Gas turbine combined cycle system load prediction method and system based on nonlinear regression model

Abstract

The invention relates to the technical field of gas turbine combined cycle systems, in particular to a load prediction method and system for a gas turbine combined cycle system based on a nonlinear regression model.The method comprises the steps that historical data of the cycle system are obtained, and the historical data comprise operation parameters, heat efficiency and system loads; based on the thermodynamic equation and the operation parameters, calculating to obtain the theoretical efficiency of the circulating system; using the operation parameters, the thermal efficiency and the theoretical efficiency to form a historical feature set; and predicting the load of the gas turbine combined cycle system. According to the method, a feature set containing real-time state monitoring, actual performance evaluation and physical limit constraint is constructed, irreversible loss in system operation is quantified, physical constraint conditions are provided for prediction, overfitting or working condition extrapolation failure caused by lack of physical significance of a pure data driving model is avoided, and prediction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas turbine combined cycle systems, and in particular to a gas turbine combined cycle system load prediction method and system based on a nonlinear regression model. Background Art

[0002] Combined cycle gas turbine (CCGT) systems, due to their high efficiency and low emissions, play an important role in power generation and comprehensive energy utilization. System load forecasting is one of the key technologies for optimizing their operating efficiency, reducing energy consumption, and ensuring grid stability. Traditional load forecasting methods mainly rely on statistical regression models based on historical data. However, the operation of CCGT systems involves complex processes with strong coupling of multiple variables (such as the coordinated operation of gas turbines, waste heat boilers, and steam turbines). Their dynamic characteristics are affected by multiple factors, including environmental parameters, equipment aging, and fuel characteristics, making it difficult for mechanism models to accurately describe actual nonlinear relationships. Traditional data-driven models, such as linear regression and shallow neural networks, have difficulty capturing the implicit correlation between high-dimensional operating parameters and load due to their single feature selection or insufficient model expression capabilities, resulting in bottlenecks in prediction accuracy. Summary of the Invention

[0003] (1) Purpose of the invention

[0004] The purpose of the present invention is to provide a method and system for gas turbine combined cycle system load prediction based on a nonlinear regression model, which calculates the theoretical efficiency of the cycle system through thermodynamic equations and operating parameters; uses operating parameters, thermal efficiency and theoretical efficiency as feature inputs into a nonlinear regression model to improve prediction accuracy.

[0005] (2) Technical solution

[0006] To solve the above problems, the present invention provides a method for load prediction of a gas turbine combined cycle system based on a nonlinear regression model, comprising:

[0007] Acquiring historical data of the circulation system, wherein the historical data includes operating parameters, thermal efficiency, and system load;

[0008] Based on thermodynamic equations and operating parameters, calculate the theoretical efficiency of the circulation system;

[0009] Using the operating parameters, thermal efficiency and theoretical efficiency, a historical feature set is constructed;

[0010] Establish a nonlinear regression model with the feature set as input variable and the system load as output variable;

[0011] Obtaining the error between the system load prediction value and the system load based on the historical feature set and the nonlinear regression model;

[0012] Based on the error, updating the nonlinear regression model parameters to obtain an updated nonlinear regression model;

[0013] The updated nonlinear regression model is used to predict the gas turbine combined cycle system load.

[0014] In another aspect of the present invention, preferably, the cycle system includes a gas turbine side and a steam turbine side, and the thermodynamic equation includes a gas turbine side thermodynamic equation and a steam turbine side thermodynamic equation;

[0015] Calculating the theoretical thermal efficiency of the gas turbine using the gas turbine side thermodynamic equation;

[0016] Calculating the theoretical thermal efficiency of the steam turbine using the steam turbine side thermodynamic equation;

[0017] The theoretical efficiency of the cycle system is obtained by using the theoretical thermal efficiency of the gas turbine side and the theoretical thermal efficiency of the steam turbine side.

