Thermal power plant AGC load OLEBS coordination control method

Abstract

The invention discloses a thermal power plant AGC load OLEBS coordination control method, and the method comprises the specific steps: employing a coordination control system with advanced control features such as biLSTM regression prediction control, adaptive control, and a recurrent neural network, and achieving the prediction of the heat energy of a boiler, so as to move the coal feed in advance, effectively compensate the inertia of the boiler, and improve the control precision of the boiler. The high-precision digital twinborn body of the specified typical 350MW supercritical cogeneration unit is realized; the unit digital twinborn body has certain reconfigurability, can simulate different unit working conditions and business states through convenient data driving and model adjustment, and can store and call the working conditions and business states respectively; a DCS logic configuration environment which is actually the same as that of the unit is realized, so that operation processes such as unit coordination control strategy and parameter optimization, test and demonstration can be conveniently carried out; an AGC coordination control strategy and logic configuration which have advanced technical characteristics and are adaptive to a specified typical 350MW supercritical cogeneration unit are designed and developed, and optimization, testing and system verification are carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of AGC load of thermal power plants, and in particular to an OL_EBS coordinated control method of AGC load of thermal power plants. Background Art

[0002] The key difficulty is to realize the online twin-type incentive virtual DCS logic configuration environment similar to the actual state of the unit, carry out the coordinated control strategy and parameter optimization, testing, demonstration and other operation processes of the unit, and establish the ol-EBs online energy balance self-calibration system. By establishing the positive and negative balance boiler thermal model, the unit parameter big data history model, and the main response strategy when establishing the optimized coordinated control system through advanced control features such as biLSTM regression predictive control, adaptive control, and recurrent neural network, it is necessary to develop big data model generation and management tools to form efficient, simple, and easy-to-use control configuration tools. The OL-EBs online energy balance intelligent control system currently has functions of coordinated control optimization and testing of units under multiple operating conditions and business formats, as well as digital model generation and system model management. However, the existing technology has many shortcomings: first, the calibration of the AGC adjustment system is slow, such as the traditional BTU calibration method; second, the boiler fuel fluctuates rapidly, which requires the adjustment method to have a higher rapid matching capability; third: the thermal power units need to be flexibly peaked, and the fuel, feed water, and air supply must be quickly adjusted according to the calorific value and carbon content of the fuel. In summary, these problems limit the optimized operation and flexible scheduling capabilities of thermal power plants. Summary of the Invention

[0003] The purpose of the present invention is to solve at least one of the technical problems existing in the prior art, and to provide a coordinated control method for the AGC load OL_EBS of a thermal power plant to realize a high-precision digital twin of a specified typical 350MW supercritical cogeneration unit; the digital twin of the unit has a certain degree of reconfigurability, and through convenient data driving and model adjustment, it can simulate different unit operating conditions and business modes, and can be stored and called separately; to realize a DCS logical configuration environment that is identical to the actual unit, which is convenient for carrying out unit coordinated control strategies and parameter optimization, testing, demonstration and other operational processes; to design and develop an AGC coordinated control strategy and logical configuration with advanced technical features that is suitable for a specified typical 350MW supercritical cogeneration unit, and has been optimized, tested and system verified.

[0004] The present invention also provides the above-mentioned AGC load OL_EBS coordinated control method of a thermal power plant, comprising the following steps:

[0005] S1. A coordinated control system that uses advanced control features such as predictive control, adaptive control, and neural networks can predict the "heat energy" of the boiler;

[0006] S2, thus setting the coal feed rate, water feed rate and water-coal ratio in advance, effectively compensating for the boiler's inertia, and ensuring the unit has a fast load response speed and smooth pressure changes;

[0007] S3. Improve the frequency regulation performance of the automatic generation control (AGC) of thermal power units and enhance the AGC frequency regulation service capability of the AGC comprehensive frequency regulation performance index (K);

[0008] S4. Realize an online high-precision digital twin of a designated typical 350MW supercritical cogeneration unit;

[0009] S5. The digital twin of the unit has a certain degree of reconfigurability. Through convenient data-driven and model adjustment, it can simulate different unit operating conditions and business formats, and can be stored and replayed separately;

[0010] S6. Realize the DCS logical configuration environment that is identical to the actual unit, facilitating the implementation of unit coordinated control strategies and parameter optimization, testing, demonstration and other operational processes;

[0011] S7. Design and develop an AGC coordinated control strategy and logic configuration with advanced technical features suitable for a typical 350MW supercritical cogeneration unit, capable of operating with the unit's online digital twin to optimize, test, and verify the unit's AGC coordinated control strategy;

[0012] Able to coordinate control technology of turbine and boiler based on precise energy balance (PEB) control;

[0013] Research on establishing a high-precision online fully-stimulated virtual DPU twin simulator:

[0014] 1) After the simulator is connected to the SIS system, it receives the system field data in a one-way manner. By processing the field data, it can realize the online twin function of the simulator and the unit;

[0015] 2) Online tracking function: The online tracking function under high load analyzes the core parameters of the on-site working conditions, calculates and simulates the online simulation conditions, and tracks the on-site AGC instructions in real time to achieve online twin simulation status tracking of the unit;

[0016] The acquired data can be processed and analyzed by using the principal component analysis (PCA) algorithm for dimensionality reduction. The formula is: Y = XU (where X is the original data matrix, U is the orthogonal basis vector, and Y is the data after dimensionality reduction).

[0017] To update the relevant parameters and states in the digital twin model, the model predictive control (MPC) algorithm can be used to optimize the update strategy.

[0018] According to a thermal power plant AGC load OL_EBS coordinated control method provided by the present invention, the single AGC response time should be no more than 50-55 seconds. The introduction of the recurrent neural network model allows for rapid positive and negative balance calculations, and the single AGC response time is no more than 50-55 seconds, which greatly improves the timeliness of load regulation and enables more flexible and timely response to load changes. The recurrent neural network allows for rapid positive and negative balance calculations, thereby improving the accuracy and efficiency of regulation.

[0019] According to a thermal power plant AGC load OL_EBS coordinated control method provided by the present invention, biLSTM is used to perform regression prediction on the database, and then a new BTU correction output is obtained based on the regression prediction model and the new fuel input. A large amount of historical data of the unit is used to establish a prediction model for coal calorific value demand, and a boiler calorific value prediction is established according to the AGC instruction requirements. The energy balance correction system of fuel, feed water, air supply, and steam temperature is corrected, and a coal calorific value demand prediction model and a boiler calorific value prediction are established. Combined with the energy balance correction system, the reasonable adjustment of key parameters such as fuel, feed water, air supply, and steam temperature is ensured, and the stability and efficiency of the overall system operation are improved. By analyzing and modeling a large amount of historical data, the accuracy of the prediction and the intelligence level of the adjustment are improved.

[0020] According to a thermal power plant AGC load OL_EBS coordinated control method provided by the present invention, the unit status is discovered by fuzzy matching, and the control loop parameters of the water, coal and wind regulation lag are quickly corrected and adjusted. The unit status is quickly identified through fuzzy matching, and the control parameters can be adjusted in real time to shorten the lag time and adapt to complex and changeable working conditions.

[0021] Beneficial effects:

[0022] The thermal power plant AGC load OL_EBS rapid adjustment method of this technical solution, the AGC frequency regulation service is based on the highest winning bid price of 12 yuan, and the market is basically cleared at 12 yuan, so the average winning bid price is calculated as 12 yuan, the highest operating value Kp of the comprehensive frequency regulation performance index is 3.6, and the average Kp value is 3.2. Combined with historical production data, the average daily frequency regulation mileage is 8926.0MW. Based on this, the average daily income is calculated as follows: daily income = 8926*[ln(3.2)+1]*12 = 231,700 yuan. The annual operating time of the unit is 300 days (considering maintenance and shutdown time), and the number of winning bid days for frequency regulation is budgeted at 150 days. The average annual income is: annual income = 8926*[ln(3.2)+1]*12*150 = 34.7549 million yuan;

