CN113433911A - Denitration device ammonia injection accurate control system and method based on concentration accurate prediction - Google Patents

Denitration device ammonia injection accurate control system and method based on concentration accurate prediction Download PDF

Info

Publication number
CN113433911A
CN113433911A CN202110736767.0A CN202110736767A CN113433911A CN 113433911 A CN113433911 A CN 113433911A CN 202110736767 A CN202110736767 A CN 202110736767A CN 113433911 A CN113433911 A CN 113433911A
Authority
CN
China
Prior art keywords
model
denitration device
prediction
ammonia injection
control
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110736767.0A
Other languages
Chinese (zh)
Other versions
CN113433911B (en
Inventor
高翔
郑成航
谭畅
张涌新
周灿
吴卫红
翁卫国
杨洋
张悠
姚龙超
刘少俊
李钦武
孙德山
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202110736767.0A priority Critical patent/CN113433911B/en
Publication of CN113433911A publication Critical patent/CN113433911A/en
Application granted granted Critical
Publication of CN113433911B publication Critical patent/CN113433911B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
    • G05B19/41865—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by job scheduling, process planning, material flow
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/33—Director till display
    • G05B2219/33273—DCS distributed, decentralised controlsystem, multiprocessor
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

The invention relates to a denitration device ammonia injection accurate control system and method based on accurate concentration prediction. According to the method, the concentration of NOx at the inlet of the denitration device is predicted in advance, accurate feedforward is provided for ammonia injection amount control of the denitration device, meanwhile, a multi-model prediction control module under a variable load working condition is established, accurate control of the ammonia injection amount of the denitration device is achieved, and the defects of large delay, large inertia and strong nonlinearity of a denitration system are overcome; under the working condition of large-range variable load, the invention greatly improves the economy and stability of ammonia injection amount control of the denitration device under the condition of ensuring that the concentration of the NOx at the outlet reaches the standard.