[0018] In another aspect of the present invention, preferably, calculating the theoretical thermal efficiency of the gas turbine side using the gas turbine side thermodynamic equation includes:

[0019] Determining the maximum temperature of the gas turbine and the compressor inlet temperature among the operating parameters;

[0020] Based on the adiabatic compression process, the compressor outlet temperature is calculated, and based on the adiabatic expansion process, the gas turbine outlet temperature is calculated;

[0021] The theoretical thermal efficiency of the gas turbine side is calculated based on the maximum temperature of the gas turbine, the compressor inlet temperature, the compressor outlet temperature and the gas turbine outlet temperature.

[0022] In another aspect of the present invention, preferably, calculating the steam turbine side theoretical thermal efficiency using the steam turbine side thermodynamic equation includes:

[0023] determining a first steam temperature and a first steam pressure before the steam enters the steam turbine in the operating parameters, and determining a first enthalpy value from a table of water vapor thermodynamic properties based on the first steam temperature and the first steam pressure;

[0024] determining a second steam temperature and a second steam pressure of the steam in the steam turbine in the operating parameters, and determining a second enthalpy value from a water vapor thermodynamic property table based on the second steam temperature and the second steam pressure;

[0025] determining a third steam temperature and a third steam pressure at the condenser outlet in the operating parameters, and determining a third enthalpy value from a water vapor thermodynamic property table based on the third steam temperature and the third steam pressure;

[0026] The steam turbine side theoretical thermal efficiency is calculated based on the first enthalpy value, the second enthalpy value, and the third enthalpy value.

[0027] In another aspect of the present invention, preferably, the theoretical efficiency of the circulation system is calculated using the following formula:

[0028] η cc =η gt +η st (1-η gt )

[0029] Among them, η cc represents the theoretical efficiency of the circulation system, η st represents the theoretical thermal efficiency of the steam turbine side, η gt It represents the theoretical thermal efficiency of the gas turbine side.

[0030] In another aspect of the present invention, preferably, the nonlinear regression model is based on a physical information neural network, and the nonlinear regression model comprises: an input layer, a hidden layer, and an output layer;

[0031] The input layer includes a plurality of neurons, each of which corresponds to a feature dimension;

[0032] The hidden layer includes an LSTM layer and a fully connected layer;

[0033] The output layer outputs a linearly activated load prediction value.

[0034] In another aspect of the present invention, preferably, updating the nonlinear regression model parameters based on the error to obtain an updated nonlinear regression model includes:

[0035] Calculating the model loss of the nonlinear regression model based on the error and a preset loss function;

[0036] Adjusting the model parameters of the nonlinear regression model based on the model loss to obtain a nonlinear regression model with adjusted parameters;

[0037] Based on a preset number of updates, the nonlinear regression model after parameter adjustment is iteratively trained to obtain an updated nonlinear regression model.

[0038] In another aspect of the present invention, preferably, adjusting the model parameters of the nonlinear regression model includes:

[0039] An adaptive moment estimation optimization algorithm is used to calculate the gradient corresponding to the model loss, and the weight matrix and bias vector in the model parameters are updated according to the gradient backpropagation direction; wherein the adaptive moment estimation optimization algorithm dynamically adjusts the learning rate of different parameters by maintaining the first-order moment estimate and the second-order moment estimate.

[0040] In another aspect of the present invention, preferably, the iterative training process based on the preset number of updates further includes:

[0041] Before each iterative training, the early stopping index is calculated based on the prediction error of the current nonlinear regression model;

[0042] If the early stopping index does not decrease in several consecutive iterations, the training is terminated early and the model with the current parameter adjustment is output as the updated nonlinear regression model.

[0043] A gas turbine combined cycle system load forecasting system based on a nonlinear regression model, comprising:

[0044] The first acquisition module is used to acquire historical data of the circulation system, including operating parameters, thermal efficiency and system load;

[0045] Calculation module: Calculates the theoretical efficiency of the circulation system based on thermodynamic equations and operating parameters;

[0046] A construction module: constructing a historical feature set using the operating parameters, thermal efficiency and theoretical efficiency;

[0047] Establishing module: establishing a nonlinear regression model with feature set as input variable and system load as output variable;

[0048] The second acquisition module is used to obtain the error between the system load prediction value and the system load based on the historical feature set and the nonlinear regression model;

[0049] An updating module: updating the nonlinear regression model parameters based on the error to obtain an updated nonlinear regression model;

[0050] Prediction module: predicts the gas turbine combined cycle system load using the updated nonlinear regression model.

