Fine ammonia injection adjusting system for coal-fired power plant

Abstract

The invention discloses a refined ammonia injection adjusting system for a coal-fired power plant, and relates to the technical field of environmental protection of coal-fired power plants. Comprising a data acquisition module which is composed of a plurality of sensors and is used for monitoring parameters in real time; the central processing module receives the data acquired by the data acquisition module and then generates a control instruction; the ammonia spraying execution module is used for receiving the control instruction of the central processing module and performing ammonia spraying treatment; the data management module is internally provided with a data storage unit, and the data storage unit is used for storing data in a topological node storage form; the data retrieval unit is used for performing data retrieval according to user input; the data analysis unit is used for analyzing the data and then providing a visual chart for a user to observe, and the sensor module comprises an infrared spectrum analyzer and a laser scattering instrument. The nitrogen oxide emission concentration is stably controlled to be 50 mg / Nm < 3 > or below, and the national environmental protection standard is met.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental protection of coal-fired power plants, and in particular to a refined ammonia injection adjustment system for coal-fired power plants. Background Art

[0002] A coal-fired power plant is also called a fire dragon power plant. A thermal power plant is abbreviated as a thermal power plant. It is a plant that uses combustibles (such as coal) as fuel to produce electricity. Its basic production process is: when the fuel is burned, it heats water to generate steam, converting the chemical energy of the fuel into thermal energy. The steam pressure drives the turbine to rotate, converting the thermal energy into mechanical energy, and then the turbine drives the generator to rotate, converting the mechanical energy into electrical energy.

[0003] The prime mover is usually a steam engine or gas turbine, and in some smaller power stations, an internal combustion engine may also be used. They all generate electricity by using the pressure drop during the process of high-temperature, high-pressure steam or gas passing through a turbine to become low-pressure air or condensed water.

[0004] During the operation of coal-fired power plants, burning coal will produce a large amount of nitrogen oxides (NOx), which are one of the main sources of air pollution and have serious impacts on human health and the environment. In order to reduce the emission of nitrogen oxides, selective catalytic reduction (SCR) technology is usually used. This technology sprays ammonia or urea and other reducing agents into the flue gas to react with nitrogen oxides to generate nitrogen and water vapor, thereby purifying the flue gas.

[0005] Traditional denitrification systems often have problems with insufficient precision in ammonia injection control, resulting in unstable denitrification effects and high ammonia escape rates.

[0006] Therefore, it is particularly important to develop an ammonia injection system that can be monitored in real time and adjusted accurately. Summary of the invention

[0007] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a refined ammonia injection adjustment system for a coal-fired power plant.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0010] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0011] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0012] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0013] Data management module.

[0014] Preferably, the data management module has:

[0015] A data storage unit, the data storage unit stores data in a topological node storage form;

[0016] A data retrieval unit, which is used to retrieve data according to user input;

[0017] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0018] Preferably: the sensor module includes an infrared spectrometer and a laser scatterer, and the parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

[0019] Preferably, the central processing module includes a main regulator and a plurality of sub-regulators.

[0020] Preferably: the control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0021] Preferably: the control logic of the secondary regulator is Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0022] Preferably: the ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

[0023] Preferably, the central processing module also has a built-in ammonia injection amount prediction unit.

[0024] Preferably: the prediction logic of the ammonia injection amount prediction unit is:

[0025] S1: Determine the expression of the gradient boosting iterative decision tree prediction function, which is x is the input sample; h t is the tth regression tree; w t is the regression tree parameter; ρ t is the weight of the tth regression tree;

[0026] S2: Determine the optimal value of the prediction function L is the loss function;

[0027] S3: Gradient boosting iterative decision tree is iterated.

[0028] Preferably: the iterative steps are as follows:

[0029] S31: Definition; f i is a weak learner;

[0030] S32: Construct training samples based on regression tree, and the objective function is: L(y i ,F(x i ))=(y i -F(x i )) 2 , (y i ,x i )∈R*R N

[0031] S33: The fitted data obtained by training the tree based on the gradient descent direction is The best fit data is

[0032] S34: Calculate the optimal step size in the gradient descent direction, which is ρ to is the initial weight of the tth regression tree

[0033] S35: Calculate the weak learner of the tth regression tree, which is f t =ρ * h t (x i ,w* );

[0034] S36: The prediction function obtained is F t (x) = F t-1 (x)+f t ;

[0035] S37: If the loss function meets the error convergence condition or the value of the obtained regression tree reaches the preset value, the iteration is terminated; if not, the iteration continues.

[0036] The beneficial effects of the present invention are:

[0037] 1. The nitrogen oxide emission concentration of the present invention is stably controlled at 50mg / Nm 3 The following meet national environmental protection standards.

[0038] 2. The ammonia escape rate of the present invention is greatly reduced, thereby reducing secondary pollution to the environment.

