Intelligent fuel control system and method of coal-fired unit for flexible peak regulation

Abstract

The invention relates to the technical field of unit fuel control, in particular to a coal-fired unit fuel intelligent control system and method oriented to flexible peak regulation. The method comprises the steps that an online sensing unit monitors and normalizes coal quality and safety environmental protection data in real time; the as-fired coal quality soft measurement unit estimates the real-time heat value of the as-fired coal on line through a mechanism-guided time sequence collaborative network; the intelligent decision-making unit generates a load change rate instruction in combination with a security constraint algorithm according to the peak regulation instruction and the unit state; and the coordination control unit dynamically calculates feed-forward instructions of the coal feeding quantity and the total air quantity based on a multivariable predictive control model, and outputs a final actuating mechanism action instruction through a control distribution algorithm. According to the method, the real-time calorific value on-line identification of the coal as fired is realized by combining the mechanism-guided time sequence collaborative network with the energy balance equation, the dynamic change of the coal quality is accurately captured, and the combustion efficiency fluctuation caused by coal quality perception lag in the traditional control is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unit fuel control, in particular to a coal-fired unit fuel intelligent control system and method for flexible peak regulation. BACKGROUND

[0002] The coal-fired unit is the core equipment of power generation in the power system, and the stability and precision of its fuel supply and combustion control directly affect the unit power generation efficiency, safe operation level and environmental protection emission standard compliance capability. With large-scale grid connection of new energy (wind power, photovoltaic, etc.), the demand for flexible peak regulation of coal-fired units in the power system has significantly increased. Flexible peak regulation requires the unit to quickly respond to the load instruction of the power grid, and to flexibly increase and decrease the load within a wide load range (such as 30%-100% of the rated load), while ensuring that key parameters such as main steam pressure and NOx concentration meet the safety and environmental protection standards.

[0003] In the prior art, the fuel control of the coal-fired unit relies on fixed parameter setting or traditional PID regulation, and the perception of changes in the quality of coal fed into the furnace (such as real-time calorific value) lags behind, which cannot dynamically adapt to changes in combustion efficiency caused by fluctuations in coal quality, and problems such as imbalance of air-coal ratio and incomplete combustion are prone to occur, therefore, a coal-fired unit fuel intelligent control system and method for flexible peak regulation are provided. SUMMARY

[0004] The present application aims to provide a coal-fired unit fuel intelligent control system and method for flexible peak regulation to solve the problem of the prior art that the perception of changes in the quality of coal fed into the furnace (such as real-time calorific value) lags behind, which cannot dynamically adapt to changes in combustion efficiency caused by fluctuations in coal quality, and problems such as imbalance of air-coal ratio and incomplete combustion are prone to occur.

[0005] To achieve the above-mentioned purpose, on the one hand, the present application aims to provide a coal-fired unit fuel intelligent control system for flexible peak regulation, comprising: an online perception unit, which monitors the quality of coal fed into the furnace and unit safety and environmental protection related data in real time through unit sensors, and normalizes the data; a soft measurement unit for coal quality fed into the furnace, which aligns the normalized coal quality fed into the furnace related data by time stamp, and performs online identification and estimation of the real-time calorific value of coal fed into the furnace through a mechanism-guided time sequence collaborative network, to obtain a real-time calorific value estimate; an intelligent decision unit, which receives the peak regulation instruction issued by the power plant and calculates the initial rate value of load change based on the monitored unit load data, and generates a load change rate instruction that fits the actual operation capacity of the unit through a safety constraint algorithm based on the unit safety and environmental protection related data; A coordination control unit receives the load change rate instruction and the real-time heat value estimation, dynamically calculates a feedforward instruction of the coal supply amount and the total air volume through a multivariate predictive control model, and outputs a final action instruction vector through a control distribution algorithm.

[0006] As a further improvement of the technical solution, the online perception unit monitors the associated data of the coal quality entering the furnace and the associated data of the unit safety and environmental protection in real time through the unit sensor, and the specific steps involved in the normalization processing of the data are as follows: The associated data of the coal quality entering the furnace and the associated data of the unit safety and environmental protection are read through the unit sensor, and the sampling period is set to 1s / time; Based on the engineering operation range of each parameter, the upper and lower limits of normalization are set, and the maximum-minimum normalization method is used to normalize the associated data of the coal quality entering the furnace and the associated data of the unit safety and environmental protection, and the data range is unified to [0, 1].

[0007] As a further improvement of the technical solution, in the coal quality entering the furnace soft measurement unit, the specific steps involved in the online identification and estimation of the real-time heat value of the coal entering the furnace through the mechanism-guided time sequence collaborative network are as follows: The normalized associated data of the coal quality entering the furnace are aligned according to the time stamp to obtain time sequence data; A mechanism-guided time sequence collaborative network is constructed, which consists of a time sequence feature extraction layer and a heat value regression layer; the time sequence feature extraction layer uses one-dimensional convolution and long short-term memory network to extract features from the input time sequence data, and the extracted feature vector is denoted as ; Based on the actual monitoring of the coal supply amount and the load data, the theoretical heat value is calculated through the energy balance equation ; The theoretical heat value is concatenated with the feature vector , and input into the heat value regression layer for prediction, and the real-time heat value estimation is output, wherein the heat value regression layer includes a fully connected layer; The mechanism-guided time sequence collaborative network is trained through historical data, and the hybrid loss function used in the training process takes the mean square error of the real-time heat value estimation and the measured heat value as the core fitting term, and combines the mechanism constraint term composed of the deviation between the real-time heat value estimation and the theoretical heat value; wherein the mechanism constraint term is used to avoid network training from falling into pure data-driven overfitting; In online operation, real-time data are collected and input into the trained network, and the real-time heat value estimation of the current coal entering the furnace is output .