[0051] (3) Beneficial effects

[0052] The above technical solution of the present invention has the following beneficial technical effects:

[0053] The application quantifies the irreversible loss in system operation by constructing a feature set containing real-time state monitoring, actual performance evaluation and physical limit constraints, provides physical constraint conditions for prediction, and avoids overfitting or working condition extrapolation failure of pure data-driven models due to lack of physical meaning. It has significant advantages in prediction accuracy, working condition adaptability and long-term stability. The nonlinear regression model is used to capture the dynamic coupling mechanism between parameters, which can accurately represent the nonlinear superposition characteristics of each parameter during load change, and improve the prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 is the overall flowchart of an embodiment of the application. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application with reference to the specific embodiments and the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present application.

[0056] Obviously, the described embodiments are part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0057] In the description of the present application, it should be noted that the terms "first", "second", "third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0058] In addition, the technical features involved in different embodiments of the application described below can be combined with each other as long as there is no conflict.

[0059] The present application will be described in more detail below with reference to the accompanying drawings. In each of the drawings, the same elements are denoted by similar reference numerals. For the sake of clarity, each part in the drawings is not drawn to scale.

[0060] Embodiment one

[0061] A gas turbine combined cycle system load prediction method based on a nonlinear regression model, Figure 1 The overall flowchart of an embodiment of the application is shown as Figure 1 As shown, it includes:

[0062] Historical data of the circulation system is obtained, including operating parameters, thermal efficiency, and system load. In an embodiment of the present invention, the operating parameters include operating parameters used to calculate the theoretical efficiency of the circulation system and operating parameters used for nonlinear regression model prediction. The operating parameters specifically include: the maximum temperature of the gas turbine and the compressor inlet temperature, a first steam temperature and a first steam pressure before steam enters the steam turbine, a second steam temperature and a second steam pressure within the steam turbine, and a third steam temperature and a third steam pressure at the condenser outlet. The operating parameters used for prediction include the gas turbine speed, intake temperature, exhaust pressure, fuel flow rate, and cooling water flow rate. Operating parameters can be obtained through sensor recordings, such as temperature parameters on the gas turbine side. The intake temperature can be obtained by a sensor installed at the compressor inlet, the exhaust temperature can be obtained by a sensor installed at the gas turbine outlet, and the combustion chamber temperature can be obtained by a high-temperature section thermocouple. The fuel flow rate can be obtained by a natural gas / liquid fuel mass flow meter, and the air flow rate can be obtained by a compressor inlet flow meter. Steam turbine operating parameters include steam pressure and temperature, which can be obtained using pressure transmitters and thermocouples; steam flow rate, which can be obtained using a vortex flowmeter; and cooling water parameters, such as circulating water flow rate and cooling water pump temperature. Thermal efficiency can be calculated as the ratio of output electrical power to fuel input energy, with system load being the output electrical power.

[0063] Based on the thermodynamic equations and operating parameters, the theoretical efficiency of the cycle system is calculated and obtained. In this embodiment, the gas turbine combined cycle system includes a gas turbine side and a steam turbine side, and the thermodynamic equations include the thermal equations of the gas turbine side and the thermal equations of the steam turbine side. The gas turbine combined cycle system includes a gas turbine side and a steam turbine side, and the thermodynamic equations include the thermal equations of the gas turbine side and the thermal equations of the steam turbine side.