Description

Denitration device ammonia injection accurate control system and method based on concentration accurate prediction
Technical Field
The invention belongs to the technical field of optimization of an ammonia injection device of a denitration system, and particularly relates to a denitration device ammonia injection accurate control system and method based on accurate concentration prediction.
Background
The combustion of coal can produce pollutants such as nitrogen oxides, sulfur dioxide, smoke dust and the like, and the pollutants can cause harm to the atmospheric environment and human health.
The SNCR/SCR coupling denitration technology is a combined technology developed by combining the SNCR technology and the SCR technology, and has the characteristics of low investment cost of the SNCR technology and high denitration efficiency of the SCR technology. However, due to the characteristics of large measurement hysteresis, high measurement accuracy and strong nonlinearity of a denitration system inlet NOx concentration, the CFB boiler adopting the SNCR and SCR coupling denitration technology has the problem that the ammonia injection amount cannot be accurately controlled, and the NOx concentration exceeds the emission standard when the ammonia injection amount is too small; when the ammonia injection amount is excessive, ammonia escapes, the atmosphere is polluted, and sulfate is generated, so that the air preheater and the catalyst are seriously blocked, and the equipment safety is influenced.
In order to improve the economy and stability of the denitration device in the ammonia injection process on the premise of ensuring that the concentration of the NOx in the outlet flue gas reaches the standard, a proper denitration device inlet NOx concentration prediction model and an outlet NOx concentration control method need to be established, and the ammonia injection amount of a denitration system is accurately controlled.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention provides a denitration device ammonia injection accurate control system and method based on accurate concentration prediction; according to the method, the concentration of NOx at the inlet of the denitration device is predicted in advance, accurate feedforward is provided for controlling the ammonia injection amount of the denitration device, and meanwhile, a multi-model prediction control module under a variable load working condition is established, so that the accurate control of the ammonia injection amount of the denitration device is realized, and the defects of large delay, large inertia and strong nonlinearity of a denitration system are overcome.
The technical scheme adopted by the invention is as follows:
an accurate control system for ammonia injection of a denitration device based on accurate concentration prediction comprises a power station information system, a denitration device inlet NOx concentration prediction model, a multi-model prediction control module and a denitration device control object;
the power station information system comprises a power plant OPC (OLE for Process control) server and DCS (distributed control System) control equipment, is in communication connection with the inlet NOx concentration prediction model of the denitration device, and transmits DCS data to the inlet NOx concentration prediction model of the denitration device in real time; calculating an outlet NOx predicted value at the current moment in advance of a pollutant Emission Continuous Monitoring System (CEMS) meter by using a denitration device inlet NOx concentration prediction model, and inputting the predicted value into a multi-model prediction control module as a feedforward;
the multi-model prediction control module takes dynamic matrix control as an inner core, takes a NOx concentration prediction model at the inlet of the denitration device as feedforward, and divides different subinterval models through typical working conditions to realize accurate control of the ammonia injection device under the working conditions of large-range variable load;
the control object of the denitration device comprises the opening degree of a valve of the ammonia spraying device of the denitration device and the frequency of an ammonia spraying pump.
Preferably, the denitration device inlet NOx concentration prediction model is a denitration device inlet NOx concentration prediction model established based on a long-short term memory neural network (LSTM) algorithm; the model aims at the problems of inaccurate measurement and large hysteresis of an inlet NOx concentration meter of a boiler denitration device, data clustering is carried out on historical typical operation working conditions of the total coal feeding quantity of the boiler, a global LSTM neural network prediction model which is suitable for the large-range variable load working conditions of the boiler is established after super parameters are optimized, the measurement hysteresis error of a CEMS system is eliminated, and the inlet NOx concentration of the denitration device is accurately predicted; the denitration device inlet NOx concentration prediction model is established through the following steps:
(1) analyzing the operation condition of the boiler unit (analyzing the mechanism of generating and removing NOx of the CFB boiler) in a mechanism, selecting parameters influencing the generation and removal of the NOx as model input characteristic variables, and taking a NOx concentration prediction value at an inlet of a denitration device as a model output value;
(2) analyzing the relation between the selected input characteristic variables and the output value by using a mutual information statistical method, calculating the mutual information value between the input characteristic variables and the output value, and forming a judgment logic by combining empirical knowledge, namely deleting the input characteristics with low mutual information value and keeping the input characteristic variables with high mutual information value;
(3) input features are normalized, and the normalization processing formula is as follows:
Figure BDA0003141929940000021
in the formula, x is an input characteristic variable, mu is a mean value of the input characteristic variable, and sigma is a standard deviation of the input characteristic variable;
(4) introducing a wavelet transform method to determine the delay between input/output characteristics, eliminating signal fluctuation caused by the fluctuation of an instrument, establishing a relation between data of an original database and different input/output characteristics, determining the response time difference between different characteristics, and finally determining the response time difference between historical and real-time data and different input/output characteristics, wherein the wavelet transform formula is as follows:
Figure BDA0003141929940000031
wherein a is a scale factor, b is a translation factor, < f (t))，ψa，b(t)>Representing the inner product of two functions, representing the taking of the conjugate;
(5) identifying the working condition of the historical data of the selected characteristic quantity, screening out a stable working condition, a purging working condition and a variable load working condition, and clustering and segmenting according to the coal feeding quantity according to the variable load working condition; considering that different boilers, operation conditions and coal quality conditions have great influence on the NOx emission characteristics of the boilers, analyzing the instantaneous total coal supply quantity which can represent the operation characteristics of the boilers by a Euclidean clustering method, finding a plurality of typical working conditions of system operation, and dividing a subinterval model according to a plurality of typical working conditions; and developing an inlet NOx concentration prediction model under the large-range variable load working condition based on the subinterval model.
(6) Modeling is carried out on the basis of an LSTM (long short term memory neural network) algorithm, a NOx concentration prediction model at the inlet of the denitration device is established on the basis of the LSTM principle, and a hyper-parameter is determined; after the super-parameters are determined, a NOx concentration prediction model of the inlet of the denitration device can be established.
Preferably, the input characteristic variables comprise coal feeding quantity, primary air quantity, secondary air quantity, hearth temperature, hearth outlet smoke temperature, main steam flow and flue gas O2Volume percent; the hyper-parameters comprise a prediction time step, a backtracking time step, the number of hidden layers and the number of hidden layer nodes.
Preferably, when a mutual information value between the input characteristic variable and the output value is calculated, a minimum redundancy maximum correlation algorithm and empirical knowledge are used as evaluation indexes, and an evaluation formula is as follows:
Figure BDA0003141929940000032
where x is the input feature variable, y is the output value, and S is the feature { x }iAnd I is a mutual information value between x and y.
Preferably, after the denitration device inlet NOx concentration prediction model is established, the super-parameter is adjusted, and then the model accuracy is verified.
Aiming at the problems of inaccurate measurement, large hysteresis and the like of a NOx concentration meter at the inlet of a boiler denitration device, the invention clusters data of historical typical operation working conditions of the total coal feeding quantity of the boiler, establishes a global LSTM neural network prediction model adapting to the variable working conditions of the boiler, and measures and corrects the NOx concentration at the inlet of the denitration device; meanwhile, a multi-model predictive control module based on the fact that the NOx concentration measurement correction at the inlet of the denitration device is feedforward is established, and the multi-model predictive control module takes Dynamic Matrix Control (DMC) as an inner core and mainly comprises three parts, namely model prediction, rolling optimization and feedback correction.
Particularly, the multi-model predictive control module adds a feedforward part on the basis of dynamic matrix control, inputs an output value of a NOx concentration prediction model at the inlet of the denitration device into the multi-model predictive control module as feedforward, and performs cooperative control on ammonia injection amount together with feedback correction.
Particularly, the multi-model prediction control module establishes a plurality of subinterval prediction models based on different load working condition sections such as full load, high load, medium load and low load, and guarantees the stability of the NOx control effect at the outlet of the denitration system under the large-range variable load working condition in a weighting switching mode under different load working conditions.
