Thermal power generating unit real-time carbon emission calculation method and device and medium

Abstract

The invention discloses a thermal power generating unit real-time carbon emission calculation method and device and a medium under the background of deep peak regulation and flexible operation. The thermal power generating unit real-time carbon emission calculation method comprises the steps that S1, coal quality dynamic back-stepping is carried out, the real-time equivalent low calorific value and the real-time equivalent carbon content are reversely deduced; s2, calculation of the carbon oxidation rate and the combustion efficiency: obtaining a basic carbon oxidation rate according to the real-time equivalent carbon content obtained by the reverse calculation, and calculating a real-time carbon oxidation rate according to the obtained basic carbon oxidation rate; calculating combustion efficiency according to the real-time carbon oxidation rate; and S3, checking and outputting the multi-source data. According to the method, the accounting deviation caused by coal quality fluctuation and load dynamic change under the deep peak regulation working condition can be solved, and the real-time performance and the accuracy are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to a real-time carbon emission calculation method for thermal power generating units under the background of deep peak regulation and flexible operation, and belongs to the technical field of thermal power carbon emission monitoring and regulation. BACKGROUND

[0002] With large-scale grid connection of new energy, thermal power generating units need to frequently participate in deep peak regulation (load fluctuation amplitude exceeding 30%, response time compressed to minutes), and the complexity of working conditions has increased dramatically.

[0003] The traditional carbon emission accounting model is designed based on stable working conditions, and uses the assumptions of "monthly average coal quality" and "fixed combustion efficiency", which cannot adapt to the dynamic characteristics of deep peak regulation:

[0004] 1) Coal quality fluctuation distortion: deep peak regulation is often accompanied by sludge blending and drift of coal composition into the furnace, and the existing model does not establish a real-time coal quality backstepping mechanism, and the energy-carbon balance error is more than ±5%;

[0005] 2) Combustion efficiency lag: the combustion characteristics are significantly different in the peak regulation stage (high load stable combustion, medium load transition, and low load stable combustion), and the fixed efficiency model ignores dynamic disturbances such as furnace temperature, fly ash carbon content, and sludge blending, and the accounting deviation is 5%-8%;

[0006] 3) Distortion of monitoring data: CEMS (Continuous Emission Monitoring System) is affected by the smoke flow field and equipment drift, and the hourly data distortion rate is 3%-5%, and lacks multi-source closed-loop verification logic, and cannot support real-time decision-making.

[0007] In the power system where deep peak regulation is normalized, real-time and accurate carbon emission data is the core basis for unit low-carbon scheduling and carbon trading settlement. Due to the lack of dynamic adaptability and error compensation in the existing technology, it is difficult to meet the accounting needs of "hourly, peak regulation working condition adaptation". Therefore, it is a key problem to be solved to build a carbon emission calculation model that adapts to deep peak regulation, and to break through the dynamic coupling difficulties of coal quality backstepping, efficiency correction, and multi-source verification. SUMMARY

[0008] The technical problem to be solved by the present application is to provide a real-time carbon emission calculation method for thermal power generating units under the background of deep peak regulation and flexible operation, which realizes hourly accurate carbon emission measurement.

[0009] To solve the above technical problems, the technical solution adopted by the present application is:

[0010] The present application first provides a real-time carbon emission calculation method for thermal power generating units under the background of deep peak regulation and flexible operation, comprising the following steps:

[0011] S1. Coal quality dynamic backstepping: Based on the energy balance equation on the boiler side and the power balance equation on the turbine side, combined with the real-time monitoring of power generation, coal quantity, main steam flow, fly ash carbon content and sludge mixing ratio, through the mixed fuel heat value conversion model and the carbon content equivalent calculation model, the real-time equivalent low heat value and the real-time equivalent carbon content are backstepped;

[0012] S2. Carbon oxidation rate and combustion efficiency calculation: The basic carbon oxidation rate is obtained according to the real-time equivalent carbon content obtained by backstepping, the real-time carbon oxidation rate is calculated according to the obtained basic carbon oxidation rate, and the combustion efficiency is calculated according to the real-time carbon oxidation rate;

[0013] S3. Multi-source data verification and output, including:

[0014] According to the carbon oxidation rate calculated in step S2, the carbon emission is calculated;

[0015] The calculated carbon emission is verified by multi-source data, and the multi-source data verification includes primary verification, secondary verification and tertiary verification, respectively:

[0016] Primary verification: Calculate the relative error between the carbon emission calculation value and the CEMS measured value:

[0017]

[0018] Wherein, Error1 is the primary verification relative error, which measures the deviation degree of the model calculation value and the CEMS measured value; The calculated carbon emission is CEMS-CO2; CEMS-CO2 is the actual measured data of the CEMS system; L is the real-time load rate of the unit;

[0019] If the primary verification relative error Error1 exceeds the corresponding load interval threshold, the secondary verification relative error Error1 is fed back to the coal quality dynamic backstepping step for correcting the heat loss coefficient in the thermodynamic balance equation;