[0008] As a further improvement of the technical solution, the intelligent decision unit includes a load instruction analysis module and a running safety constraint module; Among them, the load command parsing module receives peak-shaving commands issued by the power plant and real-time monitored unit load data, and calculates the initial rate of load change based on the boiler's thermal storage characteristics; The safety constraint module receives the initial rate value mentioned above, combines it with the unit's safety and environmental protection related data, and outputs a load change rate command through the safety constraint algorithm.

[0009] As a further improvement to this technical solution, the specific steps involved in calculating the initial rate of load change based on the boiler's thermal storage characteristics in the load command parsing module are as follows: Receive peak-shaving instructions from the power plant and the current actual load, and calculate the load deviation by the difference; Based on the boiler's heat storage characteristics, the upper limit of the safe rate of load change is determined, thus obtaining the maximum allowable rate of unit load change. Based on the load deviation and the maximum permissible rate, if the absolute value of the load deviation does not exceed the product of the maximum permissible rate and the sampling time, the initial rate value is calculated by dividing the load deviation by the sampling time; if it exceeds the product, the initial rate value is the maximum permissible rate with the same sign as the load deviation.

[0010] As a further improvement to this technical solution, the specific steps involved in outputting the load change rate command through the safety constraint algorithm in the operation safety constraint module are as follows: Based on the rated pressure of the unit, the upper limit of the safe range of the main steam pressure is set. and lower limit Set upper limits for NOx concentrations in accordance with national environmental protection standards. : If the main steam pressure is close to the boundary of the safe range, the initial rate value is multiplied by the pressure safety factor to obtain the intermediate rate value. ; If the main steam pressure is not close to the boundary of the safe range, and the NOx concentration exceeds the upper limit standard, the initial rate value is multiplied by the emission penalty factor to obtain the load change rate command. ; If the main steam pressure is close to the boundary of the safe range and the NOx concentration exceeds the upper limit standard, the intermediate rate value is multiplied by the emission penalty factor to obtain the load change rate command. ; If the real-time monitored main steam pressure and NOx concentration are both within the safe threshold, the initial rate value will be directly used as the final load change rate command. Output.

[0011] As a further improvement to this technical solution, the coordination and control unit includes a predictive control module and a control command allocation module; The prediction control module receives the load change rate instruction and the real-time heat value estimation value, and dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable prediction control model. The control instruction distribution module converts the feedforward instruction into a final action instruction vector that can be directly issued to the bottom layer execution mechanism through a control distribution algorithm. The final action instruction vector includes a coal feeder speed instruction, a primary air fan opening degree instruction, and a secondary air door opening degree instruction.

[0012] As a further improvement of the technical solution, the specific steps involved in dynamically calculating the feedforward instruction of the coal supply amount and the total air volume through a multivariable prediction control model in the prediction control module are as follows: receiving a load change rate instruction and a real-time heat value estimation value combining a preset unit efficiency, calculating a coal supply amount change rate basic value through a unit energy balance equation; based on the coal supply amount change rate basic value, combining a preset air-coal ratio, calculating a total air volume change rate basic value through an air-coal ratio relationship; based on the coal supply amount change rate basic value and the total air volume change rate basic value, calculating a basic feedforward value of the coal supply amount and the total air volume that meets the load change demand through a first-order difference method; constructing a multivariable prediction control model, which is obtained through multi-objective weighted optimization by a load dynamic model and an oxygen amount dynamic model; wherein the load dynamic model is established with the coal supply amount as the manipulated variable and the unit load as the controlled variable, and is used for predicting the unit load; the oxygen amount dynamic model is established with the total air volume as the manipulated variable and the flue gas oxygen amount as the controlled variable, and is used for predicting the flue gas oxygen amount; multi-objective weighted optimization is performed according to the unit load output by the load dynamic model and the flue gas oxygen amount output by the oxygen amount dynamic model, and a target function is set with the minimum weighted sum of the load error and the oxygen amount deviation in the future prediction time domain as the target; In each control period, the coal supply amount and the total air volume basic feedforward value are used as the initial guess for optimization solution, and a sequential quadratic programming algorithm is used to solve the target function minimization problem with operation constraints; The operation constraints include coal supply amount constraints to ensure that the coal supply amount is within the upper and lower limits of the unit stable combustion range, total air volume constraints to ensure that the total air volume is within the upper and lower limits of the safe air volume, coal supply amount change rate constraints to ensure that the change amount of the coal supply amount in adjacent periods does not exceed the upper limit of the coal supply amount change rate, and total air volume change rate constraints to ensure that the change amount of the total air volume in adjacent periods does not exceed the upper limit of the total air volume change rate; solving the above target function with operation constraints ​Minimize the problem, get the optimal control sequence, take the first element, concatenate by column and transpose to get the current time feedforward instruction .

