtain a surrogate model; and optimizing the main steam temperature control strategy based on the surrogate model and a fast simulated annealing algorithm.

[0190] Those skilled in the art will understand that embodiments of this application can be provided as methods, 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 embodied 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.

[0191] 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 a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions 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.

[0192] 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.

[0193] 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 1One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0194] In the description of this specification, the references to terms such as "an embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0195] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above descriptions are merely specific embodiments of this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for optimizing control of main steam temperature in deep peak shaving of a steam turbine, characterized in that, The method comprises the following steps: Parameterized modeling is performed on a main steam temperature curve of a steam turbine deep peak shaving process according to obtained temperature parameters, and a main steam temperature change curve is generated; Samples are obtained based on the main steam temperature change curve, and finite element numerical simulation is performed on the samples, so as to obtain an optimization data set; Model training is performed by using the optimization data set, and a proxy model is obtained; Main steam temperature control strategy optimization is performed based on the proxy model and a fast simulated annealing algorithm.

2. The method of claim 1, wherein, The temperature parameters comprise an initial temperature, an end temperature, a temperature change rate of each section and corresponding operation time; the parameterized modeling is performed on the main steam temperature curve of the steam turbine deep peak shaving process according to the obtained temperature parameters, and the main steam temperature change curve is generated, which comprises the following steps: A main steam temperature change model is constructed according to the initial temperature, the end temperature, the temperature change rate of each section and the corresponding operation time; The main steam temperature change model is used to generate the main steam temperature change curve.

3. The method of claim 1, wherein, The samples are obtained based on the main steam temperature change curve, and the finite element numerical simulation is performed on the samples, so as to obtain the optimization data set, which comprises the following steps: A sampling space is constructed based on the main steam temperature change curve; The Latin hypercube sampling method is used to sample in the sampling space, so as to obtain the samples; The samples are input into a pre-constructed finite element model to perform thermal-structural coupling numerical simulation, so as to calculate a corresponding rotor temperature field and a maximum equivalent stress; The optimization data set is constructed according to the main steam temperature change curve, the rotor temperature field and the maximum equivalent stress.

4. The method of claim 1, wherein, The model training is performed by using the optimization data set, and the proxy model is obtained, which comprises the following steps: A first deep neural network model for predicting a steam turbine rotor temperature field and a second deep neural network model for predicting a maximum equivalent stress of a rotor are constructed; The optimization data set is used to iteratively train the first deep neural network model and the second deep neural network model, so as to obtain the proxy model.

5. The method of claim 1, wherein, The main steam temperature control strategy optimization is performed based on the proxy model and the fast simulated annealing algorithm, which comprises the following steps: The temperature parameters are used as design variables, and the proxy model is called to predict a maximum equivalent stress of a steam turbine rotor under different temperature control strategies; Under the constraint condition that the maximum equivalent stress does not exceed a preset material allowable stress, the fast simulated annealing algorithm is used to obtain an optimal temperature parameter combination with the shortest main steam temperature adjustment time as an optimization target.

6. A device for optimizing control of main steam temperature in deep peak shaving of a steam turbine, characterized in that, The method comprises the following steps: A curve generation unit is configured to perform parameterized modeling on a main steam temperature curve of a steam turbine deep peak shaving process according to obtained temperature parameters, and generate a main steam temperature change curve; A data set generation unit is configured to obtain samples based on the main steam temperature change curve, and perform finite element numerical simulation on the samples, so as to obtain an optimization data set; A proxy model generation unit is configured to perform model training by using the optimization data set, and obtain a proxy model; A control optimization unit is configured to perform main steam temperature control strategy optimization based on the proxy model and a fast simulated annealing algorithm.

7. The turbine deep peaking main steam temperature control optimization apparatus of claim 6, wherein, The temperature parameters comprise an initial temperature, an end temperature, a temperature change rate of each section and corresponding operation time; The curve generation unit comprises: a temperature change model construction module, configured to construct a main steam temperature change model according to the initial temperature, the end temperature, the temperature change rate of each section and the corresponding operation time; a curve generation module, configured to generate the main steam temperature change curve by using the main steam temperature change model.

8. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the method in any one of claims 1 to 5 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program implements the method in any one of claims 1 to 5 when executed by a processor.

10. A computer program product, characterised in that, The computer program product comprises a computer program, and the computer program implements the method in any one of claims 1 to 5 when executed by a processor.

Images