A method for predicting start-up stress of a steam turbine rotor of a combined cycle unit

Abstract

The present disclosure provides a method and system for predicting the starting stress of a combined cycle unit steam turbine rotor. In the starting stage of the steam turbine main steam control valve opening degree of 0%-100%, the steam turbine speed, the main steam control valve opening degree and the high-pressure cylinder metal temperature are obtained in real time. The metal temperature is subjected to anti-interference noise reduction processing, and the surface temperature and its change rate are extracted. The rotor core temperature and its change rate are calculated based on a one-dimensional non-steady-state heat conduction model. The speed, surface temperature data and core temperature data are input into a pre-trained three-layer neural network stress prediction model, and the rotor real-time stress estimate is output. Through parameter simplification, anti-interference signal processing and lightweight neural network architecture, the present disclosure realizes millisecond-level real-time prediction of stress values, and significantly improves the unit starting speed and rotor life management accuracy.

Description

Technical Field

[0001] This invention relates to the field of cross-technology of gas-steam combined cycle power generation and industrial heating, and particularly to a method and system for predicting the start-up stress of a combined cycle turbine rotor. Background Technology

[0002] As the global energy structure accelerates its transition to a low-carbon model, natural gas power generation, as an important transitional energy source between traditional thermal power and renewable energy, plays an irreplaceable supporting role in ensuring the safe and stable operation of the power system. Gas-steam combined cycle units, with their power generation efficiency of up to 60% (compared to the average efficiency of approximately 45% for conventional coal-fired units) and peak-shaving characteristics of rapid start-up and shutdown within 15 minutes, have become the mainstream technology for natural gas power generation.

[0003] However, the frequent start-up and shutdown operations of combined cycle turbine units, which involve grid peak shaving, average 1.2-1.8 times per day, subject the turbine rotor to periodic high-temperature shocks. During the start-up phase, the thermal stress amplitude generated on the surface and central bore of the turbine rotor exceeds 300 MPa. This thermal stress originates from the internal constraint effect caused by uneven heating of the material. When metal components undergo unsteady-state heat transfer, a temperature gradient forms between their surface and interior, causing mutual constraints on the expansion and deformation of different regions, thereby creating a stress field within the material.

[0004] For thick-walled rotating components like steam turbine rotors, this is particularly evident during the unit startup phase: the rapid introduction of high-temperature steam causes the rotor surface to heat up and expand rapidly, while the rotor interior exhibits a significant temperature lag due to the thermal conductivity of the material. This inconsistency in radial temperature distribution causes the surface layer to bear pressure.

[0005] Excessive transient thermal stress not only accelerates the material damage process but also directly threatens the safe operation of the unit. When the local stress peak exceeds the material's bearing capacity, it will trigger the sudden propagation of macroscopic cracks, which can lead to catastrophic consequences such as rotor fracture in severe cases.

[0006] Therefore, there is an urgent need for a method that can predict the start-up stress of the turbine rotor in a combined cycle unit during the start-up phase. Summary of the Invention

[0007] A first aspect of this disclosure provides a method for predicting the start-up stress of a turbine rotor in a combined cycle unit, comprising the following steps: The turbine parameters during the start-up phase are obtained, including turbine speed, main steam control valve opening, and high-pressure cylinder metal temperature. The start-up phase is defined as the turbine main steam control valve opening ranging from 0% to 100%. The target high-pressure cylinder metal temperature is obtained by performing noise reduction processing on the metal temperature of the high-pressure cylinder. The instantaneous value and time derivative value of the target high-pressure cylinder metal temperature are extracted as the rotor surface temperature and the rotor surface temperature change rate. The rotor surface temperature and the rotor surface temperature change rate are input into a pre-trained radial heat transfer model, and the rotor core temperature is output. The rotor core temperature change rate is calculated based on the rotor core temperature. The calculation of the rotor core temperature is divided into two steps: the first step is the radial heat transfer model calculation, and the second step is the first-order inertia calculation.

[0008] The turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate are input into a pre-built stress prediction model, and the real-time stress estimate of the rotor is output through the stress prediction model.

[0009] In conjunction with the first aspect, when the instantaneous rate of change of the metal temperature of the high-pressure cylinder exceeds a preset threshold, the historical data moving average is used to replace the current value.