[0013] As a further improvement of the technical solution, in the control instruction distribution module, the specific steps involved in converting the feedforward instruction into the final action instruction vector that can be directly issued to the bottom layer execution mechanism by using the control distribution algorithm are as follows: Receive total air volume feedforward instruction Based on the preset matching relationship, the total air volume feedforward instruction is distributed to primary air and secondary air by proportional distribution to obtain primary air volume control increment and secondary air volume control increment; Receive coal supply amount feedforward instruction Based on the preset coal amount-speed conversion coefficient of the coal feeder, the coal feeder speed instruction is obtained through linear conversion ; Based on the primary air volume control increment and the preset air volume-opening conversion coefficient of the primary air fan, the primary air fan opening instruction is obtained through linear conversion ; Based on the secondary air volume control increment and the preset air volume-opening conversion coefficient of the secondary air door, the secondary air door opening instruction is obtained through linear conversion ; Based on the coal feeder speed instruction , the primary air fan opening instruction and the secondary air door opening instruction , the final action instruction vector is obtained through column concatenation and transposition .

[0014] On the other hand, the present application provides a flexible peak-oriented coal-fired unit fuel intelligent control method, which is used in the flexible peak-oriented coal-fired unit fuel intelligent control system described in any one of the above, comprising the following steps: The online sensing unit monitors the associated data of the coal quality entering the furnace and the associated data of the unit safety and environmental protection in real time through the unit sensor, and normalizes the data; The normalized associated data of the coal quality entering the furnace are aligned according to the time stamp, and the online identification and estimation of the real-time calorific value of the coal entering the furnace are carried out through the mechanism-guided time sequence collaborative network to obtain the real-time calorific value estimation value; The intelligent decision unit receives the peak regulation instruction issued by the power plant and calculates the initial speed value of the load change based on the monitored unit load data, and generates the load change speed instruction that fits the actual operation capacity of the unit through the safety constraint algorithm based on the normalized associated data of the unit safety and environmental protection; The coordination control unit receives the load change rate instruction and the real-time heat value estimation value, dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model, and outputs the final action instruction vector through a control distribution algorithm.

[0015] Compared with the prior art, the present application has the following beneficial effects: 1、In the present application, the soft measurement unit of the coal quality into the furnace aligns the normalized coal quality correlation data according to the time stamp through the mechanism-guided time sequence cooperative network, realizes online identification of the real-time heat value of the coal into the furnace in combination with the theoretical heat value calculated by the energy balance equation and the full connection layer, can accurately capture the dynamic change of the coal quality, avoids the fluctuation of the combustion efficiency caused by the sensing lag of the traditional control due to the coal quality, and simultaneously generates the load change rate instruction in combination with the heat storage characteristics of the boiler and the safety constraint algorithm of the intelligent decision unit, so that the peak shaving process is adapted to the actual operation capacity of the unit, risks such as over-standard main steam pressure and over-limit NOx emission are effectively avoided, and the safe and environmentally-friendly operation of the unit is ensured.

[0016] 2、In the present application, the coordination control unit constructs a load and oxygen dynamic model by taking the coal supply amount and the total air volume as the manipulated variables through a multivariable predictive control model, solves the multi-objective weighted optimization problem in combination with a sequential quadratic programming algorithm, can dynamically output accurate feedforward instructions of the coal supply amount and the total air volume, and realizes accurate regulation of the wind and coal in combination through a control distribution algorithm to convert the feedforward instructions into the instructions of the coal feeder speed and the fan opening degree, etc., so as to avoid the imbalance of the wind and coal ratio in the traditional control, improve the boiler combustion efficiency and the peak shaving response speed of the unit, and adapt to the flexible peak shaving demand under the new energy grid connection. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The figure is a whole flow chart of the present application.

[0018] The meanings of the various numbers in the figure are as follows: 1, online sensing unit; 2, soft measurement unit of the coal quality into the furnace; 3, intelligent decision unit; 31, load instruction analysis module; 32, operation safety constraint module; 4, coordination control unit; 41, predictive control module; 42, control instruction distribution module. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0020] Embodiment 1: Please refer to Figure 1As shown, a coal-fired unit fuel intelligent control system for flexible peak regulation is provided, comprising An online perception unit 1, which monitors the coal quality correlation data and unit safety and environmental protection correlation data in real time through unit sensors, and normalizes the data; In this embodiment, the coal quality correlation data and unit safety and environmental protection correlation data are read through unit sensors, and the sampling period is set to 1s / time (the sampling period is less than the control period ΔT=1min to avoid data loss); The coal quality correlation data includes coal supply (unit: t / h, engineering operation range is 30%-110% of the rated coal supply), primary air volume (unit: Nm³ / h, engineering operation range is 25%-105% of the rated primary air volume), secondary air volume (unit: Nm³ / h, engineering operation range is 25%-105% of the rated secondary air volume), flue gas oxygen content (unit: %, engineering operation range is 2%-8%), unit load (unit: MW, engineering operation range is 30%-100% of the rated load), and flue gas temperature (unit: ℃, engineering operation range is 800-1200℃), and pressure (unit: kPa, engineering operation range is -5 to 5kPa); The unit safety and environmental protection correlation data includes main steam pressure (unit: MPa, engineering safety range is 95%-105% of the rated pressure of the unit) and NOx concentration (unit: mg / m³, engineering compliance range is 0-50mg / m³, meeting the national ultra-low emission standard); Based on the engineering operation range of each parameter, the upper and lower limits of normalization are set, and the maximum-minimum normalization method is used to normalize the coal quality correlation data and unit safety and environmental protection correlation data, and the data range is unified to [0, 1]; The formula is: Wherein, is the original data, is the engineering operation lower limit of the parameter, is the engineering operation upper limit of the parameter, is the normalized data.