[0064] Calculating the theoretical thermal efficiency of the gas turbine side using the gas turbine side thermodynamic equation; in this embodiment, calculating the theoretical thermal efficiency of the gas turbine side using the gas turbine side thermodynamic equation includes:

[0065] Determining the maximum temperature of the gas turbine and the compressor inlet temperature among the operating parameters;

[0066] Based on the adiabatic compression process, the compressor outlet temperature is calculated, and based on the adiabatic expansion process, the gas turbine outlet temperature is calculated;

[0067] In this embodiment, the compressor performs adiabatic (isentropic) compression on the air. The compression process requires external work, which increases the internal energy of the air and the temperature. The temperature increase is determined by the pressure ratio π. The compressor outlet temperature is calculated using the following formula:

[0068]

[0069] Wherein, T2 represents the compressor outlet temperature, T1 represents the compressor inlet temperature, π represents the ratio of the compressor outlet to the compressor inlet pressure, and γ is the specific heat ratio of air. In this embodiment, the value is 1.4.

[0070] The high-temperature gas expands isentropically in the turbine to perform work, and the temperature drops. The expansion process releases energy, the gas temperature drops, and work is performed externally. The gas turbine outlet temperature is calculated using the following formula:

[0071]

[0072] Among them, T3 represents the maximum temperature of the gas turbine, that is, the combustion chamber outlet temperature, the temperature before the turbine; T4 represents the gas turbine outlet temperature.

[0073] The theoretical thermal efficiency of the gas turbine side is calculated based on the maximum temperature of the gas turbine, the compressor inlet temperature, the compressor outlet temperature, and the gas turbine outlet temperature. The thermal efficiency is the ratio of the net output work to the input heat. The heat released by the fuel in the combustion chamber corresponds to the temperature difference T3-T2, and the turbine work T3-T4 minus the compressor work T2-T1. Therefore, the theoretical thermal efficiency of the gas turbine side is calculated using the following formula:

[0074]

[0075] Among them, η gt Represents the theoretical thermal efficiency of the gas turbine. Based on the isentropic compression and expansion process of the Brayton cycle, the temperature change is deduced from the pressure ratio and the specific heat ratio of air, ultimately giving the theoretical thermal efficiency of the gas turbine.

[0076] The steam turbine side theoretical thermal efficiency is calculated using the steam turbine side thermodynamic equation. In this embodiment, the steam turbine side theoretical thermal efficiency is calculated using the steam turbine side thermodynamic equation, including:

[0077] determining a first steam temperature and a first steam pressure before the steam enters the steam turbine in the operating parameters, and determining a first enthalpy value from a table of water vapor thermodynamic properties based on the first steam temperature and the first steam pressure;

[0078] determining a second steam temperature and a second steam pressure of the steam in the steam turbine in the operating parameters, and determining a second enthalpy value from a water vapor thermodynamic property table based on the second steam temperature and the second steam pressure;

[0079] determining a third steam temperature and a third steam pressure at the condenser outlet in the operating parameters, and determining a third enthalpy value from a water vapor thermodynamic property table based on the third steam temperature and the third steam pressure;

[0080] The theoretical thermal efficiency of the steam turbine side is calculated based on the first enthalpy value, the second enthalpy value, and the third enthalpy value. In this embodiment, the theoretical thermal efficiency of the steam turbine side is calculated using the following formula:

[0081]

[0082] Among them, η st represents the theoretical thermal efficiency of the steam turbine. h1 represents the first enthalpy, h2 represents the second enthalpy, and h3 represents the third enthalpy. The work done by the steam expansion in the turbine is expressed as the first enthalpy minus the second enthalpy, and the heat absorbed by the steam in the boiler is expressed as the first enthalpy minus the third enthalpy. Based on a simplified Rankine cycle model, the theoretical thermal efficiency is calculated from the difference in enthalpy between the initial and final steam states.

[0083] The theoretical efficiency of the cycle system is obtained by using the theoretical thermal efficiency of the gas turbine side and the theoretical thermal efficiency of the steam turbine side.

[0084] The theoretical efficiency of the circulation system is calculated using the following formula:

[0085] η cc =η gt +η st (1-η gt )

[0086] Among them, η cc represents the theoretical efficiency of the circulation system, η st represents the theoretical thermal efficiency of the steam turbine side, η gt It represents the theoretical thermal efficiency of the gas turbine side.