[0010] In conjunction with the first aspect, the radial heat transfer model used in the first step of rotor core temperature calculation satisfies the one-dimensional unsteady-state heat conduction equation: , in, The rotor surface temperature, For time, The thermal diffusivity of the rotor material is... For the rotor radial coordinates.

[0011] In conjunction with the first aspect, the second step of calculating the first-order inertial equation for the rotor core temperature is as follows:

[0012] in, This is the rotor core temperature time lag coefficient.

[0013] In conjunction with the first aspect, the calculation step size of the time differential value is not less than 10 seconds, and it is achieved through the forward difference method: , in, The rate of change of rotor surface temperature. The current temperature of the high-pressure cylinder metal. for The metal temperature of the high-pressure cylinder a few seconds ago.

[0014] In conjunction with the first aspect, the stress prediction model is a three-layer hidden layer neural network, with the hidden layer activation function being ReLU and the output layer being a linear function; The neural network is pre-trained using a training set generated by finite element simulation. The training set covers the turbine inlet warm-up stage when the turbine main steam regulating valve opening ranges from 0% to 100%, and the load ramp-up stage when the unit load increases from the initial load to full load.

[0015] In conjunction with the first aspect, when inputting the turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate into the pre-constructed stress prediction model, the turbine speed needs to be converted into a squared characteristic quantity and normalized. , in, The turbine speed, This is the lower limit of the turbine speed range. Upper limit of turbine speed range This is the normalized squared characteristic of the rotational speed.

[0016] A second aspect of this disclosure provides a system for predicting the start-up stress of a combined cycle turbine rotor, comprising: The data acquisition module is used to acquire turbine speed, main steam regulating valve opening and high-pressure cylinder metal temperature in real time; The signal processing module is used to reduce noise in the high-pressure cylinder metal temperature and calculate the rotor surface temperature and the rate of change of rotor surface temperature. The thermal state analysis module includes a pre-trained radial heat transfer model and a first-order inertial equation, used to calculate the rotor core temperature based on the rotor surface temperature and the rate of change of the rotor surface temperature, and to calculate the rotor core temperature change rate based on the rotor core temperature. The stress prediction module includes a pre-built stress prediction model, which is used to output real-time stress estimates based on turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate.

[0017] In conjunction with the second aspect, the signal processing module includes: The noise reduction unit processes the high-pressure cylinder metal temperature using a Kalman filter algorithm, and when the instantaneous rate of change of the high-pressure cylinder metal temperature exceeds a preset threshold, it uses a historical data moving average to replace the current value. The differential calculation unit performs forward differential calculations with a fixed time step of ≥10 seconds and outputs the rotor surface temperature change rate.

[0018] A third aspect of this disclosure provides an electronic device comprising: One or more processors; A storage unit is used to store one or more programs that, when executed by one or more processors, enable the one or more processors to implement the method for predicting the start-up stress of a combined cycle turbine rotor.

[0019] A fourth aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it can implement the method for predicting the start-up stress of a combined cycle turbine rotor.

[0020] Beneficial Effects: This disclosure provides a method and system for predicting the start-up stress of a combined cycle turbine rotor. During the start-up phase, from 0% to 100% of the turbine's main steam control valve opening, the turbine speed, main steam control valve opening, and high-pressure cylinder metal temperature are acquired in real time. The metal temperature is processed to reduce noise and interference, and the surface temperature and its rate of change are extracted. The rotor core temperature and its rate of change are calculated based on a one-dimensional unsteady-state heat conduction model. The speed, surface temperature, and core temperature data are input into a pre-trained three-layer neural network stress prediction model, which outputs a real-time rotor stress estimate. This disclosure achieves millisecond-level real-time stress prediction through parameter simplification, anti-interference signal processing, and a lightweight neural network architecture, significantly improving the unit's start-up speed and the accuracy of rotor life management. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the combined cycle unit structure according to an embodiment of the present disclosure; Figure 2 This is a flowchart illustrating a method for predicting the impact of turbine differential expansion on the start-up time of a combined cycle unit, according to an embodiment of this disclosure. Figure 3 This is a schematic diagram of the structure of a system for predicting the impact of turbine differential expansion on the start-up time of a combined cycle unit, according to an embodiment of this disclosure. Figure 4 An electronic device according to an embodiment of this disclosure. Detailed Implementation

[0022] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those disclosed herein.

