A steam turbine cold end optimization method and system based on mechanism and data hybrid driving modeling

Abstract

The application discloses a steam turbine cold end optimization method and system based on mechanism and data hybrid driving modeling, and comprises the following steps: establishing a steam turbine cold end system mechanism model, and identifying to-be-identified characteristic parameters of each model; steam turbine cold end system data acquisition; characteristic parameter historical data preprocessing and steady state judgment; steady state data working condition division; performing working condition division on the steady state data of the cold end system to obtain a steady state working condition cluster of the cold end system; establishing a characteristic identification data model with the operating parameters of the cold end system and the to-be-identified characteristic parameters of the mechanism model as inputs and the condenser pressure as output; mechanism model characteristic regression; establishing a regression prediction model of the steam turbine exhaust steam parameters based on all the steady state working conditions of the cold end system; mechanism model characteristic updating; steam turbine cold end optimization model establishment; and cold end optimization model solving. The application establishes a mechanism and data hybrid driving optimization model in the full working condition range of the steam turbine cold end, realizes optimization guidance of steam turbine cold end operation adjustment, and reduces the heat consumption of the steam turbine.

Description

TECHNICAL FIELD

[0001] The application relates to a steam turbine cold end optimization method and system, in particular to a steam turbine cold end optimization method and system based on mechanism and data hybrid driving modeling. BACKGROUND

[0002] As an important part of a thermal power generating unit, the heat loss of a steam turbine cold end system accounts for the largest proportion in a thermal system and is the part with the largest energy saving potential. Through cold end optimization, the power generation cost can be reduced, which is of great significance for improving the economy of unit operation and reducing energy consumption.

[0003] Traditional steam turbine cold end operation optimization methods are mainly based on thermal test and condenser variable condition calculation model, but the optimization results of these methods are not necessarily accurate. The main reason is that after the unit has been operated for a period of time, the actual operation performance of the unit will change due to condenser heat exchange surface pollution, equipment aging and other reasons, deviating from the test results and calculation model, and cannot reflect the actual operation of the unit in real time. At present, the operation of the steam turbine cold end system mainly relies on the experience of the operation personnel for adjustment, and lacks real-time adjustment and economy. SUMMARY

[0004] The purpose of the application is to provide a steam turbine cold end optimization method and system based on mechanism and data hybrid driving modeling, to realize the optimization guidance of steam turbine cold end operation adjustment, to reduce the heat consumption of the steam turbine, and to improve the overall economic benefit of the steam turbine generating unit.

[0005] The technical scheme of the application comprises the following steps:

[0006] Establishing a mechanism model of the steam turbine cold end system, and identifying the characteristic parameters of each model to be identified;

[0007] Data acquisition of the steam turbine cold end system: collecting relevant process parameters, and the collected data are used for historical data working condition extraction and real-time optimization calculation;

[0008] Characteristic parameter historical data preprocessing and steady state judgment: extracting the recent operation data of the cold end system from the historical data, judging the steady state working condition of the preprocessed working condition data, and retaining the steady state working condition data as the data basis for characteristic identification of the mechanism model of the cold end system;

[0009] Steady state data working condition division: dividing the steady state data of the cold end system into working conditions, and obtaining the steady state working condition cluster of the cold end system after the working condition division is completed;

[0010] Intelligent identification of the characteristic parameters of the mechanism model: in each working condition cluster of the cold end system, a characteristic identification data model is established with the cold end system operation parameters and the characteristic parameters of the mechanism model to be identified as inputs and the condenser pressure as output;

[0011] Mechanism model characteristic regression: After completing the identification of the characteristics of each working condition cluster, the C working condition parameters at the center of each working condition cluster are used as variables to perform regression fitting of the mechanism model characteristic parameters between different working condition clusters;

[0012] Steam turbine exhaust parameter modeling: Based on all steady-state operating conditions of the cold-end system, a regression prediction model for steam turbine exhaust parameters is established;

[0013] Mechanism model characteristic update: Based on the mechanism model of characteristic identification and the turbine exhaust parameter prediction model, a turbine cold end hybrid model is constructed. The hybrid model prediction error is used as the criterion to determine whether the mechanism model characteristic parameters should be updated;

[0014] Establishment of a turbine cold-end optimization model: Based on a characteristic identification mechanism model and a turbine exhaust parameter prediction model, the turbine cold-end optimization model is established with the goal of maximizing the turbine net power after deducting the incremental power consumption of the circulating pump.