[0020] Secondary verification: Calculate the ash carbon balance error Δ2:

[0021]

[0022] Wherein, M slag The carbon content of the ash and slag, C slag The mass fraction of carbon in the ash and slag; The actual emission of CO2 in the flue gas;

[0023] If |Δ2|>5%, the ash carbon balance error Δ2 is fed back to the carbon content equivalent calculation link to adjust the coefficients k and c of the unburned carbon compensation term ΔC unburn ;

[0024] ΔCunburn = kgΔ2 + c

[0025] wherein k is a weight, and c is an intercept;

[0026] Third-level verification: calculate the monitoring load change rate dL / dt and the load change amount |ΔL|:

[0027]

[0028] When dL / dt < -0.10 min -1 and |ΔL| > 0.08, the load change rate dL / dt and the load change amount |ΔL| are fed back to the combustion efficiency correction link to obtain the furnace temperature correction factor α under load fluctuation T :

[0029] α T = α T,static · [1 + 0.5·|dL / dt| + 2.0·|ΔL|]

[0030] wherein α T,static is the conventional furnace temperature correction factor when the load is stable, and the experience value is 1.0; after the load is stable, the furnace temperature correction factor α T returns to α T ,static

[0031] The real-time carbon emission amount verified by multi-source data is output.

[0032] In step S3, the calculated carbon emission amount is:

[0033]

[0034] wherein B is a fuel amount tensor, B = [B coal ,B sludge ] T , wherein B coal is the amount of coal, and B sludge is the amount of sludge; K conv is a conversion coefficient tensor, is a Kronecker product symbol, used for multiplication operation between tensors; C mix is the real-time equivalent carbon content; η c,actual is the real-time carbon oxidation rate.

[0035] In step S2, the real-time carbon oxidation rate is:

[0036] η c,actual = η c,base · α T · β Ash · γ s

[0037] wherein η c,actual is the real-time carbon oxidation rate; η c,base is the base carbon oxidation rate; α T is the furnace temperature correction factor; β Ash is the fly ash unburned carbon loss factor; γ s is the sludge co-combustion kinetics loss factor;

[0038] The combustion efficiency is:

[0039] ζ actual = η c,actual · η other

[0040] wherein ζ actual is the combustion efficiency; η other is the non-carbon element combustion efficiency correction term, which is an empirical value of 0.99-1.0. The base carbon oxidation rate η c,base is:

[0041] η c,base = a0+a1·L+a2·C mix

[0042] wherein a0 is the design base carbon oxidation rate, a1 is the load rate influence coefficient, a2 is the real-time equivalent carbon content influence coefficient, L is the load rate, C mix is the real-time equivalent carbon content;

[0043] By covering the historical operating condition data in different load and carbon content intervals, the uniquely determined values of a0, a1 and a2 are fitted.

[0044] In step S3, the method for correcting the heat loss coefficient in the thermodynamic equilibrium equation is:

[0045] Define the deviation function:

[0046]

[0047] wherein E is the deviation of the calculated value from the measured value CEMS-CO2 of the CEMS; n is the sample number, is the calculated value of the i-th sample, is the measured value; by taking the partial derivative of E with respect to k1, k2 and k3 by the least square method and setting the partial derivative to 0, the iterative updating equation group is obtained:

[0048]

[0049] Solve the equation group to iteratively update the values of k1, k2 and k3 until the deviation E converges.

[0050] In step S3, the relationship between the primary check relative error and the corresponding load interval threshold value is as follows:

[0051]

[0052] The mixed fuel heat value conversion model is as follows:

[0053]

[0054] Wherein, Q net,mix is the real-time equivalent low heat value of the mixed fuel; Q coal is the received base low heat value of the coal; Q sludge is the received base low heat value of the sludge; R s is the sludge mixing ratio; ΔQ kin is the kinetic loss term.

[0055] The kinetic loss term is as follows:

[0056]

[0057] Wherein, L is the real-time load rate.

[0058] In step S1, the carbon content equivalent calculation model is as follows:

[0059] C mix =C ar ·(1-R s )+C sludge ·R s -ΔC unburn

[0060] Wherein, C mix is the real-time equivalent carbon content of the mixed fuel; C ar is the received base carbon content of the coal; R s is the sludge mixing ratio; C sludge is the received base carbon content of the sludge; ΔC unburn is the unburned carbon compensation term.

[0061] In step S1, the piecewise Newton method is used for optimization and solution, and the real-time equivalent low heat value and the real-time equivalent carbon content are obtained by back calculation; wherein the BFGS algorithm is used for the high load section with a load rate of more than 70%, and the Trust-Region algorithm is used for the deep peak regulation section with a load rate of less than 50%.

[0062] The present application also provides a real-time carbon emission calculation device for a thermal power unit, comprising:

[0063] One or more processors;

[0064] A memory for storing one or more programs;

[0065] When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the real-time carbon emission calculation method of the thermal power generating unit as provided above.