[0023] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0024] It should be understood that although the terms first, second, third, etc., may be used to describe various information in embodiments of this disclosure, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of embodiments of this disclosure, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0025] like Figure 1 The diagram shown is a schematic representation of a combined cycle power unit according to an embodiment. This unit consists of a gas turbine module, a waste heat recovery module, and a steam power module. The gas turbine module follows the Brayton cycle principle and includes an axially connected compressor 1, a combustion chamber 2, and a multi-stage axial flow turbine 3.

[0026] After ambient air is adiabatically compressed by the compressor, it is combusted with natural gas at constant pressure in the combustion chamber to generate high-temperature gas that drives the three blades of the multi-stage axial turbine to do work.

[0027] The exhaust temperature of the multi-stage axial flow turbine 3 remains at a high level of 600℃, and is guided to the waste heat boiler 4 through the transition section for heat recovery.

[0028] The waste heat boiler configured in the unit adopts a three-pressure reheat thermal layout, and integrates a three-stage heat exchange unit consisting of economizer 5, evaporator 6 and superheater 7.

[0029] After being preheated by economizer 5, the feedwater enters the steam drum-evaporator system to complete the vaporization process. The generated saturated steam absorbs the waste heat of the flue gas in the superheater to form superheated steam. This superheated steam is introduced into the high-pressure cylinder of the steam turbine 8 through the main steam pipeline to expand and do work, and together with the gas turbine, drives the generator 9 to output electrical energy.

[0030] After the exhaust steam from turbine 8 is condensed into liquid water in condenser 10, it is pressurized by feedwater pump 11 and returned to waste heat boiler 4, completing the closed thermal cycle.

[0031] Figure 2 This is a flowchart illustrating a method for predicting the start-up stress of a combined cycle turbine rotor according to an embodiment of the present disclosure, including: S101: Obtain turbine parameters during the start-up phase, including turbine speed, main steam control valve opening, and high-pressure cylinder metal temperature. The start-up phase is when the turbine main steam control valve opening ranges from 0% to 100%. For example, during the cold start-up of a combined cycle unit, the operator initiates a stress prediction program through the power plant's distributed control system (DCS). At this time, the turbine's main steam control valve begins to gradually open from its closed state (0% opening), and the system automatically activates the data acquisition thread. The speed signal is obtained from a magnetoresistive sensor installed at the front end of the turbine's high-pressure rotor. This sensor sends the pulse frequency (6 pulses per revolution) generated by the shaft rotation cutting magnetic lines of force to the DCS analog input module, which linearly maps it to a standard signal of 0-3600 rpm via a frequency-to-voltage converter. The main steam control valve opening data is acquired through the LVDT displacement transmitter of the electro-hydraulic actuator. The displacement of its core is proportional to the valve opening, and it outputs a 4-20mA current signal corresponding to the 0-100% opening range.

[0032] The high-pressure cylinder metal temperature monitoring uses two redundant thermocouples, which are embedded in the rim of the first-stage impeller of the high-pressure cylinder. The temperature measuring point is close to the rotor surface. The millivolt signal of the thermocouple is input to the DCS temperature acquisition card after cold junction compensation.

[0033] S102: The high-pressure cylinder metal temperature is subjected to noise reduction processing to obtain the target high-pressure cylinder metal temperature, and the instantaneous value and time derivative value of the target high-pressure cylinder metal temperature are extracted as the rotor surface temperature and the rotor surface temperature change rate. Furthermore, the calculation step size of the time differential value is not less than 10 seconds, and it is achieved through the forward difference method: , in, The rate of change of rotor surface temperature. The current temperature of the high-pressure cylinder metal. for The metal temperature of the high-pressure cylinder a few seconds ago.