[0021] The coal-fired unit fuel intelligent control system for flexible peak regulation also includes a coal quality soft measurement unit 2, which aligns the normalized coal quality correlation data by timestamp, and performs online identification and estimation of the real-time calorific value of the coal in the furnace through a mechanism-guided time sequence collaborative network, to obtain a real-time calorific value estimate. In this embodiment, the normalized coal quality correlation data is aligned with the timestamp to obtain time series data; Specifically, the standard clock of the unit DCS system (calibrated with NTP, precision millisecond level) is used as the only time source, and a timestamp under this benchmark (format: YYYY-MM-DD HH:MM:SS) is supplemented for each item of normalized coal quality correlation data (coal supply, primary air volume, etc.); Combined with the system control period ΔT=1min, the normalized high-frequency data of 1s / time is aggregated in a "1min time window", that is, the starting time of each minute is taken as the timestamp of this group of data, and the arithmetic mean of the 60 1s-level data in the window is taken as the time series value of the parameter corresponding to this time; The time series values of each parameter after aggregation are spliced in the field order of "timestamp-coal supply-primary air volume-…-flue gas pressure" to form a two-dimensional time series data set; A mechanism-guided time series collaborative network is constructed, which consists of a time series feature extraction layer and a heat value regression layer; the time series feature extraction layer uses one-dimensional convolution (Conv1D) and long short-term memory network (LSTM) to extract features from the input time series data, and the extracted feature vector is denoted as ; One-dimensional convolution is used for local feature extraction: wherein, is the convolution kernel weight, is the input time series data, is the output feature map; is one-dimensional convolution; is the ReLU activation function; Long short-term memory network is used for time series modeling: wherein, is the hidden state of the current time; is the hidden state of the previous time, representing the short-term and dynamic memory of the model for the boiler operation state at the previous sampling time; is the cell state of the previous time, representing the long-term and steady-state memory accumulated from earlier historical times and passed to the previous time; the final extracted feature vector is the output of the last hidden layer of the LSTM, i.e. ; The feature vector obtained after the time series feature extraction layer contains information related to the real-time heat value of the coal entering the furnace, which is contained in the input physical parameters; Based on the actual monitored coal supply and load data, the theoretical heat value is calculated through the energy balance equation; Energy balance equation: wherein, is the boiler efficiency, which is fitted from historical operation data (dimensionless, typically 0.85-0.95); is the coal supply amount, is the unit load; is the theoretical heat value (unit: kJ / kg); Specifically, in the present embodiment, based on the real-time monitored coal supply amount and unit load , and considering the non-combustible components and moisture content in the coal, the theoretical heat value is corrected; first, the original coal supply amount is corrected to the effective combustible coal supply amount: wherein, is the mass moisture content of the coal (dimensionless, typical range 0.05-0.15); represents the mass fraction of ash (dimensionless, typical range 0.10-0.25); represents the effective combustible coal supply amount; then the theoretical heat value : wherein, is the theoretical heat value; represents the unit power generation efficiency (dimensionless coefficient, typical value range 0.40-0.45); by correcting the theoretical heat value, not only the overall efficiency of the unit thermal system is considered, but also the influencing factors such as moisture and ash content of the coal are introduced, so that the calculation of the theoretical heat value is closer to the actual operation condition, and the influence of the coal quality change on the unit combustion process can be more accurately reflected; the theoretical heat value is concatenated with the feature vector according to column, input into the heat value regression layer for prediction, and the real-time heat value estimation value is output, wherein the heat value regression layer comprises a fully connected layer; heat value regression layer: wherein, represents a concatenation operation, represents a linear fully connected layer; is the real-time heat value estimation value (unit: kJ / kg) The mechanism-guided time series collaborative network is trained through historical data, and a hybrid loss function is used in the training process, taking the mean square error (MSE) of the real-time heat value estimation value and the measured heat value as the core fitting term, and combining a mechanism constraint term composed of the deviation between the real-time heat value estimation value and the theoretical heat value; wherein the mechanism constraint term is used to avoid network training from falling into pure data-driven overfitting; wherein, is the hybrid loss function; is the core fitting term, which guarantees the accuracy of the heat value estimation, is the mean square error loss; As a mechanism constraint term, L1 regularization loss is adopted, which has the core function of avoiding network training from falling into pure data-driven overfitting, and ensuring that the real-time calorific value estimate is consistent with the measured data and meets the energy conservation law of boiler combustion; As the weighing coefficient, it is usually 0.1-0.3, which needs to be adjusted according to the fluctuation range of the unit coal quality (the greater the coal quality fluctuation, The value can be appropriately increased to 0.25-0.3 to enhance the mechanism constraint); The measured calorific value (unit: kJ / kg) is determined by the oxygen bomb calorimetry method to measure the bomb calorific value of the coal sample, and then converted into the received base low calorific value according to the total sulfur content, hydrogen content and other data of the coal sample, that is, the measured calorific value (unit: kJ / kg); the measurement error of this method is ≤0.2%, which meets the industrial grade calorific value calibration accuracy requirement; In online operation, real-time data are collected and input into the trained network, and the real-time calorific value estimate of the current coal fed into the furnace is output .