[0087] The operating parameters, thermal efficiency, and theoretical efficiency are used to construct a historical feature set. This historical feature set aligns data from different sampling frequencies to a unified timestamp, handles missing values ​​and outliers, and eliminates dimensionality effects. Furthermore, lagged features, such as operating parameters from the past hour, are added to capture time dependencies. Operating parameters, including gas turbine speed, inlet air temperature, exhaust pressure, fuel flow rate, and cooling water flow rate, directly describe the system's real-time operating status. However, relying solely on these parameters may cause the model to ignore implicit system efficiency degradation or operating condition deviations. Therefore, in this embodiment, thermal efficiency and theoretical efficiency are also included. Thermal efficiency reflects the actual performance under the combined influence of multiple dynamic factors in actual operation, such as equipment aging, environmental conditions, fuel quality fluctuations, component wear, and control system response delays. Thermal efficiency accounts for nonlinear losses and uncertainties in actual operation. Theoretical efficiency, calculated based on thermodynamic equations, represents the theoretical performance limit of the system under ideal conditions and reflects the inherent characteristics of the system design parameters. The combination of these two approaches can quantify irreversible losses in system operation, such as heat transfer temperature differences, friction losses, and leakage, by comparing the efficiency differences between actual and ideal conditions. This provides physical constraints for prediction, avoiding the overfitting and operating condition extrapolation failures that can occur in purely data-driven models due to their lack of physical meaning. A feature space has been constructed that encompasses real-time state monitoring, operating parameters, actual performance evaluation, thermal efficiency and physical limit constraints, and theoretical efficiency. This approach offers significant advantages in prediction accuracy, operating condition adaptability, and long-term stability, making it suitable for complex thermal systems with strong nonlinear and time-varying characteristics.

[0088] Establishing a nonlinear regression model with the feature set as an input variable and the system load as an output variable; in this embodiment, the nonlinear regression model is based on a physical information neural network, and the nonlinear regression model includes: an input layer, a hidden layer, and an output layer;

[0089] The input layer includes several neurons, each corresponding to a feature dimension. The input layer is the interface between the model and external data. It contains several neurons, each corresponding to a feature dimension. These feature dimensions together constitute the input feature set.

[0090] The hidden layers include an LSTM layer and a fully connected layer. LSTM is used to handle long-term dependencies in time series data. In system load forecasting, the time series characteristics of historical load data are crucial for accurately predicting future loads. LSTM effectively processes time series information by introducing memory cells and gating mechanisms. The fully connected layer, located after the LSTM layer, further integrates and transforms the features output by the LSTM layer. Each neuron in a fully connected layer is connected to all neurons in the previous layer. The input is linearly transformed using a weight matrix and bias vector, and nonlinear factors are introduced using activation functions.

[0091] The output layer outputs a linearly activated load prediction value. The output layer uses a simple linear transformation to map the features output by the hidden layer to the prediction space of the system load.

[0092] Based on the historical feature set and the nonlinear regression model, an error between the system load prediction value and the system load is obtained, and based on the error, a parameter of the nonlinear regression model is updated to obtain an updated nonlinear regression model; comprising:

[0093] According to the error, based on a preset loss function, the model loss of the nonlinear regression model is calculated; the preset loss function can be mean square error, mean absolute error, Huber loss function, cross entropy loss function, etc.

[0094] Based on the model loss, the model parameters of the nonlinear regression model are adjusted to obtain a nonlinear regression model with adjusted parameters; further, the model parameters of the nonlinear regression model are adjusted, including: using an adaptive moment estimation optimization algorithm to calculate the gradient corresponding to the model loss, and updating the weight matrix and bias vector in the model parameters according to the gradient backpropagation direction; wherein the adaptive moment estimation optimization algorithm dynamically adjusts the learning rate of different parameters by maintaining the first-order moment estimation and the second-order moment estimation.

[0095] Based on a preset number of updates, the nonlinear regression model after parameter adjustment is iteratively trained to obtain an updated nonlinear regression model. Furthermore, the iterative training process based on the preset number of updates further includes:

[0096] Before each iterative training, the early stopping index is calculated based on the prediction error of the current nonlinear regression model;

[0097] If the early stopping metric does not decrease over several consecutive iterations, training is terminated early and the model with the current parameter adjustment is output as the updated nonlinear regression model. In this embodiment, the early stopping metric is if the error decreases by less than a threshold, such as if the decrease is less than 0.1% for five consecutive iterations, then training is terminated early.

