Load prediction control method for low calorific value gas turbine based on dual redundant digital twin

Abstract

The present invention discloses a method for predicting and controlling the load of a low calorific value gas turbine based on a dual redundant digital twin. A dynamic balance thermodynamic mechanism model of the load of a low calorific value gas turbine is constructed as the first digital twin of the load of the low calorific value gas turbine; the key cross-section thermodynamic parameters of the gas path characteristics of the low calorific value gas turbine are extracted to establish the second digital twin of the load of the low calorific value gas turbine; the first digital twin and the second digital twin are connected in parallel to establish a dual redundant digital twin, and the real-time operation data of the low calorific value gas turbine is processed to predict the load change in advance, so as to realize the safety monitoring of the low calorific value gas turbine and the dynamic adjustment control of the load deviation. The present invention realizes prediction and monitoring control, has strong anti-interference ability, can effectively detect the abnormality of the unit, reduce false alarms, and improve the safety of the unit operation. It has the self-learning ability of the digital twin, improves the real-time and accuracy of the dynamic tracking of the load, and reduces the operation and maintenance costs.

Claims

1. A load prediction control method for low calorific value gas turbine based on dual redundant digital twins, Features: The method comprises the following steps: Step 1: Construct a dynamic balance thermodynamic mechanism model of the load of the low calorific value gas turbine as the first digital twin of the low calorific value gas turbine load; Step 2: extracting thermodynamic parameters of key sections of gas path characteristics of the low calorific value gas turbine, and establishing a second digital twin of the low calorific value gas turbine load according to the thermodynamic parameters of key sections of gas path characteristics of the low calorific value gas turbine; Step 3: Connect the first digital twin and the second digital twin in parallel to establish a dual redundant low calorific value gas turbine load digital twin; Step 4: Use the dual redundant low calorific value gas turbine load digital twin to process the real-time operation data of the low calorific value gas turbine, predict the load change in advance, and realize the safety monitoring of the low calorific value gas turbine and the dynamic adjustment control of the load deviation; Step 5: After dynamically adjusting the control, collect real-time operating data feedback and update it to the dual redundant low calorific value gas turbine load digital twin.

2. According to claim 1, the low calorific value gas turbine load prediction control method based on dual redundant digital twins, Features: The step 1 is specifically as follows: using the load dynamic balance thermodynamic mechanism model of the following formula as the first digital twin, and calculating the low calorific value gas turbine load through the first digital twin: P.S out1 (P T -P AC / η m )η ms -P GC Among them, P out1 represents the load of the low calorific value gas turbine predicted by the first digital twin, P T Indicates the power output of the turbine, P AC Indicates the power consumption of the air compressor, P GC Indicates the power consumption of the gas compressor, η m represents the efficiency of the compressor turbine shaft, η ms represents the efficiency of the gas turbine output shaft, η CB represents the efficiency of the combustion chamber, represents the mass flow rate at the air compressor inlet, represents the mass flow rate of blast furnace gas, represents the mass flow rate of coke oven gas, Represents the mass flow rate of secondary air, H u62 Indicates the calorific value of blast furnace gas entering the combustion chamber, H u7 Indicates the calorific value of coke oven gas entering the combustion chamber; T 41 represents the turbine inlet temperature, T 42 represents the turbine outlet temperature, T 62 Indicates the gas compressor outlet temperature, T 61 Indicates the inlet temperature of the blast furnace gas compressor, T 7 Indicates the temperature of coke oven gas entering the combustion chamber, T 22 Indicates the outlet temperature of the air compressor, T 21 Indicates the inlet temperature of the air compressor; is the turbine inlet specific heat, is the turbine outlet specific heat, represents the specific heat of coke oven gas, represents the specific heat of air compressor outlet, represents the specific heat of air compressor inlet, Indicates the specific heat of blast furnace gas entering the combustion chamber, Indicates the specific heat at the gas compressor inlet.

3. The low calorific value gas turbine load prediction control method based on dual redundant digital twin according to claim 1, Features: In the first digital twin, ranges of key performance parameters related to design and manufacturing accuracy are set.

4. The low calorific value gas turbine load prediction control method based on dual redundant digital twin according to claim 1, It is characterized in that In step 2, the extracted key cross-section thermodynamic parameters of the gas path characteristics of the low calorific value gas turbine include: atmospheric pressure P 11 , air compressor inlet temperature T 21 , air compressor inlet static pressure P 21 , radial intake chamber intake differential pressure dP 1 , air compressor outlet pressure P 22 , air compressor outlet temperature T 22 , Blast furnace gas mass flow Coke oven gas mass flow Blast furnace gas calorific value H u62 , turbine exhaust pressure P 42 , turbine exhaust temperature T 42 , Gas compressor inlet pressure P 61 , Gas compressor inlet temperature T 61 , Gas compressor outlet pressure P 62 , Gas compressor outlet temperature T 62 .

5. The low calorific value gas turbine load prediction control method based on dual redundant digital twin according to claim 1, It is characterized in that The second step is specifically: according to the thermodynamic parameters of the key cross-section of the gas path characteristics of the low calorific value gas turbine, support vector machine regression SVR machine learning is used to establish the second digital twin of the low calorific value gas turbine load of the following formula: Among them, P out2 represents the load of the low calorific value gas turbine predicted by the second digital twin, and the atmospheric pressure P 11 , air compressor inlet temperature T 21 , air compressor inlet static pressure P 21 , radial intake chamber intake differential pressure dP 1 , air compressor outlet pressure P 22 , air compressor outlet temperature T 22 , Blast furnace gas mass flow Coke oven gas mass flow Blast furnace gas calorific value H u62 , turbine exhaust pressure P 42 , turbine exhaust temperature T 42 , Gas compressor inlet pressure P 61 , Gas compressor inlet temperature T 61 , Gas compressor outlet pressure P 62 , Gas compressor outlet temperature T 62 .

6. The low calorific value gas turbine load prediction control method based on dual redundant digital twin according to claim 1, Features: The second digital twin is set to support vector machine regression SVR function as: k(x,x i )=(r+γx T x i ) 3 Where f represents the load function predicted by the second digital twin, x represents the 15-dimensional key cross-section thermodynamic parameter input vector, Ptest represents the measured load, i represents the number of the running data sample, and m represents the number of the running data samples. represents the i-th Lagrange multiplier on the right side, α i represents the i-th Lagrange multiplier on the left side, represents the jth Lagrange multiplier on the right side, α j represents the jth Lagrange multiplier on the left, the atmospheric pressure P 11 , air compressor inlet temperature T 21 , air compressor inlet static pressure P 21 , radial intake chamber intake differential pressure dP 1 , air compressor outlet pressure P 22 , air compressor outlet temperature T 22 , Blast furnace gas mass flow Coke oven gas mass flow Blast furnace gas calorific value H u62 , turbine exhaust pressure P 42 , turbine exhaust temperature T 42 , Gas compressor inlet pressure P 61 , Gas compressor inlet temperature T 61 , Gas compressor outlet pressure P 62 , Gas compressor outlet temperature T 62 , b represents the model parameters to be determined, x i represents the i-th running data, x j represents the j-th operating data, γ represents the polynomial coefficient to be determined, r represents the polynomial coefficient to be determined, T represents the vector transpose, and ∈ is the allowable deviation between the predicted load and the measured load of the second digital twin.

7. The low calorific value gas turbine load prediction control method based on dual redundant digital twin according to claim 1, It is characterized in that The step three is specifically as follows: The real-time operation data of the low calorific value gas turbine is input into the first digital twin and the second digital twin of the low calorific value gas turbine load to predict the respective low calorific value gas turbine loads. By connecting the predicted load of the first digital twin and the predicted load of the second digital twin in parallel, a dual redundant low calorific value gas turbine load digital twin predicted load is established: Where P outsr Load predicted by the dual redundant low heating value gas turbine load digital twin, ΔP 1 is the deviation between the predicted load and the measured load of the first digital twin, ΔP 2 is the deviation between the predicted load and the measured load of the second digital twin, P out1 represents the load of the low calorific value gas turbine predicted by the first digital twin, P out2 represents the load of the low heating value gas turbine predicted by the second digital twin; Dynamically control the mass flow of blast furnace gas into low calorific value gas turbines and coke oven gas mass flow The dual-redundant low calorific value gas turbine load digital twin load prediction model is used to realize the prediction and dynamic control of the low calorific value gas turbine load under all conditions of speed increase and load increase.

Images