[0022] The coal-fired unit fuel intelligent control system facing flexible peak regulation further comprises an intelligent decision unit 3, which receives the peak regulation instruction issued by the power plant and the unit load data monitored in real time to calculate the initial rate value of load change, and generates a load change rate instruction consistent with the actual operation capacity of the unit based on the safety and environmental protection related data of the unit and through a safety constraint algorithm; In this embodiment, the intelligent decision unit 3 comprises a load instruction analysis module 31 and a running safety constraint module 32; The load instruction analysis module 31 receives the peak regulation instruction issued by the power plant and the unit load data monitored in real time, and calculates the initial rate value of load change according to the heat storage characteristics of the boiler; Specifically, the power plant peak regulation instruction and the current actual load are received, and the load deviation is calculated by difference; Wherein, The load deviation is, The external load instruction is, The current actual load is; According to the heat storage characteristics of the boiler, the safety rate upper limit of load change is determined to obtain the maximum allowable rate of unit load change; Wherein, The maximum allowable rate is, The maximum load change capacity of the unit per unit time (MW / min), The sampling time (min); The heat storage time constant of the boiler (min) is based on the design parameters and historical operation data of the boiler, and is usually 5-15 min; According to the load deviation and the maximum allowable rate, if the absolute value of the load deviation does not exceed the product of the maximum allowable rate and the sampling time, the initial rate value is calculated by dividing the load deviation by the sampling time; if it exceeds, the initial rate value takes the maximum allowable rate with the same positive and negative as the load deviation; If , then ; Otherwise Wherein, is the initial rate value (unit: MW / min); is a sign function; The running safety constraint module 32 receives the above initial rate value, combines the unit safety and environmental protection correlation data, and outputs the load change rate instruction through the safety constraint algorithm; Specifically, taking the unit rated pressure as the benchmark, the upper limit and the lower limit of the safety interval of the main steam pressure are set; The upper limit standard of the NOx concentration is set according to the national environmental protection standard: Wherein, is the main steam pressure, is the NOx concentration; and are the upper limit and the lower limit of the safety interval (unit rated pressure ± 5%); is the upper limit standard of the NOx concentration, which is set according to the national environmental protection standard; The initial rate value is adjusted through the safety constraint algorithm: If the main steam pressure approaches the boundary of the safety interval, the initial rate value is multiplied by the pressure safety coefficient to obtain the intermediate rate value ; Wherein, if approaches or , the rate is reduced: Wherein, is the pressure safety coefficient, ; is the safe load change rate; is the intermediate rate value after the main steam pressure constraint; If the main steam pressure does not approach the boundary of the safety interval, and the NOx concentration exceeds the upper limit standard, the initial rate value is multiplied by the emission penalty coefficient to obtain the load change rate instruction ; Wherein, is the emission penalty coefficient, ; is the load change rate instruction (unit: MW / min); If the main steam pressure is close to the boundary of the safety interval and the NOx concentration exceeds the upper limit standard, the intermediate rate value is multiplied by the emission penalty coefficient to obtain the load change rate instruction ; If the real-time monitored main steam pressure and NOx concentration are within the safety threshold, the initial rate value is directly taken as the final load change rate instruction Output.