[0098] The updated nonlinear regression model is used to predict the gas turbine combined cycle system load.

[0099] This invention constructs a feature set encompassing real-time state monitoring, actual performance evaluation, and physical limit constraints to quantify irreversible losses during system operation. This provides physical constraints for prediction and avoids overfitting or operating condition extrapolation failures that occur in purely data-driven models due to their lack of physical meaning. It offers significant advantages in predictive accuracy, operating condition adaptability, and long-term stability. By utilizing a nonlinear regression model to capture the dynamic coupling mechanism between parameters, it can accurately characterize the nonlinear superposition characteristics of each parameter during load changes, improving prediction accuracy.

[0100] Example 2

[0101] A gas turbine combined cycle system load forecasting system based on a nonlinear regression model, comprising:

[0102] The first acquisition module is used to acquire historical data of the circulation system, including operating parameters, thermal efficiency and system load;

[0103] Calculation module: Calculates the theoretical efficiency of the circulation system based on thermodynamic equations and operating parameters;

[0104] A construction module: constructing a historical feature set using the operating parameters, thermal efficiency and theoretical efficiency;

[0105] Establishing module: establishing a nonlinear regression model with feature set as input variable and system load as output variable;

[0106] The second acquisition module is used to obtain the error between the system load prediction value and the system load based on the historical feature set and the nonlinear regression model;

[0107] An updating module: updating the nonlinear regression model parameters based on the error to obtain an updated nonlinear regression model;

[0108] Prediction module: predicts the gas turbine combined cycle system load using the updated nonlinear regression model.

[0109] It should be understood that the above-described specific embodiments of the present invention are merely illustrative or illustrative of the principles of the present invention and do not constitute limitations of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included within the scope of protection of the present invention. In addition, the appended claims are intended to cover all variations and modifications that fall within the scope and metes and bounds of the appended claims, or equivalents thereof.

[0110] The present invention has been described above with reference to the embodiments thereof. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. The scope of the present invention is defined by the appended claims and their equivalents. Those skilled in the art may make various substitutions and modifications without departing from the scope of the present invention, and such substitutions and modifications are intended to fall within the scope of the present invention.

[0111] Although the embodiments of the present invention have been described in detail, it should be understood that the various changes, substitutions, and alterations could be made hereto without departing from the spirit and scope of the invention.

[0112] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A method for load prediction of a gas turbine combined cycle system based on a nonlinear regression model, characterized in that: include: Acquiring historical data of the circulation system, wherein the historical data includes operating parameters, thermal efficiency, and system load; Based on thermodynamic equations and operating parameters, calculate the theoretical efficiency of the circulation system; Using the operating parameters, thermal efficiency and theoretical efficiency, a historical feature set is constructed; Establish a nonlinear regression model with the feature set as input variable and the system load as output variable; Obtaining the error between the system load prediction value and the system load based on the historical feature set and the nonlinear regression model; Based on the error, updating the nonlinear regression model parameters to obtain an updated nonlinear regression model; The updated nonlinear regression model is used to predict the gas turbine combined cycle system load.

2. The prediction method according to claim 1, characterized in that The circulation system includes a gas turbine side and a steam turbine side, and the thermodynamic equation includes a gas turbine side thermodynamic equation and a steam turbine side thermodynamic equation; Calculating the theoretical thermal efficiency of the gas turbine using the gas turbine side thermodynamic equation; Calculating the theoretical thermal efficiency of the steam turbine using the steam turbine side thermodynamic equation; The theoretical efficiency of the cycle system is obtained by using the theoretical thermal efficiency of the gas turbine side and the theoretical thermal efficiency of the steam turbine side.