[0066] The application also provides a storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the steps of the real-time carbon emission calculation method of the thermal power generating unit as provided above.

[0067] Compared with the prior art, the application has the following beneficial effects:

[0068] (1) Dynamic adaptation of deep peak regulation: through stage division + multi-factor correction, the calculation deviation of combustion efficiency is compressed from ±8% to ±

[0069] 2.3%, covering a wide range of fluctuations of 10% to 100% of the rated load;

[0070] (2) Breakthrough in coal quality backstepping accuracy: thermodynamic double balance + segmented optimization algorithm, coal quality parameter backstepping error is reduced from ±5% to ±

[0071] 1.8%, solving the distortion of coal quality under blending combustion conditions;

[0072] (3) Multi-source closed loop self-correction: three-level verification to build a "collection -> calculation -> verification -> feedback" closed loop, total carbon emission error ≤

[0073] ±1.2%, supporting real-time scheduling at the hour level;

[0074] (4) Significant engineering value: providing accurate data for carbon trading and low-carbon scheduling of thermal power generating units under deep peak regulation, and helping the "double carbon" goal of the power system to be achieved. BRIEF DESCRIPTION OF DRAWINGS

[0075] Figure 1 The flowchart of the calculation method of the application;

[0076] Figure 2 The dynamic backstepping flowchart of coal quality;

[0077] Figure 3 The peak regulation stage correction logic diagram;

[0078] Figure 4 The multi-source data verification closed loop flowchart;

[0079] Figure 5 The carbon emission tensor operation simplification diagram. DETAILED DESCRIPTION

[0080] The application is further described below in combination with the actual operation data of the 14th unit of Xiangqian Power Plant in May 2025 and the accompanying drawings. It should be understood that the embodiments are only used to illustrate the application and do not limit the scope, and modifications fall within the scope defined by the claims.

[0081] A real-time carbon emission calculation method for thermal power generating units in a deep peak-shaving flexible operation background, as shown in Figure 1 , includes the following steps:

[0082] Step 1, data acquisition:

[0083] Collect real-time data of the unit DCS system (power generation, main steam flow, furnace temperature, flue gas temperature, ambient temperature, sampling frequency 1 minute / time), data of the coal metering system (coal quantity, sludge mixing ratio, sampling frequency 5 minutes / time), data of the fly ash detection device (fly ash carbon content, sampling frequency 30 minutes / time), as well as historical data of coal quality test reports in the past three months, unit design parameters, and hourly average data of flue gas CO2 concentration and flue gas flow of the CEMS system.

[0084] Step 2, dynamic backstepping of coal quality:

[0085] The dynamic backstepping of coal quality is as shown in Figure 2

[0086] 1) Input layer: receive real-time monitoring data, including power generation, coal quantity, main steam flow, fly ash carbon content, sludge mixing ratio, flue gas temperature, and ambient temperature;

[0087] 2) Calculation layer:

[0088] Based on the thermodynamic equilibrium equation and the power balance of the steam turbine side, the energy balance of the boiler side and the power balance of the steam turbine side are calculated:

[0089] The thermodynamic equilibrium equation is:

[0090] B·Q net,ar ·η b =F steam ·(h vap -h sw )+k1(T exh -T amb )+k2Ash fly +k3(1-η b )

[0091] Wherein, B is the fuel quantity tensor (unit: kg / h), if it is a mixed fuel of coal and sludge, the form is

[0092] B coal is the coal quantity, B sludge is the sludge quantity; Q net,ar ​GCV: gross calorific value of fuel (unit: kJ / kg), real-time equivalent value obtained by backstepping through mixed fuel calorific value conversion model; η b η: boiler efficiency (dimensionless), reflecting the effective degree of energy conversion of the boiler, which needs to be corrected according to real-time working conditions; F steam F: main steam flow (unit: kg / h), obtained by real-time monitoring; h vap h: steam enthalpy (unit: kJ / kg), corresponding to the enthalpy of main steam parameters; h sw h: feedwater enthalpy (unit: kJ / kg), determined according to the temperature and pressure of feedwater; k1: correction coefficient of flue gas heat loss (dimensionless), related to the structure of the boiler and the characteristics of flue gas, fitted by historical data; T exh T: flue gas temperature (unit: ℃), monitored in real time; T amb T: ambient temperature (unit: ℃), monitored in real time; k2: correction coefficient of fly ash heat loss (dimensionless), related to the carbon content of fly ash and heat loss, fitted by experience; Ash fly Ash: carbon content of fly ash (unit: %), detected in real time; k3: correction coefficient of unburned carbon heat loss (dimensionless), adapted to the change of unburned carbon under deep peak regulation conditions.