[0023] The coal-fired unit fuel intelligent control system for flexible peak regulation further comprises a coordination control unit 4, which receives the load change rate instruction and the real-time calorific value estimate, dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model, and outputs the final action instruction vector through a control distribution algorithm; In the embodiment, the coordination control unit 4 comprises a predictive control module 41 and a control instruction distribution module 42. The predictive control module 41 receives the load change rate instruction and the real-time calorific value estimate, and dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model. Specifically, the load change rate instruction and the real-time calorific value estimate are received, the preset unit efficiency is combined, the coal supply amount change rate basic value is calculated through a unit energy balance equation, and the coal supply amount change rate basic value is calculated through a unit energy balance equation. Unit energy balance equation: Wherein, is the coal supply amount change rate basic value (unit: t / h²); is the unit efficiency (power generation efficiency), which represents the proportion of fuel combustion heat release converted into electric energy; 3600 is a unit conversion constant of 1 hour = 3600 seconds, and 1000 is a unit conversion constant of 1 ton = 1000 kilograms; Based on the coal supply amount change rate basic value, the preset air-fuel ratio is combined, and the total air volume change rate basic value is calculated through an air-fuel ratio relationship. Air-fuel ratio relationship: Wherein, is the total air volume change rate basic value (unit: Nm³ / h²); is the current air-fuel ratio, which is determined by historical optimization data; Based on the coal supply amount change rate basic value and the total air volume change rate basic value, the coal supply amount and the total air volume basic feedforward value meeting the load change demand are calculated through a first-order difference method. Coal amount basic feedforward value: Total air volume basic feedforward value: Wherein, is the current time, is the control period (min), is the coal quantity at the current time (unit: t / h), is the total air quantity at the current time (the sum of the primary air quantity and the secondary air quantity, unit: Nm3 / h); is the coal quantity-based feedforward value, is the total air quantity-based feedforward value; a multivariable predictive control model is constructed, the multivariable predictive control model being obtained through multi-objective weighted optimization of a load dynamic model and an oxygen quantity dynamic model; wherein the load dynamic model is established by taking the coal quantity as a manipulated variable and the unit load as a controlled variable, and is used for predicting the unit load; Load dynamic model: wherein, is a load attenuation coefficient, reflecting the load drop rate of the unit due to heat dissipation, working medium loss and other factors when there is no additional fuel input, and is fitted based on the unit no-load experimental data, and is usually taken as 0.02-0.05 (unit: 1 / min); is the current power grid basic load (MW); is the unit load at the current time (unit: MW), is the unit load at the next time; the oxygen quantity dynamic model is established by taking the total air quantity as a manipulated variable and the flue gas oxygen quantity as a controlled variable, and is used for predicting the flue gas oxygen quantity; Oxygen quantity dynamic model: wherein, is an air quantity influence coefficient on the oxygen quantity, and is linearly fitted based on the boiler combustion experimental data (changing the total air quantity and recording the oxygen quantity change), and is usually taken as 1×10-5-5×10-5 (unit: % / (Nm3)); is the ideal theoretical air quantity under the current combustion condition, and can be calculated according to the ideal air-fuel ratio; is the flue gas oxygen quantity at the current time (unit: %), is the flue gas oxygen quantity at the next time; the unit load output by the load dynamic model and the flue gas oxygen quantity output by the oxygen quantity dynamic model are subjected to multi-objective weighted optimization, so as to minimize the weighted sum of the load error and the oxygen quantity deviation in the future time domain, and a target function is set as the target : wherein, is the load target value at the time, and the calculation formula is: , is the current load; is the best flue gas oxygen quantity set value, and is determined according to the boiler combustion optimization curve; is the weight coefficient of the load, and is taken as 0.6-0.8 (dimensionless), is the weight coefficient of oxygen content, whose value range is 0.2-0.4 (dimensionless) and needs to satisfy ; is the total number of prediction time domain, usually taking a value of 3-8 (i.e. predicting the parameter change in the future 3-8 min); In each control cycle, the coal supply amount and the total air volume basic feedforward value are taken as the initial guess for optimization solution, and the sequential quadratic programming (SQP) algorithm is used to solve the objective function with operation constraints Minimization problem, the operation constraints include the coal supply amount constraint to ensure that the coal supply amount is within the lower and upper limits of the unit stable combustion, the total air volume constraint to ensure that the total air volume is within the lower and upper limits of the safe air volume, the coal supply amount change rate constraint to ensure that the change amount of the coal supply amount in adjacent cycles does not exceed the upper limit of the coal supply amount change rate, and the total air volume change rate constraint to ensure that the change amount of the total air volume in adjacent cycles does not exceed the upper limit of the total air volume change rate; In each control cycle, the current coal supply amount and the total air volume basic feedforward value are taken as the initial state to solve the following optimization problem: , The coal supply amount constraint is satisfied: Wherein, is the lower limit of the unit stable combustion, which is about 30%-40% of the rated coal supply amount; is the upper limit of the unit stable combustion, which is determined by the maximum output of the coal mill and the coal feeder, and is usually 105%-110% of the rated coal supply amount; The total air volume constraint is satisfied: Wherein, is the lower limit of the safe air volume to ensure that the furnace does not burn fuel too much (avoiding deflagration), which is usually 25%-35% of the rated total air volume; is the upper limit of the safe air volume, which is determined by the maximum output of the air feeder, and is usually 105% of the rated air volume; The coal supply amount change rate constraint is satisfied: Wherein, is the change amount of the coal supply amount in adjacent cycles; is the upper limit of the coal supply amount change rate, which is determined by the characteristics of the coal feeder and the coal mill system, and is usually 2%-5% of the rated coal supply amount per minute; The total air volume change rate constraint is satisfied: Wherein, is the change amount of the total air volume in adjacent cycles; is the upper limit of the total air volume change rate, which is determined by the performance of the air feeder actuator, and is usually 3%-6% of the rated total air volume per minute; This system uses the Sequential Quadratic Programming (SQP) algorithm for online rolling optimization. Due to its super-first-order convergence and high efficiency in constraint handling, the SQP method is very suitable for real-time optimization and control problems of medium scale and high nonlinearity, such as those in this system. In each iteration, the SQP method approximates the original nonlinear programming (NLP) problem by constructing and solving a quadratic programming (QP) subproblem. The steps are as follows: Construct the Lagrangian function: in, For decision variables ( , ), As constraints, For Lagrange multipliers; At the current iteration point At this point, a second-order approximation of the Lagrange function is performed (using the BFGS method to update the Hessian matrix). The constraints are linearized to the first order, resulting in a QP problem: satisfy ; The active-set method is used to solve this quadratic programming problem to obtain the search direction. ; Along the search direction Perform a one-dimensional search (such as an Armijo line search) to determine the step size. And update the iteration point: ; Check if the convergence conditions (such as the KKT conditions) are met. If the conditions are met, output the optimal solution; otherwise, continue iterating. Solve the above objective function with operational constraints. The problem is to minimize the optimal control sequence, then concatenate and transpose the first element to obtain the feedforward instruction at the current time step. ; in, This is a feedforward instruction for coal feed rate. This is the total air volume feedforward command. For the feedforward instruction of the output; The control command allocation module 42 uses a control allocation algorithm to convert the feedforward command into a final action command vector that can be directly issued to the underlying actuator. The final action command vector includes the coal feeder speed command, the primary air fan opening command, and the secondary damper opening command. Receive total air volume feedforward command Based on the preset ratio, the total air volume feedforward command is allocated to the primary air and secondary air through proportional allocation, thereby obtaining the primary air volume control increment and the secondary air volume control increment. The proportional allocation formula is: in, To control the increase in primary air volume, This is the increment for secondary air volume control, in Nm³ / h; and The distribution coefficients for primary and secondary winds satisfy... Its value is dynamically adjusted based on the combustion optimization curve; Receive coal feedforward command Based on the preset coal quantity-speed conversion coefficient of the coal feeder, the coal feeder speed command is obtained through linear conversion. ; The linear transformation formula is: in, This is the feeder speed command, in rpm; The coal quantity to speed conversion coefficient of the coal feeder (unit: rpm / (t / h)) is determined by the characteristics of the coal feeder; Based on the primary air volume control increment and the preset primary air volume-opening conversion coefficient, the primary air fan opening command is obtained through linear transformation. ; The linear transformation formula is: in, This is a single fan opening command, in percentages (%). The air volume-opening conversion factor of the primary air fan (unit: % / (Nm³ / h)) is determined by the characteristics of the fan; Based on the secondary air volume control increment and the preset air volume-opening conversion coefficient of the secondary damper, the secondary damper opening command is obtained through linear transformation. ; The linear transformation formula is: in, This is the secondary damper opening command, in percentages (%). The air volume-opening conversion coefficient of the secondary damper (unit: % / (Nm³ / h)) is determined by the damper characteristics; Based on coal feeder speed command Primary air fan opening command and secondary damper opening command The final action command vector is obtained by concatenating and transposing the columns. : in, This is the final action command vector, which will be used subsequently. middle Issued to the feeder actuator to adjust speed, issued to the primary air fan actuator to adjust the opening degree, issued to the secondary air door actuator to adjust the opening degree, to realize the coordinated control of fuel and air volume.