The invention also provides a denitration device ammonia injection accurate control method based on accurate concentration prediction, wherein the ammonia injection accurate control system is embedded in a denitration DCS in a modular form and is communicated with the denitration DCS in real time, and when the combustion condition of a boiler changes, the output value of a denitration device inlet NOx concentration prediction model is used as a feedforward to be input into a multi-model prediction control module; meanwhile, the multi-model prediction control module identifies the current load working condition on line, carries out feedback correction on the ammonia injection amount at the current moment, carries out rolling optimization on the ammonia injection amount by combining the set value of the outlet NOx through the set feedforward feedback proportion, and sends a control instruction to the control object of the denitration device to increase or reduce the ammonia injection amount, thereby realizing the emission control on the concentration of the outlet NOx.
Preferably, the method specifically comprises the following steps:
step 1: firstly, establishing an ammonia injection accurate control system, establishing real-time communication with an original DCS (distributed control System) of a denitration device, and embedding the ammonia injection accurate control system in the denitration DCS in a module form;
step 2: after an ammonia injection accurate control system hardware platform is constructed, an NOx concentration prediction model at the inlet of the denitration device is established based on an algorithm technology;
and step 3: after the establishment of a denitration device inlet NOx concentration prediction model is completed, an optimal output value is obtained according to input characteristic variables, the influence of time difference of each input characteristic on the output characteristic is eliminated, the optimal prediction effect is obtained, and the requirement of controlling feedforward is met;
and 4, step 4: the multi-model predictive control comprises an outlet NOx feedforward predictive model and an outlet NOx control model, wherein the outlet NOx feedforward predictive model is the step response relation of air volume, coal volume and inlet NOx concentration to outlet NOx, and the outlet NOx control model is the step response relation of ammonia injection volume/air volume to outlet NOx concentration; based on the mutual information screening and screening results, the optimal air volume, the coal volume and the inlet NOx concentration are used as input parameters of a feedforward model for controlling the outlet NOx; preferably, the ammonia injection amount/air volume is used as an input parameter of an outlet NOx control quantity model;
and 5: selecting dynamic matrix predictive control as a basic control strategy of a multi-model predictive control module, wherein a prediction model in the dynamic matrix predictive control is a step response model, can be represented by a first-order inertial system model with pure lag, and obtains optimal parameters of the first-order inertial system with pure lag in each subarea by a particle swarm algorithm;
step 6: step 5, after a multi-model predictive control module is established, further adjusting parameters of the step response model in a field debugging mode; during field debugging, the working condition changes, and the main change amount in the model is a proportion link KpiAnd respond to the link TpiAnd a hysteresis loop TdiRemain substantially unchanged;
and 7: the multi-model prediction control module establishes a plurality of subinterval prediction models based on different load working condition sections such as full load, high load, medium load, low load and the like, and ensures the stability of the NOx control effect of the outlet of the denitration device under the working condition of large-range variable load in a weighting switching mode;
and 8: the multi-model prediction control module inputs the current air volume, the coal volume and the NOx concentration output value at the inlet of the denitration device into a corresponding subinterval prediction model as feedforward according to the current load working condition in the control process, then performs feedforward-feedback cooperative control on the ammonia injection volume together with a feedback correction module in the dynamic matrix control model, calculates the ammonia injection optimization volume du of the next control, then transmits the ammonia injection optimization volume to a denitration control object to calculate the valve opening degree and the ammonia injection pump frequency of the ammonia injection device of the denitration device, and finally performs the calculation of the ammonia injection volume du of the next step on the outlet NOx concentration subjected to the ammonia injection control again through the feedback correction module to realize the accurate control of the ammonia injection device.
Preferably, the multi-model prediction control module divides the working conditions in an Euclidean distance clustering mode according to the temperature of the outlet of the hearth and establishes sub-interval models under different working conditions.
Preferably, in the step response model building phase, the first-order inertia system model with pure hysteresis represents:
Figure RE-GDA0003233332100000051
in the formula, KPiRepresents a proportional element, TPiRepresenting a response element, TdiRepresenting a pure hysteresis link, Xi(s) represents system input, Y(s) represents system output, Gi(s) represents the system transfer function;
the method comprises the following steps of obtaining optimal parameters of a first-order inertial system with pure lag in each sub-region by adopting an improved particle swarm optimization algorithm of dynamic inertial weight, and searching a multi-model predictive control module under a specified typical working condition sub-region, wherein the formula for adjusting the dynamic inertial weight is as follows:
Figure BDA0003141929940000052
wherein w is the inertial weight, R is the current iteration number, RmaxIs the maximum iterationThe number of times.
Preferably, in each working condition subinterval model operation interval, setting the model according to actual proportion, inertia and delay parameters of outlet NOx concentration relative to ammonia injection quantity, total air quantity and coal supply quantity input parameters; the set value of the outlet NOx is selected to be 35-45 mg/m3。
The invention has the beneficial effects that:
according to the method, the concentration of NOx at the inlet of the denitration device can be accurately predicted, and then the NOx is input into the multi-model prediction control module under the variable load working condition with the dynamic response matrix as the kernel in a feedforward mode, so that the defects of large inertia, large delay and strong nonlinearity of the control of a denitration system are overcome; under the working condition of large-range variable load, the invention greatly improves the economy and stability of ammonia injection amount control of the denitration device under the condition of ensuring that the concentration of the NOx at the outlet reaches the standard.
Drawings
FIG. 1 is a flow chart of the establishment of a model for predicting NOx concentration at an inlet of a denitration device according to the present invention;
FIG. 2 is a schematic diagram of the present invention;
FIG. 3 is a control effect diagram of the present invention applied to a certain CFB power plant load-up condition;
FIG. 4 is a control effect diagram of the present invention applied to a load reduction condition of a CFB power plant;
FIG. 5 is a control effect diagram of the present invention applied to a CFB power plant under a stable load condition;
FIG. 6 is a diagram showing the long-term control effect of the present invention applied to a CFB power plant and a comparison between the long-term control effect and the original control effect.
Detailed Description
In order to more clearly illustrate the technical solution of the present invention, the following further describes the present invention with reference to the accompanying drawings and examples, but the scope of the present invention is not limited thereto. In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present invention. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without one or more of these specific details. It is to be understood that the practice of the invention is not limited to the following examples, and that various changes and/or modifications may be made without departing from the scope of the invention.
Example 1
Referring to fig. 1 and 2, the invention provides an inlet NO-based boiler unit for the characteristics of large-range variable-load working condition operation of a CFB boiler unitxThe accurate control system for the ammonia spraying of the denitration device with accurate concentration prediction can accurately control the ammonia spraying device; the ammonia injection accurate control system comprises a power station information system, a denitration device inlet NOx concentration prediction model, a multi-model prediction control module and a denitration device control object;
when a unit is controlled in real time, a NOx concentration prediction model at an inlet of a denitration device is in communication connection with the OPC server of the power plant, and DCS data are transmitted to the NOx concentration prediction model at the inlet of the denitration device in real time; calculating an outlet NOx predicted value at the current moment in advance of a CEMS (continuous emission monitoring System) meter by using a denitration device inlet NOx concentration prediction model, and inputting the outlet NOx predicted value into a multi-model prediction control module as feedforward;
the multi-model prediction control module takes dynamic matrix control as an inner core, takes a NOx concentration prediction model at the inlet of the denitration device as feedforward, and divides different subinterval models through typical working conditions to adapt to strong nonlinearity of the denitration system, so that the accurate control of the ammonia injection device under the large-range variable load working condition is realized;
the control object of the denitration device comprises the opening degree of a valve of an ammonia spraying device of the denitration device and the frequency of an ammonia spraying pump, wherein the opening degree of the valve and the flow rate of the ammonia water have a nonlinear relation.
The denitration device inlet NOx concentration prediction model is a denitration device inlet NOx concentration prediction model established based on a long-term short-term memory neural network (LSTM) algorithm; aiming at the problems of inaccurate measurement and large hysteresis of an inlet NOx concentration meter of a denitration device of a CFB boiler, the model carries out data clustering on the historical typical operation condition of the total coal feeding quantity of the boiler, establishes a global LSTM neural network prediction model suitable for the large-range variable load variable condition of the boiler after optimizing the super parameters, eliminates the measurement hysteresis error of a CEMS system, and carries out accurate prediction on the inlet NOx concentration of the denitration system; the denitration device inlet NOx concentration prediction model is modeled in a mode of combining mechanism and data, and the establishment steps are as follows:
(1) data were retrieved from the database: analyzing the running condition of the CFB unit (analyzing the mechanism of generating and removing NOx of the CFB boiler) in a mechanism, selecting parameters influencing the generation and removal of the NOx as model input characteristic variables, and taking a NOx concentration predicted value at an inlet of a denitration device as a model output value; the characteristic variables influencing the generation and removal of NOx comprise coal supply quantity, primary air quantity, secondary air quantity, hearth temperature, hearth outlet smoke temperature, main steam flow and smoke O2Volume percent;
(2) mutual information characteristic quantity selection: analyzing the relationship between the selected input characteristic variable and the output value by using a mutual information statistical method, calculating a mutual information value between the input characteristic variable and the output value, and forming a judgment logic; when a mutual information value between an input characteristic variable and an output value is calculated, a minimum redundancy maximum correlation algorithm and empirical knowledge are used as evaluation indexes, and an evaluation formula is as follows:
Figure BDA0003141929940000071
where x is the input feature variable, y is the output value, and S is the feature { x }iThe set of the (I) is the mutual information value between x and y;
screening out input features with strong relevance according to an evaluation formula, deleting input features with low mutual information values, and reserving input feature variables with high mutual information values;
(3) and (3) data standardization treatment: because of the difference of orders of magnitude between partial input features, in order to avoid influencing the updating speed of the model weight, the input features need to be standardized, and the standardized feature distribution is normal distribution with 0 as the mean value and 1 as the standard deviation; the normalization process formula is as follows:
Figure BDA0003141929940000081
in the formula, x is an input characteristic variable, mu is a mean value of the input characteristic variable, and sigma is a standard deviation of the input characteristic variable;
(4) determining characteristic quantity delay through wavelet transformation: entry NO due to input characteristics determined by the control modelXThe concentration influence has the time sequence, so a 'time-scale (time-frequency) analysis method of signals' is introduced, namely a wavelet transformation method is introduced to determine the delay between input/output characteristics, eliminate the signal fluctuation caused by the fluctuation of an instrument, link the data of an original database with different input/output characteristics, determine the response time difference between different characteristics, and finally determine the response time difference between history and real-time data and different input/output characteristics, wherein the wavelet transformation formula is as follows:
Figure BDA0003141929940000082
where a is a scale factor, b is a translation factor, < f (t), ψa，b(t)>Representing the inner product of two functions, representing the taking of the conjugate;
(5) analyzing data conditions: identifying the working condition of the historical data of the selected characteristic quantity, screening out a stable working condition, a purging working condition and a variable load working condition, clustering and segmenting according to the coal feeding quantity aiming at the variable load working condition, and selecting Euclidean distance clustering in a clustering mode;
considering that different boilers, operation conditions and coal quality conditions have large influence on the NOx emission characteristics of the boilers, analyzing the instantaneous total coal supply quantity which can represent the operation characteristics of the boilers by an Euclidean clustering method, finding a plurality of typical working conditions of system operation, and dividing a subinterval model according to a plurality of typical working conditions, (namely a variable working condition LSTM model I, a variable working condition LSTM model II and a variable working condition LSTM model III) to develop an inlet NOx concentration prediction model under a large-range variable load working condition based on the subinterval model;
(6) selecting an LSTM algorithm as a basis for modeling, establishing a NOx concentration prediction model of a denitration device inlet based on an LSTM neural network principle, and determining four hyper-parameters of prediction time step (prediction time step), backtracking time step (look-back time step), hidden layer number (hidden layers) and hidden layer node number (hidden node); the prediction time step is determined by actual prediction requirements and accuracy together, the backtracking time step is an optimal solution obtained by pre-training under the same network structure, and the number of hidden layers and the number of hidden layer nodes are obtained by optimizing the network hidden layer structure by adopting a particle swarm optimization algorithm; after the four super parameters are determined, a NOx concentration prediction model of the inlet of the denitration device can be established;
(7) after an LSTM-based denitration device inlet NOx concentration prediction model is established, the super-parameters are adjusted, and then the model accuracy is verified.
The off-line LSTM denitration device inlet NOx concentration prediction model established in the method has a good prediction effect, can accurately predict the denitration device inlet NOx concentration in advance under the large-range variable load working condition, and reduces or even eliminates the hysteresis caused by CEMS meter delay.
Aiming at the problems of inaccurate measurement, large hysteresis and the like of a NOx concentration meter at the inlet of a boiler denitration device, the invention clusters data of historical typical operation working conditions of the total coal feeding quantity of the boiler, establishes a global LSTM neural network prediction model adapting to the variable working conditions of the boiler, and measures and corrects the NOx concentration at the inlet of the denitration device; meanwhile, a multi-model predictive control module based on the denitration device inlet NOx concentration measurement correction as feedforward is established, wherein the multi-model predictive control module takes Dynamic Matrix Control (DMC) as an inner core and mainly comprises three parts of model prediction, rolling optimization and feedback correction.
Particularly, the multi-model predictive control module adds a feedforward part on the basis of dynamic matrix control, inputs an output value of a NOx concentration prediction model at the inlet of the denitration device into the multi-model predictive control module as feedforward, and performs cooperative control on ammonia injection amount together with feedback correction.
Particularly, the multi-model prediction control module establishes a plurality of subinterval prediction models based on different load working condition sections such as full load, high load, medium load and low load, and guarantees the stability of the NOx control effect at the outlet of the denitration system under the large-range variable load working condition in a weighting switching mode under different load working conditions.
The method for accurately controlling the ammonia spraying of the denitration device based on the accurate concentration prediction of the system is adopted, the accurate ammonia spraying control system is embedded in a denitration DCS in a module mode and is in real-time communication with the denitration DCS, and when the combustion condition of a boiler changes, the output value of an inlet NOx concentration prediction model of the denitration device is used as feedforward to be input into a multi-model prediction control module; meanwhile, the multi-model prediction control module identifies the current load working condition on line, carries out feedback correction on the ammonia injection amount at the current moment, carries out rolling optimization on the ammonia injection amount by combining the set value of the outlet NOx through the set feedforward feedback proportion, and sends a control instruction to the control object of the denitration device to increase or reduce the ammonia injection amount, thereby realizing the emission control on the concentration of the outlet NOx.
The ammonia injection accurate control method specifically comprises the following steps:
step 1: firstly, establishing an ammonia injection accurate control system, which comprises a server, an upper computer, a DO/IO interface, a control cabinet, a data acquisition instrument, a data transmitter, a data processing center, a newly-added online monitoring instrument (optional) and the like, so as to form an online control system, establishing real-time communication with an original DCS (distributed control system) of a denitration device, and embedding the ammonia injection accurate control system into the denitration DCS in a module manner;
step 2: after an ammonia injection accurate control system hardware platform is constructed, an NOx concentration prediction model at the inlet of the denitration device is established based on an algorithm technology; the method for establishing the denitration device inlet NOx concentration prediction model is as described in embodiment 1;
and step 3: after the establishment of a denitration device inlet NOx concentration prediction model is completed, an optimal output value is obtained according to input characteristics, the influence of time difference of each input characteristic on the output characteristics is eliminated, the optimal prediction effect is obtained, and the requirement as control feedforward is met;
and 4, step 4: the multi-model predictive control comprises an outlet NOx feedforward predictive model and an outlet NOx control model, wherein the outlet NOx feedforward predictive model is the step response relation of air volume, coal volume and inlet NOx concentration to outlet NOx, and the outlet NOx control model is the step response relation of ammonia injection volume/air volume to outlet NOx concentration; based on the mutual information screening and screening results, the optimal air volume, the coal volume and the inlet NOx concentration are used as input parameters of a feedforward model for controlling the outlet NOx; preferably, the ammonia injection amount/air volume is used as an input parameter of an outlet NOx control quantity model;
and 5: selecting dynamic matrix predictive control as a basic control strategy of a multi-model predictive control module, wherein a prediction model in the dynamic matrix predictive control is a step response model, can be represented by a first-order inertial system model with pure lag, and obtains optimal parameters of the first-order inertial system with pure lag in each subarea by a particle swarm algorithm;
step 6: step 5, after a multi-model predictive control module is established, further adjusting parameters of the step response model in a field debugging mode; during field debugging, the working condition changes, and the main change amount in the model is a proportion link KpiAnd respond to the link TpiAnd a hysteresis loop TdiRemain substantially unchanged;