[0093] The power balance equation on the steam turbine side is:

[0094] P e = F steam · (h sup -h cond ) / (3600·HR ref )·η m ·η g ·ξ(L)

[0095] Wherein, P e : generator output power (unit: MW), measured in real time; F steam : main steam flow (unit: kg / h), defined in the same way as in the thermodynamic balance equation; h sup : main steam enthalpy (unit: kJ / kg), corresponding to the main steam parameters;

[0096] h cond : condensate enthalpy (unit: kJ / kg), determined according to the parameters of the condenser; HR ref : unit rated heat rate (unit: kJ / kWh), design reference parameter; η m : mechanical efficiency of steam turbine generator unit (dimensionless), reflecting mechanical transmission loss; η g : generator efficiency (dimensionless), reflecting energy conversion loss in the power generation link;

[0097] ξ(L): load correction factor (dimensionless), related to the load rate L of the unit, fitted for efficiency deviation under deep peak regulation conditions.

[0098] Real-time equivalent low heat value and real-time equivalent carbon content are calculated based on a mixed fuel heat value conversion model and a carbon content equivalent calculation model;

[0099] The mixed fuel heat value conversion model is:

[0100]

[0101] wherein, Q net,mix is the real-time equivalent low heat value of the mixed fuel (unit: kJ / kg); Q coal is the received base low heat value of the coal (unit: kJ / kg); Q sludge is the received base low heat value of the sludge (unit: kJ / kg); R s is the sludge mixing ratio (unitless); ΔQ kin is a kinetic loss term (unitless), which is enabled when R s > 5%;

[0102]

[0103] wherein, L is the real-time load rate.

[0104] The carbon content equivalent calculation model is:

[0105] C mix = C ar ·(1-R s )+C sludge ·R s -ΔC unburn

[0106] wherein, C mix is the real-time equivalent carbon content of the mixed fuel (unit: %); C ar is the received base carbon content of the coal (unit: %); R s is the sludge mixing ratio (unitless); C sludge is the received base carbon content of the sludge (unit: %); ΔC unburn is an unburned carbon compensation term (unit: %).

[0107] The unburned carbon compensation term

[0108] wherein, Ash fly is the carbon content of fly ash.

[0109] 3) Optimization layer:

[0110] The segmented quasi-Newton method is adopted, the BFGS algorithm is used in the high load section (the initial value of iteration is taken from the coal quality backstepping result of the previous 1 hour, the iteration precision is e-5, and the maximum iteration number is 20), and the Trust-Region algorithm is used in the deep peak regulation section (the initial value of the trust region radius is 0.1, which is dynamically adjusted according to the iteration effect, if the residual error is reduced, the radius is expanded by 1.2 times, otherwise it is reduced by 0.5 times), to solve the objective function (minimize the energy deviation of the boiler and the steam turbine side); the real-time equivalent low calorific value Q of the mixed fuel is obtained by backstepping net,mix The real-time equivalent carbon content C mix .

[0111] The energy deviation minimization objective function is constructed:

[0112] minf = |Q boiler -Q turbine| |

[0113] Wherein, Q boiler is the boiler side energy; Q turbine is the steam turbine side energy;

[0114] Q boiler = B·Q net,ar ·η b

[0115]

[0116] Wherein, B is the fuel quantity tensor; Q net,ar is the received low calorific value; η b is the boiler efficiency; F steam is the main steam flow; h sup -h cond is the enthalpy difference; HR ref is the rated heat rate, efficiency correction term; the real-time equivalent low calorific value Q of the mixed fuel is obtained by backstepping net,mix , the real-time equivalent carbon content C mix .

[0117] 4) Output layer: output the real-time equivalent low calorific value Q net,mix and the real-time equivalent carbon content C mix .

[0118] Step 3, combustion efficiency and carbon oxidation rate calculation:

[0119] The real-time equivalent carbon content C mix obtained by backstepping in step 2 is input into the fitting model to obtain the basic carbon oxidation rate η c,base ; the basic carbon oxidation rate is obtained by a multi-dimensional correlation model of “design reference + load rate + real-time equivalent carbon content”;

[0120] η c,base=a0+a1·L+a2·C mix ;

[0121] Where a0 is the design baseline carbon oxidation rate (dimensionless, calibrated by the unit design operating conditions, e.g., a0 = 0.92), a1 is the load rate influence coefficient (dimensionless, fitted by historical operating conditions, e.g., a1 = -0.05, characterizing the degree of correction of the load rate L to the baseline carbon oxidation rate), and a2 is the real-time equivalent carbon content influence coefficient (unit: %). -1 Based on historical operating conditions, if a2 = 0.03 is taken, it represents C. mix The degree of correction to the baseline carbon oxidation rate), where L is the loading rate (dimensionless). ), C mix The real-time equivalent carbon content (unit: %); the model is trained by historical operating data covering different loads and carbon content ranges, and the unique values ​​of a0, a1, and a2 are fitted to ensure the uniqueness of the basic carbon oxidation rate calculation.

[0122] Then, based on the baseline carbon oxidation rate, calculate the real-time carbon oxidation rate:

[0123] η c,actual =η c,base ·α T ·β Ash ·γ s

[0124] Where, η c,actual For real-time carbon oxidation rate; η c,base Based on carbon oxidation rate, α T β is the furnace temperature correction factor. Ash γ is the unburned carbon loss factor in fly ash; s The kinetic loss factor for sludge co-firing;

[0125] Calculate the combustion efficiency ζ based on the real-time carbon oxidation rate. actual :

[0126] ζ actual =η c,actual ·η other

[0127] Where, η other This is a correction term for the combustion efficiency of non-carbon elements, with an empirical value of 0.99-1.0.

[0128] Step 4: Multi-source data verification and output:

[0129] Step 41: Calculate carbon emissions:

[0130]

[0131] in, The hourly average carbon emissions (unit: kg / h) are derived through tensor calculations based on the fuel quantity tensor, mixed carbon content, carbon oxidation rate, and conversion coefficient, reflecting the carbon emissions calculated by the model; B is the fuel quantity tensor, B = [B coal B sludge ] T B coal For coal consumption, B sludge For sludge quantity; K conv For the transformation coefficient tensor, The Kronecker product symbol is used for multiplication operations between tensors; C mix η represents the real-time equivalent carbon content. c,actual For real-time carbon-oxygen ratio;

[0132] Step 42, Multi-source data verification:

[0133] Multi-source data verification process, such as Figure 4 As shown, the initial verification (CEMS real-time comparison) is performed first:

[0134] Calculate the relative error between the calculated carbon emission values ​​and the measured values ​​by CEMS:

[0135]

[0136] CEMS-CO2 is the hourly average CO2 emission value in flue gas (unit: kg / h) monitored in real time by the CEMS system and converted from standard flow and concentration, representing the actual measured data of emissions; L is the unit's real-time load rate (dimensionless), used to distinguish deep peak shaving conditions; Error1 is the primary verification relative error (dimensionless), which measures the degree of deviation between the model's calculated value and the CEMS measured value.

[0137] If the error exceeds the threshold of the corresponding load range, the error Error1 will be fed back to the coal quality dynamic back-calculation link to correct the heat loss coefficients k1, k2, and k3 in the thermodynamic balance equation;

[0138] The method for correcting the heat loss coefficient in the thermodynamic equilibrium equation is as follows:

[0139] Define the deviation function:

[0140]

[0141] Where E is the calculated value. Deviation from the measured CEMS-CO2 value; n is the sample size. It is the calculated value of the i-th sample. These are measured values; by taking the partial derivatives of E with respect to l1, l2, and l3 using the least squares method and setting the partial derivatives to 0, we obtain the iteratively updated system of equations:

[0142]

[0143] Solve the equation set, iteratively update the value of l1, l2, l3, until the deviation E converges (for example, the change of E of adjacent two iterations is less than 10 -6 ).

[0144] The relationship between the error and the corresponding load interval threshold value is:

[0145]

[0146] Then, secondary verification (ash carbon closed loop balance) is performed:

[0147] Calculate the ash carbon balance error Δ2:

[0148]

[0149] If |Δ2|>5%, the ash carbon balance error Δ2 is fed back to the carbon content equivalent calculation link to adjust the coefficient of the unburned carbon compensation term ΔC unburn ;

[0150] The unburned carbon compensation term ΔC unburn is:

[0151] ΔC unburn = kgΔ2+ c

[0152] Where the weight k and the intercept c are the coefficients;

[0153] Adjustment method: when |Δ2|>5%, collect the current ash carbon content M slag (kg / h), the ash carbon content C slag (%) and the ash carbon balance error Δ2(%), take the historical "ash carbon balance error Δ2-actual unburned carbon compensation amount" as a sample, and fit the objective function by the least square method to solve iteratively to make E converge (change less than 10 -6 ), and update k, c to match the current working condition (initial k is 0.6, and c is 0.05).

[0154] Then, tertiary verification (load jump stability) is performed:

[0155] Monitor the load change rate dL / dt(unit: min -1 , the calculation method is "the difference between the current load and the load at the previous 1 minute, divided by the time interval (1 minute)") and the load change |ΔL|(dimensionless, the calculation method is When dL / dt<-0.10min -1and | AL | > 0.08, the two types of data (load change rate dL / dt and load change | AL |) are fed back to the combustion efficiency correction link to temporarily improve the carbon oxidation rate correction factor sensitivity (i.e. in the deep peak shaving condition of rapid load fluctuation, the furnace temperature correction factor a is increased for a short time T The response coefficient of the furnace temperature fluctuation makes the carbon oxidation rate correction more matched with the dynamic combustion characteristics; and when the load tends to be stable, the conventional response coefficient is restored.

[0156] Figure 3 The modification logic diagram for the peak shaving stage is shown in the figure, which intuitively embodies two layers of logic:

[0157] First, the peak shaving stage is divided and the dominant correction factor is determined: the RPS section (high load stable combustion, L > 50% L max ) is dominated by furnace temperature fluctuation disturbance, and the furnace temperature correction factor a T ; the DPS section (medium load transition, 40%≤L≤50%L max ) is dominated by fly ash unburned carbon loss, and the fly ash unburned carbon loss factor b Ash ; the DPSC section (low load stable combustion, L < 40% L max ) is dominated by sludge blending dynamics loss, and the sludge blending dynamics loss factor g s .

[0158] Second, the "verification-correction-calculation" closed loop: in step 4 "multi-source data verification", if the three-level verification (load jump stability) triggers the correction requirement (such as dL / dt < -0.10min - 1 and | AL | > 0.08), the peak shaving branch of step 3 "combustion efficiency and carbon oxidation rate calculation" is driven, and the multi-factor coupling correction model is updated c,actual = η c,base · a T · b Ash · g s , and finally fed back to the carbon emission calculation to ensure the calculation accuracy in the deep peak shaving condition.

[0159] Peak shaving stage division:

[0160] 1) RPS section (high load stable combustion, L > 50% L max ): dominated by furnace temperature fluctuation disturbance;

[0161] 2) DPS section (medium load transition, 40%≤L≤50%L max ): dominated by fly ash unburned carbon loss;

[0162] 3) DPSC section (low load stable combustion, L < 40% L max ): dominated by sludge blending dynamics loss.

[0163] Multi-factor coupling correction model: η c,actual =η c,base gα T gβ Ash gγ s

[0164] 1) Furnace temperature correction (RPS section): α T =1-0.005||T furnace -1500||

[0165] 2) Fly Ash Correction (DPS Segment): β Ash =1-0.15log(Ash) fly ×100)

[0166] 3) Firing correction (DPSC section):

[0167] Where, η c,actual For real-time carbon oxidation rate; η c,base The baseline carbon oxidation rate is fitted by a design baseline and load correlation model; α T β is the furnace temperature correction factor. Ash γ is the unburned carbon loss factor in fly ash; s Ash is the kinetic loss factor for sludge co-firing. fly R represents the carbon content of fly ash. s T is the sludge blending ratio; furnace This refers to the furnace temperature.

[0168] Step 43: Data Output

[0169] After all verifications are passed, the carbon emission figures will be output. Together with the primary verification error Error1, they form a complete closed loop of "calculation-verification-correction".

[0170] Figure 5 A simplified diagram illustrating carbon emission tensor computation shows the dimension matching and computation process for four tensors: fuel quantity tensor B (2×1 dimension), real-time equivalent carbon content tensor C. mix (Dimension 1×1), Carbon Oxidation Rate Tensor η c (1×1 dimension), transformation coefficient tensor K cov (Dimension 2×1), the CO2 emission tensor is obtained through Hadamard product operation (Dimension 2×1), and the synchronization mechanism of each tensor with 1 hour as the timestamp is marked.

[0171] Implementation Cases

[0172] Step 1, Data Source:

[0173] 1) Historical data: In May, the coal consumption of Unit 14 in Jianbi Power Plant was 203102 tons, the power generation was 41569 million kWh (the average load was 68.25%), the fly ash carbon content was 0.53%, the slag carbon content was 1.59%, and the hourly CEMS-CO2 monitoring data (e.g. 5 / 10:00 emission 21472.02 kg / h);

[0174] 2) Assumed parameters: Unit rated load L max = 600 MW, sludge mixing ratio R s = 8% (typical deep peak shaving blending ratio), furnace design temperature T design = 1500℃.

[0175] Model architecture association:

[0176] As shown in Figure 1 , the revised text is: "Input the historical operation data of the last month, use the least squares method to fit the heat loss coefficients k1, k2, k3 to make the deviation between the calculated and measured values of the boiler side energy balance minimum" (e.g. heat loss coefficients k1 = 0.02, k2 = 0.015, k3 = 0.01).

[0177] Step 2, dynamic backstepping of coal quality (taking 5 / 1 0:00 operating conditions as an example)

[0178] 1. Substitute the thermodynamic equilibrium equation:

[0179] Boiler side energy balance (take the main steam flow F steam = 380 t / h, flue gas temperature ambient temperature T amb = 25℃):

[0180] 380·Q net,mix ·0.92 = 380·(3300-420) + 0.02(135-25) + 0.015·0.53 + 0.01·(1-0.92)

[0181] (Note: h sup = 3300 kJ / kg, h fw = 420 kJ / kg, boiler efficiency η b = 0.92 is the design value).

[0182] 2. Mixed fuel property conversion:

[0183] Turbine side power balance (power generation P e= 11.55 MW, corresponding load rate L = 11.55 / 600 ≈ 19.25%. Here, a contradiction is found: the original data "monthly average load 68.25%", which needs to be corrected to the actual load of 410 MW (68.25% Lmax) at 0:00 on May 1st, the main steam flow is corrected to F steam = 620 t / h, recalculate:

[0184]

[0185] (Note: heat rate HR ref = 7000 kJ / kWh, mechanical / generating efficiency η m = 0.98, η g = 0.99.)

[0186] 3. Optimization solution verification:

[0187] Using BFGS algorithm (since L = 68.25% > 70% is not true, the actual 68.25% belongs to RPS section, still using BFGS), the calculation gets Q net,mix Deviation from thermodynamic equilibrium solution <1.5%, verifying the effectiveness of coal quality backstepping (correlation Figure 2 : Coal quality backstepping flowchart).

[0188] Step 3, adaptive correction of combustion efficiency (RPS stage, L = 68.25% L max ):

[0189] 1. Peak shaving stage determination:

[0190] 68.25% > 50%, belongs to RPS section (high load stable combustion), the dominant correction factor is the furnace temperature stability coefficient α T .

[0191] 2. Furnace temperature correction factor calculation:

[0192] Assume the furnace temperature at 0:00 on May 1st is Then: α T = 1-0.005·|1480-1500| = 0.99

[0193] 3. Dynamic carbon oxidation rate solution:

[0194] Basic carbon oxidation rate η c,base = 97.78% (design value, correlation coal quality report), then:

[0195] η c,actual = 97.78% × 0.99 × 1 × 1 ≈ 96.80%

[0196] (Note: the fly ash correction of DPS section and the blending correction of DPSC section are not triggered, so βAsh =γ s =1; association Figure 3 (Diagram illustrating the correction logic during peak shaving.)

[0197] Step 4: Multi-source data closed-loop verification (taking May 1st, 0:00 as an example)

[0198] 1. Zhang's quantitative carbon emission calculation:

[0199] Fuel quantity tensor B = [203102 / 31 / 24, 0] T ≈[275.1t / h,0] T (Sludge is not considered at this time, B) sludge =0, related Figure 5 :

[0200] (Simplified diagram of carbon emission tensor computation);

[0201] 2. Carbon emission calculation:

[0202]

[0203] The calculation yields:

[0204] 3. Three-level verification:

[0205] 1) Initial verification (CEMS comparison): CEMS monitoring value 21472.02 kg / h, error

[0206] |(21350-21472) / 21472|≈0.57%<3% (meets the RPS threshold);

[0207] 2) Secondary verification (ash and slag carbon balance): Calculated value of carbon compensation item ≈ 122 kg / h, total carbon balance error

[0208] |(21350+122)-(21472+0)|≈0<5%;

[0209] 3) Level 3 verification (load transition): The load stabilized at 0:00 on May 1st (dL / dt≈0), and no emergency correction was required.

[0210] (related) Figure 4 (Diagram of multi-source verification closed loop).

[0211] Implementation effect verification

[0212] Verification was made using data from the Jianbi Power Plant in May:

[0213] (1) Coal quality back estimation error:

[0214] Monthly average Q net,mixThe deviation of backstepping value and test value is 1.8% (better than 5% error of traditional model);

[0215] (2) Combustion efficiency deviation:

[0216] The deviation of RPS segment real-time efficiency correction is 2.3% (8% of traditional model);

[0217] (3) Total carbon emission error:

[0218] The deviation of hourly accounting value and CEMS+ash compensation value is 1.2% (about 5% of traditional model), verifying the dynamic adaptability of the application.

Claims

1. A method for calculating real-time carbon emissions of thermal power units under the background of deep peak shaving and flexible operation, characterized in that, Includes the following steps: S1. Dynamic Coal Quality Back-Calculation: Based on the boiler-side energy balance and turbine-side power balance equations, combined with real-time monitoring of power generation, coal quantity, main steam flow, fly ash carbon content and sludge blending ratio, the real-time equivalent lower heating value and real-time equivalent carbon content are back-calculated through the mixed fuel calorific value conversion model and carbon content equivalent calculation model. S2. Calculation of carbon oxidation rate and combustion efficiency: The basic carbon oxidation rate is obtained based on the real-time equivalent carbon content obtained by reverse calculation, and the real-time carbon oxidation rate is calculated based on the obtained basic carbon oxidation rate. Calculate combustion efficiency based on real-time carbon oxidation rate; S3. Multi-source data verification and output, including: Calculate the carbon emissions based on the carbon oxidation rate obtained in step S2. The calculated carbon emissions are verified using multi-source data, which includes primary verification, secondary verification, and tertiary verification, as follows: Preliminary verification: Calculate the relative error between the calculated carbon emission values ​​and the CEMS measured values. Error1 is the primary validation relative error, which measures the degree of deviation between the model's calculated value and the CEMS measured value. is the calculated carbon emissions; CEMS-CO2 is the actual measured data of emissions from the CEMS system; L is the real-time load rate of the unit. If the primary verification relative error Error1 exceeds the threshold of the corresponding load range, the primary verification relative error Error1 will be fed back to the coal quality dynamic back-calculation step to correct the heat loss coefficient in the thermodynamic balance equation. Secondary verification: Calculate the carbon balance error Δ2 in ash and slag: Among them, M slag Carbon content and C content of ash slag The mass fraction of carbon in the ash residue; This represents the actual CO2 emissions in the flue gas. If |Δ2|>5%, the carbon balance error Δ2 of the ash residue will be fed back to the carbon content equivalent calculation stage to adjust the unburned carbon compensation term ΔC. unburn The coefficients k and c; ΔC unburn =kgΔ2+c Where k is the weight and c is the intercept; Level 3 verification: Calculate the monitored load change rate dL / dt and the load change amount |ΔL|: When dL / dt < -0.10min -1 When |ΔL|>0.08, the load change rate dL / dt and the load change amount |ΔL| are fed back to the combustion efficiency correction stage to obtain the furnace temperature correction factor α under load fluctuation. T : a T =a T,static ·[1+0.5·|dL / dt|+2.0·|ΔL|] Where, α T,static The standard furnace temperature correction factor under stable load is taken as an empirical value of 1.0; after the load stabilizes, the furnace temperature correction factor α... T Restored to α T static; Real-time carbon emissions verified by multi-source data Output.

2. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 1, characterized in that, In step S3, the carbon emissions are calculated. for: Where B is the fuel quantity tensor, B = [B coal B sludge ] T B coal For coal consumption, B sludge For sludge quantity; K conv For the transformation coefficient tensor, This is the Kronecker product symbol, used for multiplication operations between tensors; C mix η represents the real-time equivalent carbon content. c,actual This represents the real-time carbon oxidation rate.

3. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 1, characterized in that, In step S2, the real-time carbon oxidation rate is: or c,actual =the c,base ·a T ·b Ash ·c s Where, η c,actual For real-time carbon oxidation rate; η c,base Basic carbon oxidation rate; α T β is the furnace temperature correction factor. Ash γ is the unburned carbon loss factor in fly ash; s The kinetic loss factor for sludge co-firing; Combustion efficiency is: g actual =the c,actual ·or other Where, ζ actual For combustion efficiency; η other This is a correction term for the combustion efficiency of non-carbon elements, with an empirical value of 0.99-1.

0.

4. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 3, characterized in that, Basic carbon oxidation rate η c,base for: η c,base =a0+a1·L+a2·C mix Where a0 is the design baseline carbon oxidation rate, a1 is the load rate influence coefficient, a2 is the real-time equivalent carbon content influence coefficient, L is the load rate, and C is the load factor. mix Real-time equivalent carbon content; By fitting historical operating data covering different load and carbon content ranges, unique and definite values ​​for a0, a1, and a2 are obtained.

5. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 3, characterized in that, In step S3, the method for correcting the heat loss coefficient in the thermodynamic equilibrium equation is as follows: Define the deviation function: Where E is the calculated value. Deviation from the measured CEMS-CO2 value; n is the sample size. It is the calculated value of the i-th sample. These are measured values; by taking the partial derivatives of E with respect to k1, k2, and k3 using the least squares method and setting the partial derivatives to 0, we obtain the iteratively updated system of equations: Solve the system of equations and iteratively update the values ​​of k1, k2, and k3 until the deviation E converges.

6. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 1, characterized in that, In step S3, the relationship between the primary verification relative error and the corresponding load interval threshold is as follows:

7. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 1, characterized in that, In step S1, the calorific value conversion model for the mixed fuel is as follows: Among them, Q net,mix Q represents the real-time equivalent lower heating value of the mixed fuel; coal Q is the net calorific value of coal received from the source; sludge The lower heating value of the sludge is obtained from the base; R s ΔQ is the sludge blending ratio. kin This is the dynamic loss term; The kinetic loss term is: Where L is the real-time load factor; In step S1, the equivalent calculation model for carbon content is as follows: C mix =C ar ·(1-R s )+C sludge ·R s -ΔC unburn Among them, C mix Real-time equivalent carbon content of blended fuels; C ar The carbon content of the coal received; R s C is the sludge blending ratio. sludge The carbon content of the sludge received; ΔC unburn This is an unburned carbon compensation item.

8. The method for calculating real-time carbon emissions of thermal power units under the flexible operation background of peak shaving as described in claim 1, characterized in that, In step S1, the piecewise quasi-Newton method is used to optimize the solution and back-calculate the real-time equivalent lower heating value and the real-time equivalent carbon content; the BFGS algorithm is used for the high load segment with a load rate >70%, and the Trust-Region algorithm is used for the deep peak shaving segment with a load rate <50%.

9. A real-time carbon emission calculation device for thermal power units, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the real-time carbon emission calculation method for thermal power units as described in any one of claims 1-8.

10. A storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the real-time carbon emission calculation method for thermal power units as described in any one of claims 1 to 8.

Images