[0024] Embodiment 2: The difference between the embodiment 2 and the embodiment 1 of the present application is that the embodiment 2 is to introduce a flexible peak-shaving-oriented intelligent fuel control method for a flexible peak-shaving-oriented intelligent fuel control system for a coal-fired unit.

[0025] The online sensing unit monitors the coal quality correlation data entering the furnace and the unit safety and environmental protection correlation data in real time through unit sensors, and normalizes the data; The coal quality soft measurement unit aligns the normalized coal quality correlation data entering the furnace according to the time stamp, and performs online identification and estimation of the real-time calorific value of the coal entering the furnace through the mechanism-guided time sequence collaborative network, to obtain a real-time calorific value estimation value; The intelligent decision unit receives the peak-shaving instruction issued by the power plant and the unit load data monitored in real time to calculate the initial rate value of load change, and based on the normalized unit safety and environmental protection correlation data, makes a decision through a safety constraint algorithm to generate a load change rate instruction that fits the actual operation capacity of the unit; The coordinated control unit receives the load change rate instruction and the real-time calorific value estimation value, dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model, and outputs the final action instruction vector through a control distribution algorithm.

[0026] The above shows and describes the basic principles, main features and advantages of the present application. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A coal-fired unit fuel intelligent control system oriented to flexible peak shaving, characterized in that: Comprise: Online perception unit (1), the online perception unit (1) is monitored by unit sensor in real time and is associated with the data of the coal quality of furnace and the safety and environmental protection associated data of unit, and the data is normalized processing; The soft measurement unit (2) of furnace coal quality, the soft measurement unit (2) of furnace coal quality is aligned with the time stamp after the normalized furnace coal quality associated data, and the real-time calorific value of furnace coal is identified and estimated on-line by mechanism guided time sequence collaborative network, and the real-time calorific value estimation value is obtained; Intelligent decision unit (3), the intelligent decision unit (3) receives the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the load change initial rate value calculated by the ​ 2. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 1, characterized in that: ​ ​ ​ 3.The coal-fired unit fuel intelligent control system for flexible peak-shaving in accordance with claim 1, characterized in that: ​ ​ ​ The time sequence feature extraction layer adopts one-dimensional convolution and long short-term memory network to extract features from the input time sequence data, and the extracted feature vector is denoted as ; Based on the actual monitoring of coal supply and load data, the theoretical heat value is calculated by energy balance equation ; The theoretical heat value With eigenvectors Concatenate by column, input heat value regression layer for prediction, output real-time heat value estimate Wherein, the heat value regression layer comprises a fully connected layer; ​ When online running, the data input is collected in real time into the trained network, and the real-time calorific value estimation value of the current coal entering the furnace is output .

4. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 1, characterized in that: ​ ​ ​ 5. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 4, characterized in that: ​ ​ According to the safety rate upper limit of load change of the boiler heat storage characteristics, the maximum allowable rate of the unit load change is obtained; According to the load deviation and the maximum allowable rate, if the absolute value of the load deviation does not exceed the product of the maximum allowable rate and the sampling time, the initial rate value is calculated by dividing the load deviation by the sampling time; if it exceeds, the initial rate value takes the maximum allowable rate with the same positive and negative as the load deviation.

6. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 4, characterized in that: In the operation safety constraint module (32), the specific steps involved in outputting the load change rate instruction by the safety constraint algorithm are as follows: Setting the upper limit of the safety range of the main steam pressure based on the rated pressure of the unit and the lower limit ; setting the upper limit standard for the concentration of NOx according to the national environmental protection standard : If the main steam pressure is close to the boundary of the safety range, the initial rate value is multiplied by a pressure safety factor to obtain an intermediate rate value ; If the main steam pressure is not close to the boundary of the safe interval and the NOx concentration exceeds the upper limit standard, the initial rate value is multiplied by an emission penalty coefficient to obtain a load change rate command ; If the main steam pressure is close to the boundary of the safe interval and the NOx concentration exceeds the upper limit standard, the intermediate rate value is multiplied by an emission penalty coefficient to obtain the load change rate command ; If both the real-time monitored main steam pressure and NOx concentration are within the safety threshold, the initial rate value is directly taken as the final load change rate command Output.

7. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 1, characterized in that: The coordination control unit (4) includes a predictive control module (41) and a control instruction distribution module (42); The predictive control module (41) receives the load change rate instruction and the real-time calorific value estimate, and dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model; The control instruction distribution module (42) converts the feedforward instruction into the final action instruction vector that can be directly issued to the bottom-level executing mechanism by using a control distribution algorithm, and the final action instruction vector includes the coal feeder speed instruction, the primary air fan opening degree instruction, and the secondary air door opening degree instruction.

8. The coal-fired unit fuel intelligent control system for flexible peak-shaving according to claim 7, characterized in that: In the predictive control module (41), the specific steps involved in dynamically calculating the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model are as follows: Receiving a load change rate instruction And real-time heat value estimate Combined with the preset unit efficiency, the coal supply change rate basic value is calculated through the unit energy balance equation. Based on the coal supply amount change rate basic value, the total air volume change rate basic value is calculated through the wind-coal ratio relationship in combination with the preset wind-coal ratio; Based on the coal supply amount change rate basic value and the total air volume change rate basic value, the coal supply amount and the total air volume basic feedforward value that meet the load change demand are calculated through the first-order difference method; A multivariable predictive control model is constructed, which is obtained by a multi-objective weighted optimization of a load dynamic model and an oxygen amount dynamic model; The load dynamic model takes the coal supply amount as the manipulated variable and the unit load as the controlled variable, and is used for predicting the unit load; The oxygen amount dynamic model takes the total air volume as the manipulated variable and the flue gas oxygen amount as the controlled variable, and is used for predicting the flue gas oxygen amount; The unit load output from the load dynamic model and the flue gas oxygen output from the oxygen dynamic model are subjected to multi-objective weighted optimization to minimize the weighted sum of load error and oxygen deviation in the future a prediction time domain to set the objective function : In each control cycle, the coal flow and total air flow base feedforward values are used as initial guesses for the optimization solution, and a sequential quadratic programming algorithm is used to solve the objective function with operational constraints minimization; The operation constraints include coal supply amount constraints to ensure that the coal supply amount is within the upper and lower limits of the unit stable combustion range, total air volume constraints to ensure that the total air volume is within the upper and lower limits of the safe air volume range, coal supply amount change rate constraints to ensure that the change amount of the coal supply amount in adjacent periods does not exceed the upper limit of the coal supply amount change rate, and total air volume change rate constraints to ensure that the change amount of the total air volume in adjacent periods does not exceed the upper limit of the total air volume change rate. Solve the above objective function with operation constraints Minimize, get the optimal control sequence, take the first element, concatenate and transpose to get the current time feedforward instruction . 9.The coal-fired unit fuel intelligent control system for flexible peak-shaving in accordance with claim 7, characterized in that: In the control instruction distribution module (42), the specific steps involved in converting the feedforward instruction into the final action instruction vector that can be directly issued to the bottom-level executing mechanism by using a control distribution algorithm are as follows: Receive total air volume feedforward instruction Based on a preset matching relationship, the total air volume feedforward instruction is distributed to primary air and secondary air through proportional distribution to obtain a primary air volume control increment and a secondary air volume control increment. Receiving coal supply feedforward instruction Obtaining coal supply speed instruction by linear conversion based on preset coal supply-conversion speed conversion coefficient of coal feeder ; Based on the primary air volume control increment and a preset primary air fan air volume-opening degree conversion coefficient, a primary air fan opening degree instruction is obtained through linear conversion ; Based on the secondary air control increment and the preset air volume-opening degree conversion coefficient of the secondary air door, the secondary air door opening degree instruction is obtained through linear conversion ; Based on the coal feeder speed instruction , primary air fan opening degree instruction and secondary air door opening degree instruction , the final action instruction vector is obtained by column splicing and transposition .

10. The intelligent fuel control method for coal-fired units facing flexible peak regulation, applied to the intelligent fuel control system for coal-fired units facing flexible peak regulation as claimed in any one of claims 1-9, characterized in that: The steps include: The online perception unit monitors the associated data of the coal quality entering the furnace and the associated data of the unit safety and environmental protection in real time through the unit sensors, and normalizes the data; The coal quality entering the furnace soft measurement unit aligns the normalized coal quality entering the furnace associated data according to the time stamp, and performs online identification and estimation on the real-time calorific value of the coal entering the furnace through the mechanism-guided time sequence cooperative network, to obtain the real-time calorific value estimate. The intelligent decision unit receives the peak regulation instruction issued by the power plant and the real-time monitored unit load data to calculate an initial load change rate value, and makes a decision based on the normalized unit safety and environmental protection correlation data through a safety constraint algorithm to generate a load change rate instruction that fits the actual operation capacity of the unit. The coordinated control unit receives the load change rate instruction and the real-time heat value estimation value, dynamically calculates the feedforward instruction of the coal supply amount and the total air volume through a multivariable predictive control model, and outputs the final action instruction vector through a control distribution algorithm.

Images