[0039] 3. The present invention improves the overall operating efficiency of the denitration system and reduces operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is an architecture diagram of a refined ammonia injection adjustment system for a coal-fired power plant proposed in the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0042] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", and "set" should be understood in a broad sense, for example, they can be fixedly connected or set, or detachably connected or set, or integrally connected or set. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0043] Embodiment 1:

[0044] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0045] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0046] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0047] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0048] Data management module, which has built-in:

[0049] A data storage unit, the data storage unit stores data in a topological node storage form;

[0050] A data retrieval unit, which is used to retrieve data according to user input;

[0051] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0052] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

[0053] Embodiment 2:

[0054] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0055] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0056] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0057] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0058] Data management module, which has built-in:

[0059] A data storage unit, the data storage unit stores data in a topological node storage form;

[0060] A data retrieval unit, which is used to retrieve data according to user input;

[0061] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0062] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

[0063] The central processing module includes a main regulator and a plurality of sub-regulators.

[0064] The control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, Ci is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0065] The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0066] Embodiment 3:

[0067] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0068] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0069] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0070] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0071] Data management module, which has built-in:

[0072] A data storage unit, the data storage unit stores data in a topological node storage form;

[0073] A data retrieval unit, which is used to retrieve data according to user input;

[0074] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0075] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

[0076] The central processing module includes a main regulator and a plurality of sub-regulators.

[0077] The control logic of the main regulator is V i =(C i -C set)×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0078] The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0079] The ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3 is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

[0080] Embodiment 4:

[0081] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0082] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0083] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0084] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0085] Data management module, which has built-in:

[0086] A data storage unit, the data storage unit stores data in a topological node storage form;

[0087] A data retrieval unit, which is used to retrieve data according to user input;

[0088] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0089] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

[0090] The central processing module includes a main regulator and a plurality of sub-regulators.

[0091] The control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0092] The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0093] The ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3 is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

[0094] The central processing module also has a built-in ammonia injection amount prediction unit.

[0095] The prediction logic of the ammonia injection amount prediction unit is:

[0096] S1: Determine the expression of the gradient boosting iterative decision tree prediction function, which is x is the input sample; h t is the tth regression tree; w t is the regression tree parameter; ρ t is the weight of the tth regression tree;

[0097] S2: Determine the optimal value of the prediction function L is the loss function;

[0098] S3: Gradient boosting iterative decision tree is iterated.

[0099] Embodiment 5:

[0100] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0101] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0102] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0103] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0104] Data management module, which has built-in:

[0105] A data storage unit, the data storage unit stores data in a topological node storage form;

[0106] A data retrieval unit, which is used to retrieve data according to user input;

[0107] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0108] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored by the sensor module include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid-based manner.

[0109] The central processing module includes a main regulator and a plurality of sub-regulators.

[0110] The control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0111] The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0112] The ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3 is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

[0113] The central processing module also has a built-in ammonia injection amount prediction unit.

[0114] The prediction logic of the ammonia injection amount prediction unit is:

[0115] S1: Determine the expression of the gradient boosting iterative decision tree prediction function, which is x is the input sample; h t is the tth regression tree; w t is the regression tree parameter; ρ t is the weight of the tth regression tree;

[0116] S2: Determine the optimal value of the prediction function L is the loss function;

[0117] S3: Gradient boosting iterative decision tree is iterated.

[0118] In the S3 step, the iterative steps are as follows:

[0119] S31: Definition; f i is a weak learner;

[0120] S32: Construct training samples based on regression tree, and the objective function is: L(y i ,F(x i ))=(y i -F(xi )) 2 , (y i ,x i )∈R*R N

[0121] S33: The fitted data obtained by training the tree based on the gradient descent direction is The best fit data is

[0122] S34: Calculate the optimal step size in the gradient descent direction, which is ρ to is the initial weight of the tth regression tree

[0123] S35: Calculate the weak learner of the tth regression tree, which is f t =ρ * h t (x i ,w * );

[0124] S36: The prediction function obtained is F t (x) = F t-1 (x)+f t ;

[0125] S37: If the loss function meets the error convergence condition or the value of the obtained regression tree reaches the preset value, the iteration is terminated; if not, the iteration continues.

[0126] Embodiment 6:

[0127] A refined ammonia injection adjustment system for a coal-fired power plant, comprising:

[0128] Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time;

[0129] The central processing module receives the data collected by the data acquisition module and then generates control instructions;

[0130] An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing;

[0131] Data management module, which has built-in:

[0132] A data storage unit, the data storage unit stores data in a topological node storage form;

[0133] A data retrieval unit, which is used to retrieve data according to user input;

[0134] The data analysis unit analyzes the data and then provides visual charts for users to observe.

[0135] The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a matrix manner.

[0136] The central processing module includes a main regulator and a plurality of sub-regulators.

[0137] The control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0138] The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

[0139] The ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3 is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

[0140] The central processing module also has a built-in ammonia injection amount prediction unit.

[0141] The prediction logic of the ammonia injection amount prediction unit is:

[0142] S1: Determine the expression of the gradient boosting iterative decision tree prediction function, which is x is the input sample; h t is the tth regression tree; w t is the regression tree parameter; ρ tis the weight of the tth regression tree;

[0143] S2: Determine the optimal value of the prediction function L is the loss function;

[0144] S3: Gradient boosting iterative decision tree is iterated.

[0145] In the S3 step, the iterative steps are as follows:

[0146] S31: Definition; f i is a weak learner;

[0147] S32: Construct training samples based on regression tree, and the objective function is: L(y i ,F(x i ))=(y i -F(x i )) 2 , (y i ,x i )∈R*R N

[0148] S33: The fitted data obtained by training the tree based on the gradient descent direction is The best fit data is

[0149] S34: Calculate the optimal step size in the gradient descent direction, which is ρ to is the initial weight of the tth regression tree

[0150] S35: Calculate the weak learner of the tth regression tree, which is f t =ρ * h t (x i ,w * );

[0151] S36: The prediction function obtained is F t (x) = F t-1 (x)+f t ;

[0152] S37: If the loss function meets the error convergence condition or the value of the obtained regression tree reaches the preset value, the iteration is terminated; if not, the iteration continues.

[0153] The nitrogen oxide emission concentration of the present invention is stably controlled at 50mg / Nm 3 The following meet national environmental protection standards.

[0154] The ammonia escape rate of the present invention is greatly reduced, thereby reducing secondary pollution to the environment.

[0155] The present invention improves the overall operation efficiency of the denitration system and reduces the operation cost.

[0156] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A refined ammonia injection adjustment system for a coal-fired power plant, characterized in that: include: Data acquisition module, which consists of a variety of sensors and is used to monitor parameters in real time; The central processing module receives the data collected by the data acquisition module and then generates control instructions; An ammonia injection execution module receives control instructions from the central processing module and performs ammonia injection processing; Data management module.

2. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 1, characterized in that: The data management module has: A data storage unit, the data storage unit stores data in a topological node storage form; A data retrieval unit, which is used to retrieve data according to user input; The data analysis unit analyzes the data and then provides visual charts for users to observe.

3. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 1, characterized in that: The sensor module includes an infrared spectrometer and a laser scatterer. The parameters monitored include nitrogen oxide concentration, oxygen concentration, temperature, ammonia injection flow rate and ammonia injection pressure. The sensor module is deployed in a grid or matrix manner.

4. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 1, characterized in that: The central processing module includes a main regulator and a plurality of sub-regulators.

5. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 4, characterized in that: The control logic of the main regulator is V i =(C i -C set )×K p ×f(L), where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, C set is the set nitrogen oxide concentration target value, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

6. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 4, characterized in that: The control logic of the sub-regulator is: Where V i is the adjustment amount of the i-th ammonia injection branch pipe, C i is the nitrogen oxide concentration in the ith region, is the average of the target values ​​of nitrogen oxide concentration in the m regions adjacent to the i-th region, K p is the proportionality coefficient, K p ∈(0.1, 1), f(L) is the load correction function, and L is the boiler.

7. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 1, characterized in that: The ammonia injection amount adjustment formula of the ammonia injection execution module is Q NH3 =K*(C NOX -C NOXtargrt )*Q fluegas , where Q NH3 is the amount of ammonia sprayed, K is the empirical coefficient C NOX is the current concentration of nitrogen oxides in the flue gas, C NOXtargrt is the target nitrogen oxide concentration, Q fluegas is the flue gas flow rate.

8. The refined ammonia injection adjustment system for a coal-fired power plant according to claim 1, characterized in that: The central processing module also has a built-in ammonia injection amount prediction unit.

9. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 8, characterized in that: The prediction logic of the ammonia injection amount prediction unit is: S1: Determine the expression of the gradient boosting iterative decision tree prediction function, which is x is the input sample; h t is the tth regression tree; w t is the regression tree parameter; ρ t is the weight of the tth regression tree; S2: Determine the optimal value of the prediction function L is the loss function; S3: Gradient boosting iterative decision tree is iterated.

10. A refined ammonia injection adjustment system for a coal-fired power plant according to claim 9, characterized in that: In the S3 step, the iterative steps are as follows: S31: Definition; f i is a weak learner; S32: Construct training samples based on regression tree, and the objective function is: L(y i ,F(x i ))=(y i -F(x i )) 2 , (y i ,x i )∈R*R N S33: The fitted data obtained by training the tree based on the gradient descent direction is The best fit data is S34: Calculate the optimal step size in the gradient descent direction, which is ρ to is the initial weight of the tth regression tree S35: Calculate the weak learner of the tth regression tree, which is f t =ρ * h t (x i ,w * ); S36: The prediction function obtained is F t (x) = F t-1 (x)+f t ; S37: If the loss function meets the error convergence condition or the value of the obtained regression tree reaches the preset value, the iteration is terminated; if not, the iteration continues.

Images