[0034] For example, when the raw signal of the high-pressure cylinder metal temperature (e.g., 538.7℃) is input from the AI-03 channel of the DCS, the system first activates the Kalman filter noise reduction unit. This unit initializes the process noise covariance Q=0.05 (characterizing natural temperature fluctuations) and the observation noise covariance R=1.2 (reflecting thermocouple measurement errors), and outputs a smooth value (e.g., 538.2℃) through a predictive-correction iterative cycle. If a temperature jump between adjacent sampling points exceeds the 5℃ / s threshold (e.g., a sudden jump from 538℃ to 580℃), it is determined to be a momentary short circuit of the thermocouple caused by steam condensate interference, and the moving average fault tolerance mechanism is immediately activated: the arithmetic mean of the first 10 valid sampling points (e.g., (536.1+537.3+...+538.5) / 10=537.9℃) is taken to replace the current outlier value, ensuring that the output target high-pressure cylinder metal temperature curve is continuously smooth.

[0035] S103: Input the rotor surface temperature and the rotor surface temperature change rate into the pre-trained radial heat transfer model, output the rotor core temperature, and calculate the rotor core temperature change rate based on the rotor core temperature; Specifically, the rotor surface temperature (e.g., 543.6℃) and its rate of change output by S102 are input into the pre-trained radial heat transfer model. This model is constructed based on a one-dimensional unsteady-state heat conduction equation: , in, The rotor surface temperature, For time, The thermal diffusivity of the rotor material is... For the rotor radial coordinates.

[0036] This model calculates the temperature value (core temperature) at the geometric center of the rotor by analyzing the physical process of the thermal state from the surface to the interior.

[0037] Specifically, based on the actual rotor dimensions (e.g., a rotor with a radius of 350mm), multiple layers of calculation nodes are divided along the radial direction. Initially, the temperature of all nodes is set to the current surface temperature value to ensure matching with the unit's cold start-up conditions.

[0038] The model dynamically simulates the heat transfer process from the rotor surface to the core based on the real-time trend of surface temperature changes (such as detecting a sudden increase in the temperature rise rate from 0.5℃ / s to 1.8℃ / s). This process takes into account the temperature delay phenomenon caused by the thermal inertia of the material (for example, when the surface temperature reaches 400℃, the core temperature only rises to 150℃).

[0039] Based on the core temperature output value of continuous time series, the temperature difference between adjacent moments is calculated at fixed time intervals (typically 10 seconds) and divided by the time step to directly obtain the core temperature change rate. For example: current core temperature: 205.3℃, core temperature 10 seconds ago: 198.6℃, the calculated change rate is: (205.3-198.6) / 10 = 0.67℃ / s.

[0040] S104: Input the turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate into the pre-built stress prediction model, and output the real-time stress estimate of the rotor through the stress prediction model.

[0041] Furthermore, the stress prediction model is a three-layer hidden layer neural network, with the hidden layer activation function being ReLU and the output layer being a linear function; The neural network is pre-trained using a training set generated by finite element simulation, which covers the start-up conditions where the turbine main steam regulating valve opening ranges from 0% to 100%.

[0042] When inputting the turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate into the pre-built stress prediction model, the turbine speed needs to be converted into a squared characteristic quantity and normalized. , in, The turbine speed, This is the lower limit of the turbine speed range. Upper limit of turbine speed range This is the normalized squared characteristic of the rotational speed.

[0043] Specifically, after preparing the thermal state parameters of the rotor surface and core, five types of key data are input into the pre-built stress prediction model: Preprocessed speed signal: The real-time speed (e.g., 2800 rpm) is converted into a dimensionless squared speed characteristic quantity to eliminate the influence of the dimension of mechanical load; Rotor surface temperature and rate of change: noise reduction results from the signal processing module (e.g., 543.6℃) and their derivatives (e.g., 1.2℃ / second); Rotor core temperature and rate of change: The internal thermal state (e.g., 218.4℃) and its changing trend (e.g., 0.83℃ / second) output by the heat transfer model.

[0044] The core operating mechanism of the stress prediction model is as follows: Model architecture: It adopts a three-layer fully connected neural network, which contains a hidden layer consisting of 128-64-32 neurons, and fuses five-dimensional input parameters through nonlinear mapping; Activation mechanism: The hidden layer uses the ReLU function to filter out negative interference, and the output layer directly generates stress physical quantities; Real-time calculation: The time for a single prediction on a general industrial computer is no more than 0.03 seconds, and the results are continuously output at a frequency of 10 Hz.