[0023] After the implementation of this project, it helps production personnel to optimize parameters so that the unit always operates at the optimal operating point. The Kp value can be optimized to a better level than units of the same level, with an average value of 3.5. The average annual income is: income = 8926*[ln(3.6)+1]*12*150=36.6473 million yuan, which can increase the annual AGC frequency regulation service income by 1.8924 million yuan. In addition, it can also predict the "thermal energy" of the boiler, thereby acting in advance to feed coal, effectively compensate for the inertia of the boiler, and ensure that the unit has a fast load response speed and smooth pressure changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments;

[0025] Figure 1 This is a diagram of the ol-EBs online energy balance self-calibration process of the thermal power plant AGC load OL_EBS rapid adjustment method of the present invention; DETAILED DESCRIPTION

[0026] This section will describe in detail the specific embodiments of the present invention. The preferred embodiments of the present invention are shown in the accompanying drawings. The purpose of the accompanying drawings is to supplement the description of the text part of the specification with graphics, so that people can intuitively and vividly understand each technical feature and the overall technical solution of the present invention, but it should not be understood as a limitation on the scope of protection of the present invention.

[0027] Reference Figure 1 The embodiment of the present invention provides a method for coordinated control of AGC load OL_EBS in a thermal power plant, which includes:

[0028] S1. A coordinated control system that uses advanced control features such as biLSTM predictive control, adaptive control, and recurrent neural networks can predict the "heat energy" of the boiler;

[0029] S2, thus feeding coal in advance, effectively compensating for the boiler's inertia, and ensuring the unit has a fast load response speed and smooth pressure changes;

[0030] S3. Improve the frequency regulation performance of the automatic generation control (AGC) of thermal power units and enhance the AGC frequency regulation service capability of the AGC comprehensive frequency regulation performance index (K);

[0031] S4. Achieve a high-precision digital twin of a designated typical 350MW supercritical cogeneration unit;

[0032] S5. The digital twin of the unit has a certain degree of reconfigurability. Through convenient data-driven and model adjustment, it can simulate different unit operating conditions and business formats, and can be stored and called separately;

[0033] S6. Realize the DCS logical configuration environment that is identical to the actual unit, facilitating the implementation of unit coordinated control strategies and parameter optimization, testing, demonstration and other operational processes;

[0034] S7. Design and develop an AGC coordinated control strategy and logic configuration with advanced technical features suitable for a specified typical 350MW supercritical cogeneration unit, and carry out optimization, testing and system verification;

[0035] Able to provide coordinated control scheme of turbine and boiler based on precise energy balance (PEB) control;

[0036] Research on the establishment of a high-precision online twin fully-stimulated virtual DPU twin simulator:

[0037] 1) After the simulator is connected to the SIS system, it receives the system field data in a one-way manner. By processing the field data, the online functions of the simulator and the unit can be realized;

[0038] 2) Online tracking function: The online tracking function under high load analyzes the core parameters of the on-site working conditions, calculates and simulates the online simulation working conditions, and tracks the on-site working conditions in real time to achieve online simulation status tracking of the unit;

[0039] The acquired data can be processed and analyzed by using the principal component analysis (PCA) algorithm for dimensionality reduction. The formula is: Y = XU (where X is the original data matrix, U is the orthogonal basis vector, and Y is the data after dimensionality reduction).

[0040] To update the relevant parameters and states in the digital twin model, the model predictive control (MPC) algorithm can be used to optimize the update strategy. The single AGC response time should not exceed 50-55s. By introducing the recurrent neural network model, fast positive and negative balance calculations can be performed, and BILSTM is used to perform regression prediction on the database. Then, based on the regression prediction model and the new fuel input, a new BTU correction output is obtained. A large amount of historical data of the unit is used to establish a prediction model for coal calorific value demand. According to the AGC instruction requirements, a boiler calorific value prediction is established, and the energy balance correction system of fuel, feed water, supply air, and steam temperature is corrected. The fuzzy matching unit state is discovered, and the control loop parameters of the water, coal, and air regulation lag are quickly corrected. As the unit operating conditions change, the powerful learning and fitting capabilities of the deep neural network algorithm are used to estimate the system parameters, and the similarity between the model output and the actual output is used as an indicator to evaluate equipment abnormalities, thereby achieving advanced warning of parameter system abnormalities.