and 7: the multi-model prediction control module establishes a plurality of subinterval prediction models based on different load working condition sections such as full load, high load, medium load, low load and the like, and ensures the stability of the NOx control effect of the outlet of the denitration system under the large-range variable load working condition in a weighting switching mode;
and 8: the multi-model prediction control module takes the current air volume, the coal volume and the NOx concentration output value at the inlet of the denitration device as feedforward in the control process, inputs the feedforward into a corresponding subinterval prediction model according to the current load working condition, then performs feedforward-feedback cooperative control on the ammonia injection volume together with a feedback correction module in the dynamic matrix control model, calculates the ammonia injection optimization volume of the next control, transmits the ammonia injection optimization volume to a denitration control object to calculate the valve opening degree and the ammonia injection pump frequency of the denitration device ammonia injection device, and finally performs the calculation of the next ammonia injection volume du on the outlet NOx concentration subjected to the ammonia injection control again through the feedback correction module, so that the accurate control of the ammonia injection device is realized.
The multi-model prediction control module takes dynamic matrix control as an inner core, and controls the ammonia injection amount of the ammonia injection device of the denitration system by overlapping step responses of all input parameters.
And the multi-model prediction control module divides working conditions in an Euclidean distance clustering mode according to the temperature of the hearth outlet and establishes sub-interval models under different working conditions. It is worth noting that the step responses of the input parameters of the prediction models in different subintervals are different, and parameter setting needs to be performed through a particle swarm algorithm or field debugging.
Preferably, in the step response model building phase, the first-order inertia system model with pure hysteresis represents:
Figure RE-GDA0003233332100000111
in the formula, KPiRepresents a proportional element, TPiRepresenting a response element, TdiRepresenting a pure hysteresis link, Xi(s) represents system input, Y(s) represents system output, Gi(s) represents the system transfer function;
the method comprises the following steps of obtaining optimal parameters of a first-order inertial system with pure lag in each sub-region by adopting an improved particle swarm optimization algorithm of dynamic inertial weight, and searching a multi-model predictive control module under a specified typical working condition sub-region, wherein the formula for adjusting the dynamic inertial weight is as follows:
Figure BDA0003141929940000112
wherein w is the inertial weight, R is the current iteration number, RmaxIs the maximum number of iterations.
Furthermore, in the actual on-line control of the power plant with the multi-model predictive control module, the operation interval of the sub-model of each working condition is controlled according to the output NOxThe model is set by the actual proportion, inertia and delay parameters of the concentration relative to the input parameters such as ammonia injection quantity, total air quantity and coal feeding quantitySo as to achieve the optimal control effect under the working condition of large-range variable load.
Further, outlet NOxThe set value is selected to be 35-45 mg/m3To ensure that the concentration of the water is not more than 50mg/m3NO ofxUnder the condition of the emission concentration limit value, the ammonia injection amount is reduced to the maximum extent, and the ammonia injection economy of the denitration system is improved.
The method adopts a method of combining data and mechanism, selects key characteristics by analyzing important influence factors influencing NOx generation, establishes an online denitration device inlet NOx concentration measurement correction model based on a long-short term memory neural network (LSTM) algorithm, predicts the NOx concentration in advance through boiler data, reduces or even eliminates CEMS concentration measurement lag error, and realizes the inlet NO of the denitration devicexAnd (4) accurate prediction of concentration.
In the process of on-line control of a power plant, an inlet NOx concentration output value output by an inlet NOx concentration prediction model and parameters such as primary air volume, coal feeding amount and the like which reflect combustion changes at the boiler side are input into a multi-model prediction control module as feed-forward in real time, and the feed-forward and the feedback correction module of a dynamic matrix model cooperatively control the ammonia injection amount of a denitration device, so that the accurate control of the ammonia injection amount of the denitration device is realized.
Example 2
The invention is described in detail by taking the actual control flow and short-term and long-term control effects of a certain 220t/h circulating fluidized bed pulverized coal furnace as an example:
(1) selecting 4 coal feeding amounts, 2 primary air amounts, 2 secondary air amounts, 10 hearth temperature, 2 hearth outlet smoke temperatures, 1 main steam flow and 1 smoke O2The volume percentage is 22 input characteristic variables, clustering is carried out based on the running coal feeding amount data of the last two months, and the clustering centers are determined to be 12.1t/h, 18.0t/h and 26.0 t/h; inputting the data and the processed data into an LSTM model, selecting a 2-layer hidden layer with a prediction time step of 14 and a backtracking time step of 30, and 128 hidden layer nodes in each layer; establishing an LSTM model according to the super-parameter configuration to predict the NOx concentration at the inlet of the denitration device;
(2) determining that M is 1, P is 80, R is 5000, Q is 0.05, the feedforward-feedback ratio is 0.01, and identifying the step response parameters of each subinterval model by adopting an improved particle swarm optimization method; and (3) carrying out load increase, load reduction, load stabilization and long-time control effect verification on the multi-model predictive control module, wherein the effects are shown in figures 3-6.
As shown in fig. 3, under the load-increasing working condition, along with the increase of the load, the ammonia injection amount is relatively high under the original control method, and the ammonia escape phenomenon is relatively serious; different from the original control method, the multi-model prediction control module has the function of predicting the concentration of NOx, can accurately judge the concentration change trend, slowly improves the ammonia injection amount along with the increase of the boiler load, and ensures that the ammonia escape amount is stably at a lower level; in addition, the fluctuation range of the outlet NOx concentration is smaller, and the standard deviation is reduced by 79.3 percent compared with the standard deviation of the original control method.
As shown in fig. 4, under the load-reducing condition, the ammonia injection amount required by the system is gradually reduced to reduce the size of the ammonia escape at the outlet, so as to avoid excessive ammonia injection; under the original control method, through the down regulation of the ammonia injection quantity for many times, the concentration of the NOx at the outlet presents an ascending trend and an overproof state appears; compared with the existing control method, the multi-model prediction control module has the advantages that the average value of the outlet NOx concentration is higher, and the ammonia injection amount is reduced more smoothly, so that the multi-model prediction control module can regulate the ammonia injection amount more clearly in the load reduction process, and the outlet NOx concentration is maintained in a certain range.
Under the working condition of conventional fluctuating load, under the two control methods of original control and multi-model predictive control, the ammonia injection amount, the outlet NOx concentration and the ammonia escape change with time as shown in figure 5; compared with the original control, the change of the ammonia injection amount under the multi-model prediction control module is relatively more stable; intelligently controlling the average value of outlet NOx to be closer to the set concentration value (40 mg/m)3) The fluctuation is small, and the standard deviation is reduced by 90.49% compared with the standard deviation of the original control method; this is because the multi-mode type prediction control module has a prediction effect and can react to the inlet NOx concentration variation in advance.
In a 220t/h coal-fired thermoelectric unit, the typical working conditions of the unit are respectively controlled by using the original control module and the multi-model predictive control moduleThe control is carried out, the load of the boiler shows periodic variation of high day and low night according to the steam supply requirement, a complete cycle is formed every 24 hours, and in order to verify the working condition under a long time, 48 hours (two complete cycles) are selected for carrying out the analysis of main parameters. FIG. 6 shows the trend of steam mass, ammonia injection mass, ammonia slip and outlet NOx concentration for two control regimes. As can be seen from the figure, the inlet steam flow concentration of the original control module is basically the same as that of the multi-model predictive control module, and the working condition load is proved to be basically the same; but the ammonia injection amount of the original control is relatively large, the ammonia escape rate is relatively high, and the concentration fluctuation of the outlet NOx is obvious; the ammonia injection amount of the multi-model prediction control module is stabilized below 150L/h, so that the ammonia water consumption in the denitration process is reduced; meanwhile, the ammonia escape rate is close to zero, so that the problem of equipment corrosion caused by ammonia escape is avoided; the concentration of the NOx at the outlet is stabilized at 40mg/m3And the amplitude fluctuation is small, so that the use amount of ammonia water and the ammonia escape rate of the multi-model predictive control module can be reduced on the premise of ensuring the stable concentration of outlet NOx, and the multi-model predictive control module is beneficial to reducing the material consumption in the denitration process and avoiding equipment corrosion.
The present invention is described in detail with reference to the embodiments, but the description is only for the specific embodiments of the present invention, and therefore, the present invention should not be construed as limiting the scope of the present invention. It should be noted that, for those skilled in the art, variations and modifications made within the scope of the present invention shall fall within the scope of the claims of the present invention without departing from the spirit of the present invention.