[0015] Solving the cold-end optimization model: Utilizing an intelligent optimization algorithm, the cold-end optimization model is optimized and solved under specific constraints to obtain the recommended operating mode for the circulating pump under the turbine's real-time operating conditions. The optimization result, X, serves as the recommended operating mode or frequency for the circulating pump in the turbine's cold-end system.

[0016] The steam turbine cold end system mechanism model includes a steam turbine slight power increase characteristic model, a condenser characteristic model, and a circulating water pump power consumption characteristic model.

[0017] The turbine power increase characteristic model is:

[0018] ΔP t =f(P t ,p k )

[0019] Where ΔP t is the turbine power increase, P t is the turbine active power, p k It is the exhaust back pressure of the steam turbine.

[0020] The condenser characteristic model includes:

[0021] Saturation temperature of turbine exhaust steam at condenser pressure t s for:

[0022] t s =t w1 +Δt+δt

[0023] Among them, t w1 is the condenser circulating water inlet temperature, °C; Δt is the condenser circulating water temperature rise, °C; δt is the condenser heat transfer end difference, °C;

[0024] The circulating water temperature rise Δt is:

[0025]

[0026] wherein, D c , D w are respectively steam turbine exhaust flow and circulating water flow, t / h; h c , h ′ c are respectively specific enthalpy of steam in condenser and specific enthalpy of condensate, kJ / kg;

[0027] The condenser heat transfer end difference δt is:

[0028]

[0029] wherein, A c is condenser heat transfer area, m 2 ; K is condenser overall heat transfer coefficient, kJ / (m 2 ·h·K).

[0030] The to-be-identified characteristic parameters include condenser heat transfer coefficient, steam turbine micro-increment power characteristic parameter, circulating pump characteristic parameter, and a total of P to-be-identified parameters.

[0031] The working condition cluster structure is as follows:

[0032]

[0033] The intelligent identification of the mechanism model characteristic parameters is specifically: using an intelligent optimization algorithm to take the deviation of the condenser pressure data model output value and the actual value as the target, inversely solving the mechanism model characteristic parameters, and obtaining the cold end system mechanism model characteristic parameters on the M working condition clusters:

[0034]

[0035] During the mechanism model characteristic updating process, if the model error is greater than a threshold value, the process of data collection of the steam turbine cold end system to the intelligent identification of the mechanism model characteristic parameters is repeated, and the mechanism model characteristic parameters are updated and regression fitted.

[0036] The steam turbine cold end optimization model is: max F = ∑P t - ∑P pump .

[0037] A steam turbine cold end optimization system based on mechanism and data hybrid driving modeling includes:

[0038] A steam turbine cold end system mechanism model establishment module: establishing a steam turbine cold end system mechanism model, and identifying to-be-identified characteristic parameters of each model;

[0039] Steam turbine cold end system data acquisition module: collect steam turbine cold end system related process parameters;

[0040] Characteristic parameter historical data preprocessing and steady state judgment module: extract the historical data of the cold end system for a certain period, preprocess the working condition data, and judge the steady state working condition of the preprocessed characteristic parameter working condition data;

[0041] Steady state data working condition division module: for the steady state data of the steam turbine cold end system, the working condition cluster for the characteristic identification of the mechanism model of the cold end system is formed;

[0042] Intelligent identification module of mechanism model characteristic parameters: for each working condition cluster generated by working condition division, the data model of the cold end system operating parameter and characteristic parameter is constructed;