[0041] Example 1: As the operating conditions of the unit change, the limit range of the traditional threshold alarm is difficult to select and the single parameter alarm mode lacks judgment on the overall system. The fuel calorific value is calculated through historical big data experience. First, based on the regression prediction model and the new fuel input, a new BTU correction instruction is obtained. The large amount of historical data of the unit is used to establish a current prediction model of the coal calorific value. According to the AGC instruction requirements, a boiler calorific value prediction is established, and the energy balance correction system of fuel, feed water, air supply, and steam temperature is corrected. The fuzzy matching unit state is discovered, and the control loop parameters of the water-coal-air regulation lag are quickly corrected. A coal calorific value demand prediction model and a boiler calorific value prediction are established. Combined with the energy balance correction system, the reasonable adjustment of key parameters such as fuel, feed water, air supply, and steam temperature is ensured. The powerful learning and fitting capabilities of the deep neural network algorithm are used to estimate the system parameters, and the similarity between the model output and the actual output is used as an evaluation index for equipment abnormality, to achieve advanced warning of parameter system abnormalities.

[0042] The embodiments of the present invention are described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Various changes can be made within the scope of knowledge possessed by ordinary technicians in the technical field without departing from the scope of the present invention.

Claims

1. A method for coordinated control of AGC load OL_EBS in a thermal power plant, characterized in that: The following steps are involved: S1. The OL_EBS coordinated control system, which uses advanced control features such as biLSTM regression predictive control, adaptive control, and recurrent neural networks, can quickly predict the "heat energy" of the boiler; S2, thereby adjusting the coal and water supply in advance, effectively compensating for the boiler's inertia, and ensuring the unit has a fast load response speed and smooth pressure changes; S3. Improve the automatic generation control frequency regulation performance of thermal power units and enhance the AGC frequency regulation service capability of the AGC comprehensive frequency regulation performance index (K); S4. Implement a high-precision digital twin model of a designated typical 350MW supercritical cogeneration unit; S5. The digital twin of the unit has a certain degree of reconfigurability. Through convenient data-driven and model adjustment, it can simulate different unit operating conditions and business formats, and can be stored and replayed separately; S6. Realize an online twin DCS logic configuration environment identical to the actual unit, facilitating the implementation of unit coordinated control strategies and parameter optimization, testing, demonstration and other operational processes; S7. Design and develop an AGC coordinated control ol_EBs solution with advanced technical features suitable for a specified typical 350MW supercritical cogeneration unit, and undergo optimization, testing and system verification; Ability to coordinate load control technology based on precise energy balance control; Able to establish high-precision online twin-type fully-stimulated virtual DPU twin simulator research; The specific methods are as follows: 1) After the simulator is connected to the DCS or SIS system, it receives the system's field data in a one-way manner. By processing the field data and applying biLSTM regression predictive control, it can quickly predict the boiler's "heat energy"; 2) Online tracking function: The online tracking function under high load analyzes the core parameters of the on-site working conditions and realizes the online simulation state tracking of the unit through the online twin simulation model and self-learning process; The acquired data can be processed and analyzed by using the principal component analysis (PCA) algorithm for dimensionality reduction. The formula is: Y = XU, where X is the original data matrix, U is the orthogonal basis vector, and Y is the data after dimensionality reduction. To update the relevant parameters and states in the digital twin model, the model predictive control (MPC) algorithm can be used to optimize the update strategy.

2. A thermal power plant AGC load OL_EBS coordinated control method according to claim 1, characterized in that: The single AGC response time should be no more than 50-55s, and the introduction of the recurrent neural network model allows for rapid positive and negative balance calculations.

3. A thermal power plant AGC load OL_EBS coordinated control method according to claim 1, characterized in that: The BILSTM is used to perform regression prediction on the database, and then a new BTU correction coefficient is obtained based on the regression prediction model and the new fuel input. The unit uses a large amount of historical data to establish a prediction model for coal calorific value demand. According to the AGC instruction requirements, the energy balance regulation system of fuel, feed water, air supply, and steam temperature is corrected through rapid calculation of the boiler calorific value.

4. A thermal power plant AGC load OL_EBS coordinated control method according to claim 1, characterized in that: The unit status of the fuzzy matching is discovered and the control loop parameters of the water, coal and wind regulation lag are adjusted quickly and correctively.

Images