Claims (10)

1. The utility model provides an accurate control system of denitrification facility ammonia spraying based on accurate prediction of concentration which characterized in that: the ammonia injection accurate control system comprises a power station information system, a denitration device inlet NOx concentration prediction model, a multi-model prediction control module and a denitration device control object;
the power station information system comprises a power plant OPC server and DCS control equipment, is in communication connection with the inlet NOx concentration prediction model of the denitration device, and transmits DCS data to the inlet NOx concentration prediction model of the denitration device in real time; calculating an outlet NOx predicted value at the current moment in advance of a pollutant emission continuous monitoring system meter by using a denitration device inlet NOx concentration prediction model, and inputting the predicted value into a multi-model prediction control module as a feedforward;
the multi-model prediction control module takes dynamic matrix control as an inner core, takes a NOx concentration prediction model at the inlet of the denitration device as feed-forward, and divides different subinterval models through typical working conditions;
the control object of the denitration device comprises the opening degree of a valve of the ammonia spraying device of the denitration device and the frequency of an ammonia spraying pump.
2. The accurate control system of denitration device ammonia injection based on accurate concentration prediction of claim 1, characterized in that: the denitration device inlet NOx concentration prediction model is established based on a long-term and short-term memory neural network algorithm; the denitration device inlet NOx concentration prediction model is established through the following steps:
(1) analyzing the operation condition of the boiler unit in a mechanism, selecting parameters influencing NOx generation and removal as model input characteristic variables, and taking a NOx concentration predicted value at an inlet of a denitration device as a model output value;
(2) analyzing the relationship between the selected input characteristic variable and the output value by using a mutual information statistical method, calculating a mutual information value between the input characteristic variable and the output value, and forming a judgment logic;
(3) input features are normalized, and the normalization processing formula is as follows:
Figure FDA0003141929930000011
in the formula, x is an input characteristic variable, mu is a mean value of the input characteristic variable, and sigma is a standard deviation of the characteristic;
(4) introducing a wavelet transform method to determine the delay between input/output characteristics, establishing a relation between data of an original database and different input/output characteristics, determining the response time difference between different characteristics, and finally determining the response time difference between historical and real-time data and different input/output characteristics;
(5) identifying the working condition of the historical data of the selected characteristic variable, screening out a stable working condition, a purging working condition and a variable load working condition, and clustering and segmenting according to the coal feeding amount according to the variable load working condition;
(6) modeling is carried out on the basis of an LSTM algorithm, a NOx concentration prediction model at the inlet of the denitration device is established on the basis of the LSTM principle, and a super parameter is determined; after the super-parameters are determined, a NOx concentration prediction model of the inlet of the denitration device can be established.
3. The accurate control system of denitration device ammonia injection based on accurate prediction of concentration of claim 2, characterized in that: the input characteristic variables comprise coal feeding quantity, primary air quantity, secondary air quantity, hearth temperature, hearth outlet smoke temperature, main steam flow and flue gas O2Volume percent; the hyper-parameters comprise a prediction time step, a backtracking time step, the number of hidden layers and the number of nodes of the hidden layers.
4. The accurate control system of denitration device ammonia injection based on accurate prediction of concentration of claim 2, characterized in that: when a mutual information value between an input characteristic variable and an output value is calculated, a minimum redundancy maximum correlation algorithm and empirical knowledge are used as evaluation indexes, and an evaluation formula is as follows:
mRMR:
Figure FDA0003141929930000021
where x is the input feature variable, y is the output value, and S is the feature { x }iAnd I is a mutual information value between x and y.
5. The accurate control system of denitration device ammonia injection based on accurate prediction of concentration of claim 2, characterized in that: after a denitration device inlet NOx concentration prediction model is established, the super-parameters are adjusted, and then the model accuracy is verified.
6. The method for accurately controlling ammonia spraying of the denitration device based on accurate concentration prediction is characterized by comprising the following steps of: embedding the ammonia injection accurate control system of any one of claims 1 to 5 in a denitration DCS in a module form, carrying out real-time communication with the denitration DCS, and inputting input characteristic variables and output values of a denitration device inlet NOx concentration prediction model into a multi-model prediction control module as feedforward when the combustion condition of a boiler changes; meanwhile, the multi-model predictive control module identifies the current load working condition on line, carries out feedback correction on the ammonia injection amount at the current moment, carries out rolling optimization on the ammonia injection amount by combining the set value of the outlet NOx through the set feedforward feedback proportion, and sends a control instruction to the control object of the denitration device to increase or reduce the ammonia injection amount, thereby realizing the emission control of the outlet NOx concentration.
7. The method for accurately controlling ammonia injection of the denitration device based on accurate concentration prediction as claimed in claim 6, wherein: the method specifically comprises the following steps:
step 1: firstly, establishing an ammonia injection accurate control system, establishing real-time communication with an original DCS (distributed control System) of a denitration device, and embedding the ammonia injection accurate control system in the denitration DCS in a module form;
step 2: after an ammonia injection accurate control system hardware platform is constructed, an NOx concentration prediction model at the inlet of the denitration device is established based on an algorithm technology;
and step 3: after the establishment of a denitration device inlet NOx concentration prediction model is completed, an optimal output value is obtained according to input characteristic variables, the influence of time difference of each input characteristic on the output characteristic is eliminated, the optimal prediction effect is obtained, and the requirement of controlling feedforward is met;
and 4, step 4: the multi-model predictive control comprises an outlet NOx concentration feedforward predictive model and an outlet NOx concentration control model, wherein the outlet NOx concentration feedforward predictive model is a step response relation of air volume, coal volume and inlet NOx concentration to outlet NOx concentration, and the outlet NOx concentration control model is a step response relation of ammonia injection volume/air volume to outlet NOx concentration;
and 5: selecting dynamic matrix predictive control as a basic control strategy of a multi-model predictive control module, wherein a prediction model in the dynamic matrix predictive control is a step response model;
step 6: step 5, after a multi-model predictive control module is established, further adjusting parameters of the step response model in a field debugging mode;
and 7: the multi-model prediction control module establishes a plurality of subinterval prediction models based on different load working condition sections, and ensures the stability of the NOx control effect at the outlet of the denitration device under the large-range variable load working condition in a weighting switching mode;
and 8: the multi-model prediction control module takes the current air volume, the coal volume and the NOx concentration output value at the inlet of the denitration device as feedforward in the control process, inputs the feedforward into a corresponding subinterval prediction model according to the current load working condition, then performs feedforward-feedback cooperative control on the ammonia injection volume together with a feedback correction module in the dynamic matrix control model, calculates the ammonia injection optimization volume of the next control, transmits the ammonia injection optimization volume to a denitration control object to calculate the valve opening degree and the ammonia injection pump frequency of the denitration device, and finally performs the next ammonia injection volume calculation again on the outlet NOx concentration subjected to the ammonia injection control through the feedback correction module, so that the accurate control of the ammonia injection device is realized.
8. The method for accurately controlling ammonia injection of the denitration device based on accurate concentration prediction as claimed in claim 7, wherein: and the multi-model prediction control module divides working conditions in an Euclidean distance clustering mode according to the temperature of the hearth outlet and establishes subinterval models under different working conditions.
9. The method for accurately controlling ammonia injection of the denitration device based on accurate concentration prediction as claimed in claim 7 or 8, wherein: in the step response model establishing stage, the first-order inertia system model with pure lag represents that:
Figure RE-FDA0003233332090000031
in the formula, KPiRepresents a proportional element, TPiRepresenting a response element, TdiRepresenting a pure hysteresis link, Xi(s) represents system input, Y(s) represents system output, Gi(s) represents the system transfer function;
the method comprises the following steps of obtaining optimal parameters of a first-order inertia system with pure lag in each subarea by adopting an improved particle swarm optimization algorithm of dynamic inertia weight, and searching a multi-model predictive control module under a defined typical working condition subarea, wherein the formula of dynamic inertia weight adjustment is as follows:
Figure RE-FDA0003233332090000041
wherein w is the inertial weight, R is the current iteration number, RmaxIs the maximum number of iterations.
10. The method for accurately controlling ammonia injection of the denitration device based on accurate concentration prediction as claimed in claim 8, wherein: in each working condition subinterval model operation interval, setting the model according to actual proportion, inertia and delay parameters of outlet NOx concentration relative to ammonia injection quantity, total air quantity and coal supply quantity input parameters; the set value of the outlet NOx is selected to be 35-45 mg/m3。
CN202110736767.0A 2021-06-30 2021-06-30 Accurate control system and method for ammonia spraying of denitration device based on accurate concentration prediction Active CN113433911B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110736767.0A CN113433911B (en) 2021-06-30 2021-06-30 Accurate control system and method for ammonia spraying of denitration device based on accurate concentration prediction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110736767.0A CN113433911B (en) 2021-06-30 2021-06-30 Accurate control system and method for ammonia spraying of denitration device based on accurate concentration prediction