[0045] Figure 3 This is a schematic diagram of a system for predicting the start-up stress of a combined cycle turbine rotor according to an embodiment of the present disclosure, including: Data acquisition module 210 is used to acquire turbine speed, main steam regulating valve opening and high-pressure cylinder metal temperature in real time; Signal processing module 220 is used to reduce noise in the high-pressure cylinder metal temperature and calculate the rotor surface temperature and the rotor surface temperature change rate. The thermal state analysis module 230 includes a pre-trained radial heat transfer model, which is used to calculate the rotor core temperature based on the rotor surface temperature and the rotor surface temperature change rate, and to calculate the rotor core temperature change rate based on the rotor core temperature. The stress prediction module 240 includes a pre-built stress prediction model for outputting real-time stress estimates based on turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and the rotor core temperature change rate.

[0046] Specifically, in the actual operation of the data acquisition module 210, during unit startup, the magnetoresistive speed sensor captures the high-pressure rotor rotation frequency (e.g., 2800 rpm) in real time and converts it into a standard electrical signal for transmission to the control cabinet. The displacement sensor of the electro-hydraulic actuator senses the main steam regulating valve opening (e.g., 85% opening) via a mechanical linkage, generating a 4-20mA current signal. Two K-type thermocouples are embedded at the root of the first-stage impeller of the high-pressure cylinder (see...). Figure 3 (Installation location), directly measuring the rotor surface temperature (e.g., 538℃). All signals are connected to the power plant's DCS system via shielded cables, and data timestamp deviation is ensured to be less than 5 milliseconds through hardware clock synchronization.

[0047] The core function of the signal processing module 220 is implemented when the raw metal temperature signal (e.g., 539.2℃) enters this module: The noise reduction unit activates the average filtering technology: Automatically identify and suppress thermocouple circuit interference (such as random fluctuations of 0.5℃). If a sudden temperature change exceeding 5°C / second is detected (e.g., a signal jump caused by steam condensation), immediately switch to the historical data moving average mode (taking the average of 10 data sets from the previous second). The differential calculation unit calculates the temperature rise rate using a fixed time window: Basic mode: Compare the current temperature value with the temperature value 10 seconds ago (e.g., (540.1-532.3) / 10=0.78℃ / second). Fast Response Mode: When the temperature rise rate is consistently higher than 3℃ / second, the step size is automatically shortened to 5 seconds to improve sensitivity.

[0048] The physical mechanism of the thermal state analysis module 230 involves inputting the surface temperature and rate of change (e.g., 540℃, 0.78℃ / second) into the preloaded radial heat transfer model and first-order inertial equations: Model kernel: Based on the thermal conductivity characteristics of 12Cr steel rotor material (thermal diffusivity 11.5 mm² / s), the model simulates the heat transfer process from the surface to the core. Real-time calculation: The rotor radius (350mm) is divided into 50 layers of calculation nodes. The boundary conditions are dynamically updated according to the surface temperature, and the rotor geometric center temperature is output (e.g., the core temperature is 210℃ when the surface temperature is 540℃).

[0049] The output rotor core temperature is input into the inertial equation to obtain the time-delayed output parameters.

[0050] Change rate generation: Calculate the core temperature difference every 10 seconds (e.g., 210℃ this time - 205℃ last time = 5℃), and obtain a change rate of 0.5℃ / second.

[0051] Industrial deployment of the stress prediction module 240, example of five-dimensional parameter input: Rotational speed characteristic (0.871), surface temperature (538℃), surface temperature rise rate (0.95℃ / second), core temperature (205℃), core temperature rise rate (0.77℃ / second).

[0052] Neural network execution flow: The input layer receives five-dimensional parameters (automatically normalized), the three hidden layers perform nonlinear transformation (128→64→32 neurons), and the output layer generates the stress value at the root of the first stage impeller of the high-pressure cylinder (e.g., 203.4 MPa).

[0053] Hardware support: On a standard industrial computer (Intel i5 processor), a single calculation takes 0.03 seconds, with 10 predictions updated per second.

[0054] Electronic device 400 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 400 may include, but is not limited to, processor 401 and memory 402. Those skilled in the art will understand that... Figure 4This is merely an example of electronic device 400 and does not constitute a limitation on electronic device 400. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0055] Processor 401 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0056] The memory 402 can be an internal storage unit of the electronic device 400, such as a hard disk or memory of the electronic device 300. The memory 402 can also be an external storage device of the electronic device 400, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 400. Furthermore, the memory 402 can include both internal and external storage units of the electronic device 400. The memory 402 is used to store the computer program 403 and other programs and data required by the electronic device. The memory 402 can also be used to temporarily store data that has been output or will be output.

[0057] The above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit it. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be included within the protection scope of this disclosure.

Claims

1. A method for predicting the start-up stress of a combined cycle turbine rotor, characterized in that, Includes the following steps: The turbine parameters during the start-up phase are obtained, including turbine speed, main steam control valve opening, and high-pressure cylinder metal temperature. The start-up phase is defined as the turbine main steam control valve opening ranging from 0% to 100%. The target high-pressure cylinder metal temperature is obtained by performing noise reduction processing on the metal temperature of the high-pressure cylinder. The instantaneous value and time derivative value of the target high-pressure cylinder metal temperature are extracted as the rotor surface temperature and the rotor surface temperature change rate. The rotor surface temperature and the rotor surface temperature change rate are input into a pre-trained radial heat transfer model, the rotor core temperature is output, and the rotor core temperature change rate is calculated based on the rotor core temperature. The turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate are input into a pre-built stress prediction model, and the real-time stress estimate of the rotor is output through the stress prediction model.

2. The method according to claim 1, characterized in that, The noise reduction process employs a Kalman filter algorithm, and when the instantaneous rate of change of the metal temperature of the high-pressure cylinder exceeds a preset threshold, the historical data moving average is used to replace the current value.

3. The method according to claim 1, characterized in that, The radial heat transfer model satisfies a one-dimensional unsteady heat conduction equation: ， in, The rotor surface temperature, For time, The thermal diffusivity of the rotor material is... For the rotor radial coordinates.

4. The method according to claim 1, characterized in that, The calculation step size for the time differential value is no less than 10 seconds, and it is achieved using the forward difference method: ， in, The rate of change of rotor surface temperature. The current temperature of the high-pressure cylinder metal. for The metal temperature of the high-pressure cylinder a few seconds ago.

5. The method according to claim 1, characterized in that, The stress prediction model is a three-layer hidden layer neural network, with ReLU activation function in the hidden layers and a linear function in the output layer; The neural network is pre-trained using a training set generated by finite element simulation, which covers the start-up conditions where the turbine main steam regulating valve opening ranges from 0% to 100%.

6. The method according to claim 1, characterized in that, When inputting the turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate into the pre-built stress prediction model, the turbine speed needs to be converted into a squared characteristic quantity and normalized. , in, The turbine speed, This is the lower limit of the turbine speed range. Upper limit of turbine speed range This is the normalized squared characteristic of the rotational speed.

7. A system for predicting the start-up stress of a combined cycle turbine rotor, characterized in that, include: The data acquisition module is used to acquire turbine speed, main steam regulating valve opening and high-pressure cylinder metal temperature in real time; The signal processing module is used to reduce noise in the high-pressure cylinder metal temperature and calculate the rotor surface temperature and the rate of change of rotor surface temperature. The thermal state analysis module includes a pre-trained radial heat transfer model, which is used to calculate the rotor core temperature based on the rotor surface temperature and the rate of change of the rotor surface temperature, and to calculate the rotor core temperature change rate based on the rotor core temperature. The stress prediction module includes a pre-built stress prediction model, which is used to output real-time stress estimates based on turbine speed, rotor surface temperature, rotor surface temperature change rate, rotor core temperature, and rotor core temperature change rate.

8. The system according to claim 7, characterized in that, The signal processing module includes: The noise reduction unit processes the high-pressure cylinder metal temperature using a Kalman filter algorithm, and when the instantaneous rate of change of the high-pressure cylinder metal temperature exceeds a preset threshold, it uses a historical data moving average to replace the current value. The differential calculation unit performs forward differential calculations with a fixed time step of ≥10 seconds and outputs the rotor surface temperature change rate.

9. An electronic device, characterized in that, include: One or more processors; A storage unit for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the method for predicting the start-up stress of a combined cycle turbine rotor according to claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it can implement the method for predicting the start-up stress of a combined cycle turbine rotor as described in claims 1-6.

Images