[0043] Mechanism model characteristic regression module: after completing the characteristic identification of each working condition cluster, the regression fitting of the mechanism model characteristic parameters is carried out based on the center working condition parameters of each working condition cluster;

[0044] Steam turbine exhaust parameter modeling module: based on all steady state working conditions of the cold end system, the regression prediction model of the steam turbine exhaust parameter is established;

[0045] Mechanism model characteristic updating module: based on the mechanism model of the characteristic identification and the prediction model of the steam turbine exhaust parameter, the hybrid model of the steam turbine cold end is constructed, and whether the mechanism model characteristic is updated is determined according to the hybrid model prediction error;

[0046] Steam turbine cold end optimization model establishment module: based on the mechanism model of the characteristic identification and the prediction model of the steam turbine exhaust parameter, the steam turbine cold end optimization model is established with the maximum net power of the steam turbine excluding the pump power consumption increment as the target;

[0047] Cold end optimization model solving module: the intelligent optimization algorithm is used to optimize and solve the cold end optimization model under certain constraints, and the pump operation mode of the steam turbine under real-time working condition is obtained.

[0048] Beneficial effects: the optimization model of the mechanism and data hybrid driving of the whole working condition range of the steam turbine cold end is established, the optimization guidance of the steam turbine cold end operation adjustment is realized, the steam turbine heat consumption is reduced, and the total economic benefit of the steam turbine generator unit is improved. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 The flowchart of the present application;

[0050] Figure 2 The steam turbine cold end optimization technical roadmap based on the mechanism and data hybrid driving modeling of the present application. DETAILED DESCRIPTION

[0051] The present application will be further described below with reference to the drawings.

[0052] Embodiment 1

[0053] As shown in Figure 1 and Figure 2 , the steam turbine cold end optimization method based on mechanism and data hybrid driving modeling of the present embodiment optimizes the steam turbine cold end operation mode according to the real-time operation condition of the steam turbine, guides the steam turbine operation personnel to optimize and adjust the cold end system, and achieves the purpose of improving the total economic benefit of the steam turbine generator unit, which specifically includes the following steps:

[0054] Step one, establish the mechanism model of the steam turbine cold end system, including the steam turbine micro-increment power characteristic, the condenser characteristic, the circulating water pump power consumption characteristic and other models, and determine the to-be-identified characteristic parameters of each model.

[0055] Steam turbine micro-increment power characteristic:

[0056] ΔP t =f(P t ,p k )

[0057] Wherein, ΔP t is the steam turbine micro-increment power, P t is the steam turbine active power, and p k is the steam turbine exhaust back pressure.

[0058] Condenser characteristic:

[0059] The saturation temperature t s of the steam turbine exhaust under the condenser pressure can be expressed as:

[0060] t s =t w1 +Δt+δt

[0061] Wherein, t w1 is the condenser circulating water inlet temperature, ℃; Δt is the condenser circulating water temperature rise, ℃; and δt is the condenser heat transfer approach, ℃.

[0062] The circulating water temperature rise Δt can be expressed as:

[0063]

[0064] Wherein, D c , D w are the steam turbine exhaust flow and the circulating water flow, respectively, t / h; h c , h′ c are the steam specific enthalpy and the condensate specific enthalpy in the condenser, respectively, kJ / kg.

[0065] The condenser heat transfer approach δt can be expressed as:

[0066]

[0067] Among them, A c is the condenser heat exchange area, m 2 ;K is the overall heat transfer coefficient of the condenser, kJ / (m 2 ·h·K).

[0068] The characteristic parameters to be identified in the cold end system mechanism model are the condenser heat transfer coefficient, the turbine power increment characteristic parameters, the circulating pump characteristic parameters, etc., with a total of P parameters to be identified.

[0069] Step 2: Data collection of the cold-end system of the steam turbine. Depending on the area where the system is deployed, relevant process parameters are collected from the unit DCS or SIS system through the interface. These parameters include generator active power, main steam pressure, main steam temperature, high-pressure control valve opening feedback, cold reheat steam pressure, reheat steam pressure, reheat steam temperature, medium-pressure control valve opening feedback, extraction pressure and temperature of each section, condenser pressure, condenser water level, condenser hot well temperature, circulating pump current and frequency, circulating water main pressure, circulating water main flow, condenser inlet and outlet water pressure and temperature, and other parameters. The collected data is used for historical data operating condition extraction and real-time optimization calculation.

[0070] Step 3: Preprocessing historical data of characteristic parameters and determining steady-state conditions. Recent operating data for the cold-end system is extracted from the historical data, and data cleaning is performed to address outliers and missing values ​​in the operating data. The preprocessed operating data is then subjected to a steady-state determination based on the R statistical test. Data that meets the threshold conditions is considered steady-state cold-end system data, and this data is retained as the data basis for identifying the characteristics of the cold-end system mechanism model.

[0071] Step 4: Steady-state data working condition division. Perform fuzzy C-means clustering on the cold-end system steady-state data. After the working condition division is completed, M cold-end system steady-state working condition clusters are obtained. The working condition cluster structure is as follows:

[0072]

[0073] Step 5: Intelligently identify the characteristic parameters of the mechanism model. For each operating condition cluster of the cold-end system, a characteristic identification data model is established, with the cold-end system operating parameters and the characteristic parameters of the mechanism model to be identified as input, and the condenser pressure as the output. This results in M ​​characteristic identification data models. Using an intelligent optimization algorithm, the deviation between the output value of the condenser pressure data model and the actual value is used as the target to inversely solve the characteristic parameters of the mechanism model. This yields M sets of cold-end system mechanism model characteristic parameters, including condenser heat transfer characteristics, turbine power increase characteristics, and circulating pump characteristics, for each of the M operating condition clusters. This allows for intelligent identification of multiple operating condition clusters of cold-end system mechanism model characteristic parameters.

[0074]

[0075] Step six, mechanism model characteristic regression, after completing the characteristic identification of each operating condition cluster, taking the operating condition parameters of each operating condition cluster center C, such as the steam turbine active power, the circulating water inlet temperature, etc. as variables, the regression fitting of the characteristic parameters of the mechanism model between different operating condition clusters is carried out, and the regression results of the characteristic parameters are as follows, so as to improve the operating condition applicable range and model accuracy of the steam turbine cold end mechanism model.

[0076]

[0077] Step seven, steam turbine exhaust parameter modeling, for the calculation of the steam turbine exhaust enthalpy and exhaust flow, based on all the steady operating conditions of the cold end system, taking the steam turbine main steam parameters, the valve opening degree, the extraction steam parameters as inputs, and taking the exhaust enthalpy and the exhaust flow as outputs, the regression prediction model MODEL_TURBINE of the steam turbine exhaust parameters is established, and the calculation of the exhaust parameters under the real-time operating condition of the steam turbine can be carried out.

[0078] Step eight, mechanism model characteristic updating, based on the mechanism model of the characteristic identification and the prediction model of the steam turbine exhaust parameters, the mixed model of the steam turbine cold end is constructed, and the mixed model prediction error (including the mean square error (MSE), the root mean square error (RMSE), and the mean absolute error (MAE)) is used as the criterion to determine whether the mechanism model characteristic parameters are updated, if the model error is greater than the threshold value, i.e. |parameter predict -parameter real |>ε, then steps two to five are repeated, and the updating and regression fitting of the mechanism model characteristic parameters such as the condenser heat transfer characteristic, the steam turbine micro-increment power characteristic, and the circulating pump characteristic are carried out.

[0079] Step nine, establishment of the steam turbine cold end optimization model, based on the mechanism model of the characteristic identification and the prediction model of the steam turbine exhaust parameters, the steam turbine net power maximum after deducting the circulating pump power consumption increment is taken as the target to establish the steam turbine cold end optimization model.

[0080] maxF =∑P t -∑P pump

[0081] Step ten, solution of the cold end optimization model: the intelligent optimization algorithm is used to solve the optimization of the cold end optimization model under certain constraints, and the recommended operating mode of the circulating pump under the real-time operating condition of the steam turbine is obtained, and the optimization result X can be used as the recommended value of the operating mode or the operating frequency of the steam turbine cold end system circulating pump, so as to reduce the heat consumption of the steam turbine and improve the overall economic benefit of the steam turbine generator unit.

[0082] X = [run_flag1,…,run_flag num ,n1,…,n num ]

[0083] The intelligent optimization algorithm includes a constrained particle swarm optimization algorithm, a grey wolf algorithm and the like, and the constraint processing method includes a penalty function method, a C-constraint processing method and a hybrid method.

[0084] Embodiment 2

[0085] The steam turbine cold end optimization system based on the mechanism and data hybrid driving modeling of the embodiment includes:

[0086] A steam turbine cold end system mechanism model establishing module: a steam turbine cold end system mechanism model is established, including a steam turbine micro-increment power characteristic, a condenser characteristic, a circulating water pump power consumption characteristic and the like, and the to-be-identified characteristic parameters of each model are determined;

[0087] A steam turbine cold end system data acquisition module: steam turbine cold end system related process parameters are acquired from a unit DCS or SIS system through an interface, including a generator active power, a main steam pressure, a main steam temperature, a high regulating valve opening degree feedback, a cold reheat steam pressure, a reheat steam pressure, a reheat steam temperature, a medium regulating valve opening degree feedback, each section steam extraction pressure and temperature, a condenser pressure, a condenser water level, a condenser hot well temperature, a circulating water pump current and frequency, a circulating water main pipe pressure, a circulating water main pipe flow, condenser inlet and outlet water pressure and temperature and the like, and the acquired data is saved as historical data, so as to perform historical data processing and working condition identification;

[0088] A characteristic parameter historical data preprocessing and steady state judgment module: historical data of a certain period of the cold end system is extracted, working condition data preprocessing is performed, for the preprocessed characteristic parameter working condition data, steady state working condition judgment is performed based on a certain algorithm rule, and steady state working condition data of the steam turbine cold end system is extracted;

[0089] A steady state data working condition division module: for the steady state data of the steam turbine cold end system, working condition division of the steam turbine cold end data is realized based on a classification algorithm, and a working condition cluster for cold end system mechanism model characteristic identification is formed;

[0090] An intelligent identification module of mechanism model characteristic parameters: for each working condition cluster generated by working condition division, a data model of cold end system operating parameters and characteristic parameters is constructed, an intelligent optimization algorithm is used for inversion of to-be-identified parameters, and intelligent identification of the mechanism model characteristic parameters of the cold end system in a single working condition cluster is realized;

[0091] A mechanism model characteristic regression module: after the characteristic identification of each working condition cluster is completed, regression fitting of the mechanism model characteristic parameters is performed based on the central working condition parameters of each working condition cluster, the model parameters of the mechanism model are applied to a wider load range, and the model precision and working condition application range of the steam turbine cold end mechanism model are improved;

[0092] A steam turbine exhaust parameter modeling module: for the calculation of the exhaust enthalpy and flow rate of the steam turbine, based on the overall steady state working condition of the cold end system, with the main steam parameters, the gate opening and the extraction parameters as inputs and the exhaust enthalpy and flow rate as outputs, a regression prediction model of the exhaust parameters of the steam turbine is established, and the exhaust parameters of the steam turbine under the working condition can be calculated;