Publications (2)

Publication Number Publication Date
CN113433911A true CN113433911A (en) 2021-09-24
CN113433911B CN113433911B (en) 2022-05-20

Family

ID=77758188

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110736767.0A Active CN113433911B (en) 2021-06-30 2021-06-30 Accurate control system and method for ammonia spraying of denitration device based on accurate concentration prediction

Country Status (1)

Country Link
CN (1) CN113433911B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113856457A (en) * 2021-09-27 2021-12-31 京能（锡林郭勒）发电有限公司 NOx emission control system for low-calorific-value lignite
CN114609986A (en) * 2022-03-16 2022-06-10 中国中材国际工程股份有限公司 Cement decomposing furnace denitration regulation and control optimization system and method based on predictive control
CN115155310A (en) * 2022-07-13 2022-10-11 浙江大学 Ammonia spraying accurate optimization method for SCR denitration system
CN115309129A (en) * 2022-10-11 2022-11-08 华电电力科学研究院有限公司 SCR denitration efficiency automatic optimization regulation and control method and system
CN115598974A (en) * 2022-09-02 2023-01-13 南京天洑软件有限公司（Cn） Denitration system model prediction control method and device based on linear system identification
CN116253446A (en) * 2023-03-24 2023-06-13 青岛思普润水处理股份有限公司 Intelligent aeration setting method for sewage treatment
CN115155310B (en) * 2022-07-13 2024-04-26 浙江大学 SCR denitration system ammonia spraying accurate optimization method

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106681381A (en) * 2017-01-03 2017-05-17 华北电力大学 SCR denitration system ammonia spraying quantity optimal control system and method based on intelligent feedforward signals
CN106842962A (en) * 2017-04-13 2017-06-13 东南大学 Based on the SCR denitration control method for becoming constraint multiple model predictive control
CN108628177A (en) * 2018-07-02 2018-10-09 大唐环境产业集团股份有限公司 A kind of SCR denitration intelligence spray ammonia optimization method and system based on model adaptation PID
CN108803309A (en) * 2018-07-02 2018-11-13 大唐环境产业集团股份有限公司 It is a kind of that ammonia optimization method and system are intelligently sprayed based on the SCR denitration of hard measurement and model adaptation
CN108837699A (en) * 2018-07-02 2018-11-20 大唐环境产业集团股份有限公司 It is a kind of that ammonia optimization method and system are intelligently sprayed based on the SCR denitration of hard measurement and PREDICTIVE CONTROL
CN109062053A (en) * 2018-08-31 2018-12-21 江苏国信靖江发电有限公司 A kind of denitration spray ammonia control method based on multivariate calibration
CN112221347A (en) * 2020-08-11 2021-01-15 华电电力科学研究院有限公司 Accurate ammonia injection control method for SCR denitration system
CN112418284A (en) * 2020-11-16 2021-02-26 华北电力大学 Control method and system for SCR denitration system of full-working-condition power station
CN112580250A (en) * 2020-11-12 2021-03-30 山东纳鑫电力科技有限公司 Thermal power generating unit denitration system based on deep learning and optimization control method
CN112619394A (en) * 2020-11-24 2021-04-09 呼和浩特科林热电有限责任公司 Denitration ammonia injection self-adaptive control method and device and denitration system