3. The prediction method according to claim 1, wherein: Calculating the theoretical thermal efficiency of the gas turbine using the gas turbine side thermodynamic equation includes: Determining the maximum temperature of the gas turbine and the compressor inlet temperature among the operating parameters; Based on the adiabatic compression process, the compressor outlet temperature is calculated, and based on the adiabatic expansion process, the gas turbine outlet temperature is calculated; The theoretical thermal efficiency of the gas turbine side is calculated based on the maximum temperature of the gas turbine, the compressor inlet temperature, the compressor outlet temperature and the gas turbine outlet temperature.

4. The prediction method according to claim 2, characterized in that Calculating the theoretical thermal efficiency of the steam turbine using the steam turbine side thermodynamic equation includes: determining a first steam temperature and a first steam pressure before the steam enters the steam turbine in the operating parameters, and determining a first enthalpy value from a table of water vapor thermodynamic properties based on the first steam temperature and the first steam pressure; determining a second steam temperature and a second steam pressure of the steam in the steam turbine in the operating parameters, and determining a second enthalpy value from a water vapor thermodynamic property table based on the second steam temperature and the second steam pressure; determining a third steam temperature and a third steam pressure at the condenser outlet in the operating parameters, and determining a third enthalpy value from a water vapor thermodynamic property table based on the third steam temperature and the third steam pressure; The steam turbine side theoretical thermal efficiency is calculated based on the first enthalpy value, the second enthalpy value, and the third enthalpy value.

5. The prediction method according to claim 2, characterized in that The theoretical efficiency of the circulation system is calculated using the following formula: or cc =the gt +n st (1st) gt ) Among them, η cc represents the theoretical efficiency of the circulation system, η st represents the theoretical thermal efficiency of the steam turbine side, η gt It represents the theoretical thermal efficiency of the gas turbine side.

6. The prediction method according to claim 1, characterized in that The nonlinear regression model is based on a physical information neural network and includes: an input layer, a hidden layer, and an output layer; The input layer includes a plurality of neurons, each of which corresponds to a feature dimension; The hidden layer includes an LSTM layer and a fully connected layer; The output layer outputs a linearly activated load prediction value.

7. The prediction method according to claim 1, wherein: Based on the error, updating the nonlinear regression model parameters to obtain an updated nonlinear regression model includes: Calculating the model loss of the nonlinear regression model based on the error and a preset loss function; Adjusting the model parameters of the nonlinear regression model based on the model loss to obtain a nonlinear regression model with adjusted parameters; Based on a preset number of updates, the nonlinear regression model after parameter adjustment is iteratively trained to obtain an updated nonlinear regression model.

8. The prediction method according to claim 7, characterized in that Adjusting the model parameters of the nonlinear regression model includes: An adaptive moment estimation optimization algorithm is used to calculate the gradient corresponding to the model loss, and the weight matrix and bias vector in the model parameters are updated according to the gradient backpropagation direction; wherein the adaptive moment estimation optimization algorithm dynamically adjusts the learning rate of different parameters by maintaining the first-order moment estimate and the second-order moment estimate.

9. The prediction method according to claim 7, characterized in that: The iterative training process based on the preset number of updates also includes: Before each iterative training, the early stopping index is calculated based on the prediction error of the current nonlinear regression model; If the early stopping index does not decrease in several consecutive iterations, the training is terminated early and the model with the current parameter adjustment is output as the updated nonlinear regression model.

10. A gas turbine combined cycle system load forecasting system based on a nonlinear regression model, characterized in that: include: The first acquisition module is used to acquire historical data of the circulation system, including operating parameters, thermal efficiency and system load; Calculation module: Calculates the theoretical efficiency of the circulation system based on thermodynamic equations and operating parameters; A construction module: constructing a historical feature set using the operating parameters, thermal efficiency and theoretical efficiency; Establishing module: establishing a nonlinear regression model with feature set as input variable and system load as output variable; The second acquisition module is used to obtain the error between the system load prediction value and the system load based on the historical feature set and the nonlinear regression model; An updating module: updating the nonlinear regression model parameters based on the error to obtain an updated nonlinear regression model; Prediction module: predicts the gas turbine combined cycle system load using the updated nonlinear regression model.

Images