[0093] A mechanism model characteristic updating module: based on the mechanism model and the exhaust parameter prediction model of the steam turbine, a mixed model of the cold end of the steam turbine is constructed, and the mixed model prediction error is used as a criterion to determine whether the characteristics of the mechanism model are updated, if the model error is greater than a threshold value, the process of intelligent identification of the mechanism model characteristics is repeated from the data collection of the cold end system of the steam turbine to the mechanism model characteristics, and the updating and regression fitting of the mechanism model characteristics such as the heat transfer coefficient of the condenser, the micro-increment power characteristics of the steam turbine and the characteristics of the circulating pump are performed;

[0094] A steam turbine cold end optimization model establishing module: based on the mechanism model and the exhaust parameter prediction model of the steam turbine, a steam turbine cold end optimization model is established with the maximum net power of the steam turbine after deducting the increment of the circulating pump power consumption as the target;

[0095] A cold end optimization model solving module: an intelligent optimization algorithm is used to solve the optimization of the cold end optimization model under certain constraints, and the circulating pump operation mode under the real-time working condition of the steam turbine is obtained, and the optimization result can be used as the recommended value of the operation mode or the operation frequency of the circulating pump of the cold end system of the steam turbine, so as to reduce the heat consumption of the steam turbine and improve the total economic benefit of the steam turbine generator unit.

[0096] According to the real-time working condition of the steam turbine, the cold end operation mode of the steam turbine is optimized, the operation personnel of the steam turbine are guided to optimize and adjust the cold end system, and the total economic benefit of the steam turbine generator unit is improved.

Claims

1. A steam turbine cold end optimization method based on mechanism and data hybrid driven modeling, characterized by: The following steps are involved: Establish a mechanism model of the steam turbine cold end system and clarify the characteristic parameters to be identified for each model; Data collection of the steam turbine cold end system: collects relevant process parameters. The collected data is used for historical data condition extraction and real-time optimization calculation; Preprocessing of historical data of characteristic parameters and steady-state judgment: Extract recent operating data of the cold-end system from historical data, make steady-state judgment on the preprocessed operating data, and retain the steady-state operating data as the data basis for identifying the characteristics of the cold-end system mechanism model; Steady-state data working condition division: The working condition of the cold-end system steady-state data is divided. After the working condition division is completed, the cold-end system steady-state working condition cluster is obtained; Intelligent identification of characteristic parameters of the mechanism model: For each operating condition cluster of the cold-end system, a characteristic identification data model is established with the cold-end system operating parameters and the characteristic parameters of the mechanism model to be identified as input, and the condenser pressure as output; Mechanism model characteristic regression: After completing the identification of the characteristics of each working condition cluster, the C working condition parameters at the center of each working condition cluster are used as variables to perform regression fitting of the mechanism model characteristic parameters between different working condition clusters; Steam turbine exhaust parameter modeling: Based on all steady-state operating conditions of the cold-end system, a regression prediction model for steam turbine exhaust parameters is established; Mechanism model characteristic update: Based on the mechanism model of characteristic identification and the turbine exhaust parameter prediction model, a turbine cold end hybrid model is constructed. The hybrid model prediction error is used as the criterion to determine whether the mechanism model characteristic parameters should be updated; Establishment of a turbine cold-end optimization model: Based on a characteristic identification mechanism model and a turbine exhaust parameter prediction model, the turbine cold-end optimization model is established with the goal of maximizing the turbine net power after deducting the incremental power consumption of the circulating pump. Solving the cold-end optimization model: Using intelligent optimization algorithms, the cold-end optimization model is optimized and solved under specific constraints to obtain the recommended operating mode of the circulating pump under the real-time operating conditions of the steam turbine.

2. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 1 is characterized in that: The steam turbine cold end system mechanism model includes a steam turbine slight power increase characteristic model, a condenser characteristic model, and a circulating water pump power consumption characteristic model.

3. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 2 is characterized in that: The turbine power increase characteristic model is: ΔP t =f(P t ,p k ) Where ΔP t is the slight increase in turbine power, P t is the turbine active power, p k It is the exhaust back pressure of the steam turbine.

4. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 2 is characterized in that: The condenser characteristic model includes: Saturation temperature of turbine exhaust steam at condenser pressure t s for: t s =t w1 +Δt+δt Among them, t w1 is the condenser circulating water inlet temperature, °C; Δt is the condenser circulating water temperature rise, °C; δt is the condenser heat transfer end difference, °C; The circulating water temperature rise Δt is: Among them, D c , D w Turbine exhaust flow and circulating water flow, t / h; h c ,h ′ c are the specific enthalpy of steam and condensate in the condenser, kJ / kg; The heat transfer end difference δt of the condenser is: Among them, A c is the condenser heat exchange area, m 2 ;K is the overall heat transfer coefficient of the condenser, kJ / (m 2 ·h·K).

5. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 1 is characterized in that: The characteristic parameters to be identified include the condenser heat transfer coefficient, the turbine power increment characteristic parameters, and the circulating pump characteristic parameters, a total of P parameters to be identified.

6. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 1 is characterized in that: The working condition cluster structure is as follows:

7. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 6 is characterized in that: The intelligent identification of the characteristic parameters of the mechanism model is specifically as follows: using an intelligent optimization algorithm to take the deviation between the output value of the condenser pressure data model and the actual value as the target, perform an inverse solution of the characteristic parameters of the mechanism model, and obtain the characteristic parameters of the cold end system mechanism model on M operating condition clusters:

8. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 1 is characterized in that: During the updating process of the mechanism model characteristics, if the model error is greater than the threshold, the process of intelligently identifying the mechanism model characteristic parameters from the steam turbine cold end system data collection is repeated to update and regress the mechanism model characteristic parameters.

9. The method for optimizing the cold end of a steam turbine based on hybrid mechanism- and data-driven modeling according to claim 3 is characterized in that: The turbine cold end optimization model is established with the goal of maximizing the net power of the turbine after deducting the incremental power consumption of the circulating pump. The turbine cold end optimization model is: maxF = ∑P t -∑P pump .

10. A steam turbine cold end optimization system based on mechanism and data hybrid driven modeling, characterized by: include: Steam turbine cold end system mechanism model establishment module: establishes the steam turbine cold end system mechanism model and clarifies the characteristic parameters to be identified for each model; Steam turbine cold end system data acquisition module: collects relevant process parameters of the steam turbine cold end system; Characteristic parameter historical data preprocessing and steady-state judgment module: extracts historical data of the cold end system for a certain period of time, performs operating condition data preprocessing, and performs steady-state operating condition judgment on the preprocessed characteristic parameter operating condition data; Steady-state data operating condition division module: divides the steady-state data of the steam turbine cold-end system into operating conditions to form operating condition clusters for cold-end system mechanism model characteristic identification; Intelligent identification module of characteristic parameters of mechanism model: for each operating condition cluster generated by operating condition division, a data model of cold end system operating parameters and characteristic parameters is constructed; Mechanism model characteristic regression module: After completing the identification of the characteristics of each working condition cluster, the regression fitting of the mechanism model characteristic parameters is performed based on the central working condition parameters of each working condition cluster; Steam turbine exhaust parameter modeling module: Based on all steady-state operating conditions of the cold-end system, a regression prediction model for steam turbine exhaust parameters is established; Mechanism model characteristic update module: Based on the mechanism model of characteristic identification and the turbine exhaust parameter prediction model, a turbine cold end hybrid model is constructed. The hybrid model prediction error is used as the criterion to determine whether the mechanism model characteristics should be updated; Steam turbine cold-end optimization model establishment module: Based on the characteristic identification mechanism model and the steam turbine exhaust parameter prediction model, the steam turbine cold-end optimization model is established with the goal of maximizing the steam turbine net power after deducting the incremental power consumption of the circulating pump; Cold end optimization model solving module: uses intelligent optimization algorithms to optimize and solve the cold end optimization model under specific constraints, and obtains the circulation pump operation mode under the real-time working conditions of the steam turbine.

Images