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106681381A (en) * 2017-01-03 2017-05-17 华北电力大学 SCR denitration system ammonia spraying quantity optimal control system and method based on intelligent feedforward signals
CN106842962A (en) * 2017-04-13 2017-06-13 东南大学 Based on the SCR denitration control method for becoming constraint multiple model predictive control
CN108628177A (en) * 2018-07-02 2018-10-09 大唐环境产业集团股份有限公司 A kind of SCR denitration intelligence spray ammonia optimization method and system based on model adaptation PID
CN108803309A (en) * 2018-07-02 2018-11-13 大唐环境产业集团股份有限公司 It is a kind of that ammonia optimization method and system are intelligently sprayed based on the SCR denitration of hard measurement and model adaptation
CN108837699A (en) * 2018-07-02 2018-11-20 大唐环境产业集团股份有限公司 It is a kind of that ammonia optimization method and system are intelligently sprayed based on the SCR denitration of hard measurement and PREDICTIVE CONTROL
CN109062053A (en) * 2018-08-31 2018-12-21 江苏国信靖江发电有限公司 A kind of denitration spray ammonia control method based on multivariate calibration
CN112221347A (en) * 2020-08-11 2021-01-15 华电电力科学研究院有限公司 Accurate ammonia injection control method for SCR denitration system
CN112580250A (en) * 2020-11-12 2021-03-30 山东纳鑫电力科技有限公司 Thermal power generating unit denitration system based on deep learning and optimization control method
CN112418284A (en) * 2020-11-16 2021-02-26 华北电力大学 Control method and system for SCR denitration system of full-working-condition power station
CN112619394A (en) * 2020-11-24 2021-04-09 呼和浩特科林热电有限责任公司 Denitration ammonia injection self-adaptive control method and device and denitration system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JIGAO FU ET.: "Control Strategy for Denitrification Efficiency of Coal-Fired Power Plant Based on Deep Reinforcement Learning", 《SPECIAL SECTION ON ADVANCES IN MACHINE LEARNING AND COGNITIVE COMPUTING FOR INDUSTRY APPLICATIONS》 *
盛锴 等: "基于改进型多模型预测控制的喷氨优化算法及其应用", 《热能动力工程》 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113856457A (en) * 2021-09-27 2021-12-31 京能（锡林郭勒）发电有限公司 NOx emission control system for low-calorific-value lignite
CN113856457B (en) * 2021-09-27 2024-04-02 京能（锡林郭勒）发电有限公司 NOx emission control system for low-heat-value lignite
CN114609986A (en) * 2022-03-16 2022-06-10 中国中材国际工程股份有限公司 Cement decomposing furnace denitration regulation and control optimization system and method based on predictive control
CN115155310A (en) * 2022-07-13 2022-10-11 浙江大学 Ammonia spraying accurate optimization method for SCR denitration system
CN115155310B (en) * 2022-07-13 2024-04-26 浙江大学 SCR denitration system ammonia spraying accurate optimization method
CN115598974A (en) * 2022-09-02 2023-01-13 南京天洑软件有限公司（Cn） Denitration system model prediction control method and device based on linear system identification
CN115598974B (en) * 2022-09-02 2024-02-20 南京天洑软件有限公司 Denitration system model predictive control method and device based on linear system identification
CN115309129A (en) * 2022-10-11 2022-11-08 华电电力科学研究院有限公司 SCR denitration efficiency automatic optimization regulation and control method and system
WO2023221446A1 (en) * 2022-10-11 2023-11-23 华电电力科学研究院有限公司 Scr denitrification efficiency automatic optimization regulation and control method and system
CN116253446A (en) * 2023-03-24 2023-06-13 青岛思普润水处理股份有限公司 Intelligent aeration setting method for sewage treatment
CN116253446B (en) * 2023-03-24 2024-01-30 青岛思普润水处理股份有限公司 Intelligent aeration setting method for sewage treatment

Also Published As

Publication number Publication date
CN113433911B (en) 2022-05-20

Similar Documents

Publication Publication Date Title
CN113433911B (en) Accurate control system and method for ammonia spraying of denitration device based on accurate concentration prediction
CN109062053B (en) Denitration ammonia injection control method based on multivariate correction
CN110368808B (en) Ammonia spraying amount control method and system for SCR flue gas denitration system
CN105629738B (en) SCR flue gas denitrification systems control method and equipment
CN104826492B (en) Improvement method for selective catalytic reduction flue gas denitrification and ammonia injection control system
CN112580250A (en) Thermal power generating unit denitration system based on deep learning and optimization control method
CN107526292B (en) A method of the regulation ammonia spraying amount based on inlet NOx concentration prediction
CN105786035B (en) Fired power generating unit SCR denitration Optimal Control System based on heuristic Prediction and Control Technology
CN110263395A (en) The power plant&#39;s denitration running optimizatin method and system analyzed based on numerical simulation and data
CN111897373B (en) Model prediction-based ammonia injection flow adjusting method for SCR denitration device
CN111829003A (en) Power plant combustion control system and control method
CN110263452A (en) Flue gas Annual distribution characteristic analysis method, system and denitrating system in a kind of flue
CN104715142A (en) NOx emission dynamic soft-sensing method for power station boiler
CN112742187A (en) Method and device for controlling pH value in desulfurization system
CN114721263B (en) Intelligent regulation and control method for cement decomposing furnace based on machine learning and intelligent optimization algorithm
Qiao A novel online modeling for NOx generation prediction in coal-fired boiler
CN107561944A (en) A kind of denitrating system adaptive prediction control method based on Laguerre model
CN112613237B (en) CFB unit NOx emission concentration prediction method based on LSTM
CN109833773A (en) A kind of NO_x Reduction by Effective ammonia flow accuracy control method
CN116720446B (en) Method for monitoring thickness of slag layer of water-cooled wall of boiler in real time
JPH08339204A (en) Autonomous adaptive optimization control system for thermal power station
CN109933884B (en) Neural network inverse control method for SCR denitration system of coal-fired unit
CN116036849A (en) CFB boiler flue gas denitration automatic control method and system
CN113488111B (en) Ammonia injection amount optimization modeling method for SCR denitration system
CN115145152A (en) Boiler combustion and denitration process collaborative optimization control method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant