A method, system and device for predicting thermal loss of coal-fired power generation

Abstract

The present application provides a kind of thermal power coal-fired heat loss prediction method, system and device, the present application is established by obtaining the heat distribution model of coal-fired boiler heat distribution data type to be predicted, by obtaining the same time interval of boiler parameter and ambient temperature, respectively calculate the heat that is not effectively utilized and the heat that is effectively utilized, form preliminary loss heat data set and preliminary utilization heat data set, and according to the total heat input in each time period, calculate measurement error value, according to the loss heat and the heat utilization calculation with loss ratio, preliminary loss heat is corrected to obtain the prediction data set of coal in each time interval, the neural network is established by prediction model of coal-fired heat loss, the optimization model of coal-fired heat loss prediction model is established, the parameter of prediction model is optimized, the future time coal-fired heat loss is predicted by optimized prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of coal-fired heat loss, and in particular to a method, system and device for predicting coal-fired heat loss in thermal power generation. Background Art

[0002] With the global energy transition and increasing environmental protection requirements, the thermal power generation industry faces increasingly severe challenges in reducing coal heat loss and improving combustion efficiency. Coal heat loss refers to the portion of coal that is not effectively converted into heat or electricity during combustion during thermal power generation. This heat loss not only wastes energy but also seriously impacts the economic profitability of power plants. Therefore, accurately predicting and optimizing coal heat loss has become a key issue in improving coal utilization efficiency and promoting sustainable development in the power industry.

[0003] In the prior art, publication number CN112465241 B discloses a method, system, device, and storage medium for predicting heat loss from coal-fired power generation. The method includes collecting burnout influencing factors of coal combustion, including coal characteristic parameters, combustion mode parameters, furnace structure parameters, and furnace operation parameters; determining characteristic quantities of coal combustion products based on a pre-created neural network model and burnout influencing factors; wherein the characteristic quantities include slag parameters, soot parameters, coal leakage parameters, and fly ash parameters; and determining the heat loss of coal combustion based on a heat loss relationship and the characteristic quantities, wherein the heat loss relationship is the relationship between the heat loss of coal and the characteristic quantities. In this application, by collecting burnout influencing factors and a neural network model, the content of slag, soot, and unburned combustible materials in the products after coal combustion is determined, thereby accurately predicting heat loss. However, this invention primarily focuses on the characteristic quantities of combustion products, with little consideration of dynamic operating conditions such as boiler parameters, ambient temperature, and combustion mode. This results in an incomplete model input variable, limiting its prediction accuracy. However, it fails to model and analyze the dynamic characteristics of heat distribution over time. This approach, which ignores time series characteristics, may fail to capture the regular changes in heat loss during actual operation, thereby reducing the accuracy and practicality of the prediction and making it difficult to adapt to the dynamic changes and complex operating conditions of thermal power plants.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to provide a method, system and device for predicting heat loss of coal-fired power generation to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for predicting heat loss from coal-fired power generation, comprising the following steps:

[0008] Step 1: Obtain the data type of the thermal power boiler heat distribution to be predicted, including boiler input heat, boiler effective utilization heat, exhaust heat loss, gas incomplete combustion heat loss, solid incomplete combustion heat loss, heat loss due to heat dissipation, and ash physical heat loss, and establish a coal heat distribution model;

[0009] Step 2: Based on the coal heat distribution model and heat distribution data type, by obtaining boiler parameters and ambient temperature at the same time interval, calculate the heat that is not effectively utilized and the heat that is effectively utilized, respectively, to form a preliminary heat loss dataset and a preliminary heat utilization dataset;

[0010] Step 3: Calculate the measurement error value based on the preliminary heat loss data set, the preliminary heat utilization data set, and the boiler input heat in each time interval. Calculate the loss-use ratio based on the heat loss and heat utilization, correct the preliminary heat loss, and obtain the corrected heat loss.

[0011] Step 4: Obtain the coal particle size within each time interval to calculate the average coal particle size value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters, and flue gas parameters. Through normalization processing, a prediction data set is formed. A coal heat loss prediction model is established through a neural network and the model is trained.

[0012] Step 5: Establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, speed up the model training, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss in the future.

[0013] Furthermore, the calculation formula of the coal heat distribution model is:

[0014] Q z =Q1+Q2+Q3+Q4+Q5+Q6+Q w

[0015] Among them, Q z is the heat input to the boiler, Q1 is the effective heat utilization of the boiler, Q2 is the heat loss of exhaust gas, Q3 is the heat loss of incomplete combustion of gas, Q4 is the heat loss of incomplete combustion of solid, Q5 is the heat loss of heat dissipation, Q6 is the physical heat loss of ash, Q w is the measurement error value.

[0016] Furthermore, the boiler parameters include exhaust volume, specific heat capacity of flue gas, flue gas temperature, concentration of combustible gas in flue gas, slag volume, coal content in slag, boiler insulation coefficient, boiler area, boiler surface temperature, specific heat capacity of slag, temperature at slag discharge, steam production, steam specific enthalpy, feed water specific enthalpy, and coal heat;

[0017] The calculation formula for the heat that is not effectively utilized is:

[0018] Q1=Q2+Q3+Q4+Q5+Q6

[0019] Among them, Q1 is the heat that is not effectively utilized, Q2 is the heat loss of flue gas, Q3 is the heat loss of incomplete combustion of gas, Q4 is the heat loss of incomplete combustion of solid, Q5 is the heat loss of heat dissipation, and Q6 is the physical heat loss of ash;

[0020] The calculation formula for exhaust heat loss Q2 is:

[0021] Q2=Gyq*Cyq*(Tyq-Thj)

[0022] Among them, Gyq is the exhaust volume, Cyq is the specific heat capacity of flue gas, Tyq is the flue gas temperature, and Thj is the ambient temperature;

[0023] The calculation formula for the heat loss Q3 due to incomplete combustion of gas is:

[0024] Q3=Gyq*Nrq*Hrq

[0025] Among them, Gyq is the exhaust volume, Nrq is the concentration of combustible gas in the flue gas, and Hrq is the heat of the combustible gas;

[0026] The calculation formula for the heat loss Q4 due to incomplete combustion of solids is:

[0027] Q4=Glz*Nc*Hc

[0028] Where Glz is the amount of slag, Nc is the coal content in the slag, and Hc is the calorific value of the coal;

[0029] The calculation formula for heat loss Q5 is:

[0030] Q5=Kbw*Sbm*(Tbm-Thj)

[0031] Where Kbw is the boiler insulation coefficient, Sbm is the boiler area, Tbm is the boiler surface temperature, and Thj is the ambient temperature;

[0032] The calculation formula for the physical heat loss of ash Q6 is:

[0033] Q6=Glz*Czz*(Tlz-Thj)

[0034] Where Glz is the amount of slag, Czz is the specific heat capacity of slag, Tlz is the temperature when slag is discharged, and Thj is the ambient temperature;

[0035] The calculation formula of the effectively utilized heat Q1 is:

[0036] Q1=Mzq*(Pzq-Pjs)

[0037] Among them, Mzq is the steam production, Pzq is the steam specific enthalpy, and Pjs is the feed water specific enthalpy.

[0038] Furthermore, the calculation formula of the boiler input heat is:

[0039] Q z =Mtc*Hc

[0040] Among them, Mtc is the coal consumption, Hc is the calorific value of coal;

[0041] The calculation formula of the measurement error value is:

[0042] Q w =Q z -Q1-Q l

[0043] Among them, Q w is the measurement error value.

[0044] Furthermore, the calculation formula of the use-loss ratio is:

[0045]

[0046] Among them, Pys is the usage-loss ratio, na is the number of data collections, in is the current in-th data collection, Q 1,in is the heat that is effectively utilized for the first time, Q l，in is the heat that is not effectively utilized for the current inth time;

[0047] The calculation formula for correcting the initial heat loss is:

[0048] Qx l =Pys*Q w +Q2+Q3+Q4+Q5+Q6

[0049] Among them, Qx l is the corrected heat loss.

[0050] Furthermore, the specific steps for calculating the coal average particle value and particle deviation are as follows:

[0051] Take multiple samples of the pulverized coal entering the furnace within each time interval, count the size of the pulverized coal particles in each sample, and calculate the average particle size and particle deviation of the coal within the time interval:

[0052] The calculation formula for the average particle size of coal is:

[0053]

[0054] Among them, Pmp is the average particle size of coal, nq is the number of sampling times, iq is the iqth sampling, nb is the number of particles in each sampling, MD ib is the size of the ibth coal particle;

[0055] The calculation formula for particle deviation is:

[0056]

[0057] Among them, Qmp is the particle deviation, nq is the number of sampling times, iq is the iqth sampling, nb is the number of coal particles in each sampling, MD ib is the size of the ibth coal particle.

[0058] Furthermore, the furnace parameters include the amount of coal fed into the furnace and the furnace temperature;

[0059] The steam parameters include feed water flow, main steam flow, main steam pressure, main steam temperature, and reheat steam temperature;

[0060] The air volume parameters include primary air volume, primary air temperature, secondary air volume, secondary air temperature, and burnout air volume;

[0061] The flue gas parameters include flue gas temperature and flue gas flow rate;

[0062] The calculation formula for the normalization process is:

[0063]

[0064] Among them, Xg is the normalized data of each parameter type, Xmax is the largest number in each parameter type, Xmin is the smallest number in each parameter type, and xi id The idth data in each parameter type, nx is the number of data in each parameter type.

[0065] Furthermore, the specific steps of establishing a coal-fired heat loss prediction model through a neural network and training the model are as follows:

[0066] The neural network is an LSTM neural network prediction model, which consists of three gates: forget gate, input gate, and output gate:

[0067] The equation of the forget gate is:

[0068] f t =δ(W f ·[h t―1 ,x t ]+b f )

[0069] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state of the previous moment, x t Input switch for current data;

[0070] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0071] The equation for the input gate is:

[0072] i t =δ(W i ·[h t-1 ,x t ]+b i )

[0073] Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t-1 is the output hidden state of the previous moment, x t is the prediction data set of the previous moment;

[0074]

[0075] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t-1 is the hidden state of the previous moment, x t Input switch for current data;

[0076] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0077] The update equation is:

[0078]

[0079] Among them, C t Current cell state, Ct-1 is the cell state at the previous moment;

[0080] The output gate equation is:

[0081] o t =(W o ·[h t-1 ,x t ]+b o )

[0082] h t =o t *tanh(C t )

[0083] Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t-1 is the hidden state of the previous moment, x t For the current data input switch, h t is the corrected heat loss at the current moment:

[0084] Controls how much of the current unit state needs to output the corrected loss heat at the current moment;

[0085] The training steps are:

[0086] The prediction data set at the previous moment is used as the input of the prediction model, and the corrected heat loss at the current moment is used as the output of the model to train the model.

[0087] The parameters optimized for the prediction model include learning rate, number of hidden units, time step, and regularization parameter;

[0088] The optimization model is based on the sparrow search optimization algorithm. The sparrow population consists of discoverers, joiners, and a certain number of scouts. The specific optimization logic is:

[0089] The discoverer's position update formula is as follows:

[0090]

[0091] Where t is the current iteration number, j is the dimension of the parameter to be optimized, j = 1, 2, ..., d, d is the total number of parameters to be optimized, X i，j is the position information of the i-th sparrow in the j-th dimension; iter max is the maximum number of iterations; α is a random number, α∈(0,1), R2 and ST are the alarm value and safety threshold; L is a 1×d-dimensional matrix; Q is a random number that conforms to the normal distribution;

[0092] The position updating formula of the joiner is as follows:

[0093]

[0094] In the formula, X P is the optimal position of the current discoverer; X worst is the current global worst position; A is a 1xd matrix, the elements of which are randomly assigned 1 or -1, and ni is the population size of the sparrow;

[0095] The position updating formula of the scout is as follows:

[0096]

[0097] In the formula, X best is the global optimal position, beta is the iteration step length, f i is the fitness value of the current sparrow individual, f g and f w are the current global optimal and worst fitness values, K is a random number, K belongs to [-1, 1], and epsilon is a constant, which prevents the denominator from being 0:

[0098] The root mean square error is used to calculate the fitness, and the calculation formula is as follows:

[0099]

[0100] In the formula, n is the sample number, x i is the prediction data set at the last moment, y i is the corrected loss heat.

[0101] The present application also provides a thermal power coal-fired heat loss prediction system, which is used for executing the thermal power coal-fired heat loss prediction method, and comprises:

[0102] A heat distribution module is configured to acquire the heat distribution data types of the thermal power boiler to be predicted, including the boiler input heat, the boiler effective utilization heat, the exhaust smoke heat loss heat, the gas incomplete combustion heat loss heat, the solid incomplete combustion heat loss heat, the heat loss heat, and the ash slag physical heat loss heat, and establish a coal-fired heat distribution model.

[0103] A heat calculation module is configured to calculate the heat not effectively utilized and the heat effectively utilized respectively by acquiring the boiler parameters and the environmental temperature at the same time interval according to the coal-fired heat distribution model and the heat distribution data types, and form a preliminary loss heat data set and a preliminary utilization heat data set.

[0104] a loss correction module, which is used to calculate a measurement error value based on a preliminary heat loss data set, a preliminary heat utilization data set, and the boiler input heat in each time interval, calculate a loss-use ratio based on the heat loss and the heat utilization, correct the preliminary heat loss, and obtain a corrected heat loss;

[0105] A model building module is used to obtain the coal particle size in each time interval to calculate the coal average particle value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters and flue gas parameters through normalization processing to form a prediction data set, establish a coal heat loss prediction model through a neural network, and train the model;

[0106] The model optimization module is used to establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, accelerate the training of the model, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss at future times.

[0107] The present invention also provides a device for predicting heat loss of coal-fired power generation, which includes a processor and a storage medium. The storage medium stores a computer program. When the computer program is executed by the processor, it can implement the above-mentioned method for predicting heat loss of coal-fired power generation.

[0108] Compared with the prior art, the present invention has the following beneficial effects: the present invention establishes a coal heat distribution model by acquiring the data type of the heat distribution of a thermal power generation boiler to be predicted, calculates the heat that is not effectively utilized and the heat that is effectively utilized by acquiring boiler parameters and ambient temperature at the same time interval, respectively, forms a preliminary heat loss data set and a preliminary heat utilization data set, calculates a measurement error value based on the total heat input in each time period, calculates the loss-use ratio based on the heat loss and the utilized heat, corrects the preliminary heat loss data set, obtains a prediction data set of coal in each time interval, establishes a coal heat loss prediction model through a neural network, establishes an optimization model of the coal heat loss prediction model, optimizes the parameters of the prediction model, and predicts coal heat loss at future moments through the optimized prediction model;

[0109] The present application calculates the unused heat and the utilized heat separately, calculates the measurement error value for the total heat of each time period, and dynamically corrects the ratio of the loss heat to the utilized heat, so as to more comprehensively reflect the source of heat loss and improve the reliability of the predicted data. The data set of the time interval is used for modeling, and the predicted data set of the current time is used as input to predict the future heat loss. This way fully embodies the dynamic characteristics of time series data, can adapt to real-time changes in power plant operation, and the prediction result is closer to the actual situation. The parameters of the neural network model are optimized by the optimization algorithm, which not only improves the prediction accuracy, but also is more efficient and stable when dealing with complex nonlinear relationships. BRIEF DESCRIPTION OF DRAWINGS

[0110] Figure 1 The present application provides a whole method flowchart;

[0111] Figure 2 The present application provides a whole system structure diagram. DETAILED DESCRIPTION

[0112] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in combination with specific embodiments.

[0113] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present application should be understood as the usual meaning understood by those skilled in the art to which the present application belongs. The "first", "second" and similar words used in the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Include" or "contain" and similar words mean that the elements or objects before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connected" or "connected" and similar words are not limited to physical or mechanical connection, but can include electrical connection, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to represent relative positional relationship, when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0114] EMBODIMENT:

[0115] Please refer to Figure 1 The present application provides a technical scheme:

[0116] A thermal loss prediction method for coal-fired power generation, the specific steps comprising:

[0117] Step 1: Obtain the data type of the thermal power boiler heat distribution to be predicted, including boiler input heat, boiler effective utilization heat, flue gas heat loss, gas incomplete combustion heat loss, solid incomplete combustion heat loss, heat dissipation loss, and ash physical heat loss, and establish a coal heat distribution model.

[0118] The thermal power generation process involves a large amount of heat transfer and conversion. The heat distribution of the boiler directly determines the effective utilization efficiency of thermal energy and the overall performance of the system. The heat distribution model is used to predict boiler heat loss. Establishing the model can provide a basis for subsequent prediction and control strategy formulation. The heat distribution model integrates the input heat and various heat loss data, and can intuitively show the flow direction and loss ratio of heat during boiler operation.

[0119] The input heat of the boiler is the calorific value of the pulverized coal, and also includes the physical sensible heat of the fuel, the heat carried by the heat entering the furnace and the heat brought in by the atomized fuel vapor. The effective utilization heat of the boiler is the heat absorbed by the water and steam in the steam drum and the heat absorbed by the boiler steam when it flows through various heating surfaces. The exhaust heat loss of the boiler is the heat carried by the hot flue gas discharged by the boiler when it passes through the economizer, air preheater, etc. and is finally discharged into the atmosphere. It is the largest item of boiler heat loss. The heat loss of incomplete combustion of gas is the heat loss caused by the incomplete combustion of combustible gases in the furnace during the combustion of pulverized coal. The heat loss of incomplete combustion of solid refers to the loss caused by the incomplete combustion of carbon in the slag, fly ash and pulverized coal. The heat dissipation loss is the heat loss dissipated by the boiler and its connected pipes because the temperature is higher than the external air temperature. The physical sensible heat loss of ash is the heat carried by the high-temperature slag when it is discharged from the furnace.

[0120] In this embodiment, the calculation formula of the coal heat distribution model is:

[0121] Q z =Q1+Q2+Q3+Q4+Q5+Q6+Q w

[0122] Among them, Q z is the heat input to the boiler, Q1 is the effective heat utilization of the boiler, Q2 is the heat loss of exhaust gas, Q3 is the heat loss of incomplete combustion of gas, Q4 is the heat loss of incomplete combustion of solid, Q5 is the heat loss of heat dissipation, Q6 is the physical heat loss of ash, Q w is the measurement error value.

[0123] Step 2: Based on the coal heat distribution model and heat distribution data type, by obtaining boiler parameters and ambient temperature at the same time interval, calculate the heat that is not effectively utilized and the heat that is effectively utilized, respectively, to form a preliminary heat loss dataset and a preliminary heat utilization dataset.

[0124] By acquiring time interval data, we can capture dynamic changes during boiler operation and provide a more accurate description of the operating conditions. At the same time, we can adapt to the dynamic load requirements of boiler operation, such as changes in heat distribution when the power generation load fluctuates, reduce the impact of random measurement noise on heat distribution analysis, and provide more stable input data.

[0125] In this embodiment, the boiler parameters include flue gas volume, flue gas specific heat capacity, flue gas temperature, concentration of combustible gas in flue gas, slag volume, coal content in slag, boiler insulation coefficient, boiler area, boiler surface temperature, slag specific heat capacity, temperature at slag discharge, steam production, steam specific enthalpy, feed water specific enthalpy, and coal heat capacity;

[0126] The calculation formula for the heat that is not effectively utilized is:

[0127] Q l =Q2+Q3+Q4+Q5+Q6

[0128] Among them, Q l is the heat that is not effectively utilized, Q2 is the heat loss due to exhaust gas, Q3 is the heat loss due to incomplete combustion of gas, Q4 is the heat loss due to incomplete combustion of solid, Q5 is the heat loss due to heat dissipation, and Q6 is the physical heat loss due to ash;

[0129] The calculation formula for exhaust heat loss Q2 is:

[0130] Q2=Gyq*Cyq*(Tyq-Thj)

[0131] Among them, Gyq is the exhaust volume, Cyq is the specific heat capacity of flue gas, Tyq is the flue gas temperature, and Thj is the ambient temperature;

[0132] The calculation formula for the heat loss Q3 due to incomplete combustion of gas is:

[0133] Q3=Gyq*Nrq*Hrq

[0134] Among them, Gyq is the exhaust volume, Nrq is the concentration of combustible gas in the flue gas, and Hrq is the heat of the combustible gas;

[0135] The calculation formula for the heat loss q4 due to incomplete combustion of solids is:

[0136] Q4=Glz*Nc*Hc

[0137] Where Glz is the amount of slag, Nc is the coal content in the slag, and Hc is the calorific value of the coal;

[0138] The calculation formula for heat loss Q5 is:

[0139] Q5=Kbw*Sbm*(Tbm-Thj)

[0140] Where Kbw is the boiler insulation coefficient, Sbm is the boiler area, Tbm is the boiler surface temperature, and Thj is the ambient temperature;

[0141] The calculation formula for the physical heat loss of ash Q6 is:

[0142] Q6=Glz*Czz*(Tlz-Thj)

[0143] Where Glz is the amount of slag, Czz is the specific heat capacity of slag, Tlz is the temperature when slag is discharged, and Thj is the ambient temperature;

[0144] The calculation formula of the effectively utilized heat Q1 is:

[0145] Q1=Mzq*(Pzq-Pjs)

[0146] Among them, Mzq is the steam production, Pzq is the steam specific enthalpy, and Pjs is the feed water specific enthalpy.

[0147] Step 3: Calculate the measurement error value based on the preliminary heat loss data set, the preliminary heat utilization data set, and the boiler input heat in each time interval. Calculate the loss-use ratio based on the heat loss and heat utilization, correct the preliminary heat loss, and obtain the corrected heat loss.

[0148] During boiler operation and calculation, when calculating the heat that is not effectively utilized and the heat that is effectively utilized separately, errors caused by measurement and calculation may occur, such as insufficient sensor accuracy or data acquisition delays, certain small influencing factors ignored by the coal heat distribution model, and fluctuations in boiler operating load that may cause uneven heat distribution in a short period of time.

[0149] The total boiler heat input is used as a reference value to ensure that the sum of heat loss and heat utilization is equal to the heat input. If errors are not corrected, the calculated results may deviate from the actual operating conditions. The corrected data set better reflects the actual operating conditions of the boiler and provides high-quality input for the prediction model.

[0150] In this embodiment, the calculation formula for the boiler input heat is:

[0151] Q z =Mtc*Hc

[0152] Among them, Mtc is the coal consumption, Hc is the calorific value of coal;

[0153] The calculation formula of the measurement error value is:

[0154] Q w =Q z -Q1-Q l

[0155] Among them, Q w is the measurement error value.

[0156] During boiler operation, the ratio of heat loss to heat utilization varies with operating conditions, such as load fluctuations and ambient temperature changes. The loss-to-energy ratio correction method dynamically reflects these changes. It also reduces cumulative deviations caused by initial calculation errors, ensuring the stability and reliability of the final data.

[0157] In this embodiment, the calculation formula of the use-loss ratio is:

[0158]

[0159] Among them, Pys is the usage-loss ratio, na is the number of data collections, in is the current in-th data collection, Q1, in is the heat that is effectively utilized for the first time, Q l，in The amount of heat that is not effectively utilized for the current time

[0160] The calculation formula for correcting the initial heat loss is:

[0161] Qx l =Pys*Q w +Q2+Q3+Q4+Q5+Q6

[0162] Among them, Qx l is the corrected heat loss.

[0163] Step 4: Obtain the coal particle size in each time interval to calculate the coal average particle value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters and flue gas parameters. Through normalization processing, form a prediction data set, establish a coal heat loss prediction model through a neural network, and train the model.

[0164] The size of coal particles directly affects the completeness of combustion. The smaller the particles, the more complete the combustion and the smaller the heat loss. Particles that are too large may lead to increased incomplete combustion losses. Large deviations in particle size indicate uneven particle distribution, which may lead to local incomplete combustion and increase heat loss. By obtaining the mean and deviation of particle size, the impact of coal powder particles on combustion effects can be quantified, improving the ability to describe combustion losses.

[0165] In this embodiment, the specific steps for calculating the coal average particle value and particle deviation are as follows:

[0166] Take multiple samples of the pulverized coal entering the furnace within each time interval, count the size of the pulverized coal particles in each sample, and calculate the average particle size and particle deviation of the coal within the time interval:

[0167] The calculation formula for the average particle size of coal is:

[0168]

[0169] Among them, Pmp is the average particle size of coal, nq is the number of sampling times, iq is the iqth sampling, nb is the number of particles in each sampling, MD ib is the size of the ibth coal particle;

[0170] The calculation formula for particle deviation is:

[0171]

[0172] Among them, Qmp is the particle deviation, nq is the number of sampling times, iq is the iqth sampling, nb is the number of coal particles in each sampling, MD ib is the size of the ibth coal particle.

[0173] In this embodiment, the furnace parameters include the amount of coal fed into the furnace and the furnace temperature;

[0174] The steam parameters include feed water flow, main steam flow, main steam pressure, main steam temperature, and reheat steam temperature;

[0175] The air volume parameters include primary air volume, primary air temperature, secondary air volume, secondary air temperature, and burnout air volume;

[0176] The flue gas parameters include flue gas temperature and flue gas flow rate.

[0177] Unit load reflects the output level of the boiler and directly affects heat distribution and losses. Furnace parameters reflect the combustion conditions within the combustion chamber and significantly influence heat losses. Steam parameters are directly related to the boiler's thermal efficiency and influence the calculation of heat utilization and heat losses. Air volume parameters determine the oxygen supply during combustion, affecting combustion completeness and exhaust heat losses. Flue gas parameters reflect incomplete combustion losses and exhaust heat losses. By comprehensively collecting these parameters, a multi-dimensional coal-fired heat loss prediction dataset can be formed, covering all key factors influencing the combustion process.

[0178] In this embodiment, the calculation formula for the normalization process is:

[0179]

[0180] Among them, Xg is the normalized data of each parameter type, Xmax is the largest number in each parameter type, Xmin is the smallest number in each parameter type, and xi id The idth data in each parameter type, nx is the number of data in each parameter type.

[0181] The numerical ranges of different parameters (such as particle size, temperature, and pressure) can vary significantly. Directly inputting these values ​​into a neural network can magnify or minimize the impact of certain parameters, affecting model training effectiveness. Normalization standardizes data to the same range [-1, 1], ensuring that the model can fairly learn the contribution of each parameter. Normalization also reduces the impact of numerical differences on gradient descent, preventing gradient explosion or vanishing, thereby improving the convergence speed of neural network training.

[0182] The core advantages of using LSTM to build a coal-fired heat loss prediction model lie in its powerful modeling capabilities for time series data, its precise capture of multivariate nonlinear relationships, and its adaptability to complex operating conditions. Compared to traditional methods, LSTM can significantly improve prediction accuracy, reduce reliance on feature engineering, and meet the real-time and robustness requirements of real-world operating conditions.

[0183] LSTMs avoid information loss over long time spans by recording important information. Using input, forget, and output gate mechanisms, they dynamically control the flow of information, extracting and retaining features important to the prediction target while ignoring irrelevant information. This allows the long-term dependencies of historical information to be captured in coal-fired heat loss prediction. For example, short-term changes in furnace parameters and load can affect current heat loss, while steam parameters and coal particle properties may have longer-term effects. Preserving multi-period features helps extract dynamic changes within past time intervals, thereby improving prediction accuracy.

[0184] In this embodiment, the specific steps of establishing a coal-fired heat loss prediction model through a neural network and training the model are as follows:

[0185] The neural network is an LSTM neural network prediction model, which consists of three gates: forget gate, input gate, and output gate:

[0186] The equation of the forget gate is:

[0187] f t =δ(W f ·[h t-1 , x t ]+b f )

[0188] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t-1 is the hidden state of the previous moment, x t Input switch for current data;

[0189] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0190] The equation for the input gate is:

[0191] i t =δ(W i ·[h t-1 ,x t ]+b i )

[0192] Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t-1 is the output hidden state of the previous moment, x t is the prediction data set of the previous moment;

[0193]

[0194] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t-1 is the hidden state of the previous moment, x t Input switch for current data;

[0195] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0196] The update equation is:

[0197]

[0198] Among them, C t Current cell state, C t-1 is the cell state at the previous moment;

[0199] The output gate equation is:

[0200] o t =(W o ·[h t-1 ,x t ]+b o )

[0201] h t =o t *tanh(C t )

[0202] Among them, tis the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t-1 is the hidden state of the previous moment, x t For the current data input switch, h t is the corrected heat loss at the current moment;

[0203] Controls how much of the current unit state needs to output the corrected loss heat at the current moment;

[0204] The training steps are:

[0205] The prediction data set at the previous moment is used as the input of the prediction model, and the corrected heat loss at the current moment is used as the output of the model to train the model.

[0206] Step 5: Establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, speed up the model training, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss in the future.

[0207] In the prediction of coal-fired heat loss in thermal power generation, building an optimization model based on the sparrow search optimization algorithm can significantly improve the performance and practical value of the prediction model. Coal-fired heat loss prediction involves complex nonlinear and multivariate relationships. The core challenge lies in how to efficiently and accurately optimize neural network hyperparameters such as the learning rate, number of hidden layers, and regularization coefficient to improve the model's predictive accuracy and robustness.

[0208] During the optimization process, the sparrow search optimization algorithm initializes the population with a random distribution, sets the neural network parameters to individual positions, and iteratively optimizes the parameters using discoverers, joiners, and a certain number of scouts. By updating the positions and velocities of individuals, it enables rapid global exploration and effectively escapes local optimality, ensuring global optimality of the parameter optimization results.

[0209] In this embodiment, the parameters optimized by the prediction model include learning rate, number of hidden units, time step, and regularization parameter;

[0210] The optimization model is based on the sparrow search optimization algorithm. The sparrow population consists of discoverers, joiners, and a certain number of scouts. The specific optimization logic is:

[0211] The discoverer's position update formula is as follows:

[0212]

[0213] Where t is the current iteration number, j is the dimension of the parameter to be optimized, j = 1, 2, ..., d, d is the total number of parameters to be optimized, X i，j is the position information of the i-th sparrow in the j-th dimension; iter max is the maximum number of iterations; α is a random number, α∈(0,1), R2 and ST are the alarm value and safety threshold; L is a 1×d-dimensional matrix; Q is a random number that conforms to the normal distribution;

[0214] The position update formula of the joiner is as follows:

[0215]

[0216] Where: X P is the optimal position of the current discoverer; X worst is the current global worst position; A is a 1×d matrix whose elements are randomly assigned 1 or -1, and ni is the size of the sparrow population;

[0217] The scout position update formula is as follows:

[0218]

[0219] Among them, X best is the global optimal position, β is the iteration step size, f i is the fitness value of the current sparrow individual, f g and f w is the current global optimal and worst fitness value, K is a random number, K∈[-1,1], ε is a constant to prevent the denominator from being 0;

[0220] The root mean square error is used to calculate the fitness, and the calculation formula is as follows:

[0221]

[0222] Among them, n is the number of samples, x i is the prediction data set of the previous moment, y i is the corrected heat loss.

[0223] See also Figure 2 The present invention further provides a thermal power generation coal-fired heat loss prediction system, which is used to execute the thermal power generation coal-fired heat loss prediction method, including:

[0224] A heat distribution module is used to obtain the type of heat distribution data of the thermal power generation boiler to be predicted, including boiler input heat, boiler effective utilization heat, exhaust heat loss, gas incomplete combustion heat loss, solid incomplete combustion heat loss, heat loss due to heat dissipation, and ash physical heat loss, and to establish a coal heat distribution model;

[0225] a heat calculation module, which is used to calculate the heat that is not effectively utilized and the heat that is effectively utilized, respectively, based on the coal heat distribution model and the heat distribution data type, by obtaining boiler parameters and ambient temperature at the same time interval, to form a preliminary heat loss data set and a preliminary heat utilization data set;

[0226] a loss correction module, which is used to calculate a measurement error value based on a preliminary heat loss data set, a preliminary heat utilization data set, and the boiler input heat in each time interval, calculate a loss-use ratio based on the heat loss and the heat utilization, correct the preliminary heat loss, and obtain a corrected heat loss;

[0227] A model building module is used to obtain the coal particle size in each time interval to calculate the coal average particle value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters and flue gas parameters through normalization processing to form a prediction data set, establish a coal heat loss prediction model through a neural network, and train the model;

[0228] The model optimization module is used to establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, accelerate the training of the model, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss at future times.

[0229] The present invention also provides a device for predicting heat loss of coal-fired power generation, which includes a processor and a storage medium. The storage medium stores a computer program. When the computer program is executed by the processor, it can implement the above-mentioned method for predicting heat loss of coal-fired power generation.

[0230] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0231] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0232] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0233] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for predicting heat loss of coal-fired power generation, characterized in that: The specific steps include: Step 1: Obtain the data type of the thermal power boiler heat distribution to be predicted, including boiler input heat, boiler effective utilization heat, exhaust heat loss, gas incomplete combustion heat loss, solid incomplete combustion heat loss, heat loss due to heat dissipation, and ash physical heat loss, and establish a coal heat distribution model; Step 2: Based on the coal heat distribution model and heat distribution data type, by obtaining boiler parameters and ambient temperature at the same time interval, calculate the heat that is not effectively utilized and the heat that is effectively utilized, respectively, to form a preliminary heat loss dataset and a preliminary heat utilization dataset; Step 3: Calculate the measurement error value based on the preliminary heat loss data set, the preliminary heat utilization data set, and the boiler input heat in each time interval. Calculate the loss-use ratio based on the heat loss and heat utilization, correct the preliminary heat loss, and obtain the corrected heat loss. Step 4: Obtain the coal particle size within each time interval to calculate the average coal particle size value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters, and flue gas parameters. Through normalization processing, a prediction data set is formed. A coal heat loss prediction model is established through a neural network and the model is trained. Step 5: Establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, speed up the model training, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss in the future; The calculation formula of the use-loss ratio is: ； in, is the loss ratio, is the number of data collection times, For the current Data collection, For the current The heat that is effectively utilized, For the current times the heat that is not effectively utilized; The calculation formula for correcting the initial heat loss is: ； in, is the corrected heat loss; The specific steps for calculating the coal average particle value and particle deviation are as follows: Take multiple samples of the pulverized coal entering the furnace within each time interval, count the size of the pulverized coal particles in each sample, and calculate the average particle size and particle deviation of the coal within the time interval: The calculation formula for the average particle size of coal is: ； in, is the average particle size of coal, is the number of sampling times, For the Sampling, is the number of coal particles in each sampling, For the The size of the coal particles; The calculation formula for particle deviation is: ； in, is the particle deviation, is the number of sampling times, For the Sampling, is the number of coal particles in each sampling, For the The size of the coal particles; The furnace parameters include the amount of coal fed into the furnace and the furnace temperature; The steam parameters include feed water flow, main steam flow, main steam pressure, main steam temperature, and reheat steam temperature; The air volume parameters include primary air volume, primary air temperature, secondary air volume, secondary air temperature, and burnout air volume; The flue gas parameters include flue gas temperature and flue gas flow rate; The calculation formula for the normalization process is: ； in, For each parameter type, the normalized data is is the largest number of each parameter type, is the smallest number of each parameter type, For each parameter type data, The number of data in each parameter type.

2. The method for predicting heat loss of coal-fired power generation according to claim 1, characterized in that: The calculation formula of the coal heat distribution model is: ； in, Input heat to the boiler, For the boiler to effectively utilize heat, Heat loss due to exhaust gas, Heat loss due to incomplete combustion of gas. Heat loss due to incomplete combustion of solids, To dissipate heat loss, The heat loss from the physical heat of the ash is is the measurement error value.

3. The method for predicting heat loss of coal-fired power generation according to claim 1, characterized in that: The boiler parameters include flue gas volume, flue gas specific heat capacity, flue gas temperature, concentration of combustible gas in flue gas, slag volume, coal content in slag, boiler insulation coefficient, boiler area, boiler surface temperature, slag specific heat capacity, slag discharge temperature, steam production, steam specific enthalpy, feed water specific enthalpy, and coal heat; The calculation formula for the heat that is not effectively utilized is: ； in, The heat that is not effectively utilized is Heat loss due to exhaust gas, Heat loss due to incomplete combustion of gas. Heat loss due to incomplete combustion of solids, To dissipate heat loss, Heat loss is due to physical heat of ash; Exhaust heat loss The calculation formula is: ； in, is the smoke exhaust volume, is the specific heat capacity of flue gas, is the flue gas temperature, is the ambient temperature; Heat loss from incomplete combustion of gas The calculation formula is: ； in, is the smoke exhaust volume, is the concentration of combustible gas in the flue gas, is the heat of combustible gas; Heat loss from incomplete combustion of solids The calculation formula is: ； in, is the slag amount, is the coal content in the slag, For coal heat; Heat loss from heat dissipation The calculation formula is: ； in, is the boiler insulation coefficient, is the boiler area, is the boiler surface temperature, is the ambient temperature; Physical heat loss of ash The calculation formula is: ； in, is the slag amount, is the specific heat capacity of slag, is the temperature when the slag is discharged, is the ambient temperature; The effectively utilized heat The calculation formula is: ； in, is the steam production, is the specific enthalpy of steam, is the feed water specific enthalpy.

4. The method for predicting heat loss from coal-fired power generation according to claim 1, characterized in that: The calculation formula of the boiler input heat is: ； in, is the coal consumption, for the heat of coal; The calculation formula of the measurement error value is: ； in, is the measurement error value.

5. The method for predicting heat loss of coal-fired power generation according to claim 1, characterized in that: The specific steps of establishing a coal-fired heat loss prediction model through a neural network and training the model are as follows: The neural network is an LSTM neural network prediction model, which consists of three gates: forget gate, input gate, and output gate: The equation of the forget gate is: ； in, is the current switch state of the forget gate, represents the sigmoid function, is the forget gate weight, is the forget gate bias parameter, is the hidden state at the previous moment, Input switch for current data; Determine how much of the unit state at the previous moment needs to be retained to the current moment; The equation for the input gate is: ； in, is the current switch state of the input gate, represents the sigmoid function, is the input gate weight, is the input gate bias parameter, is the output hidden state at the previous moment, is the prediction data set of the previous moment; ； in, Candidate cell states, To represent the hyperbolic tangent function, is the candidate cell weight, is the candidate cell state bias parameter, is the hidden state at the previous moment, Input switch for current data; Determine how much of the network's input data needs to be saved to the unit state at the current moment; The update equation is: ； in, Current cell state, is the cell state at the previous moment; The output gate equation is: ； ； in, is the current switch state of the output gate, represents the sigmoid function, is the output gate weight, is the output gate bias parameter, is the hidden state at the previous moment, Input switch for current data, is the corrected heat loss at the current moment; Controls how much of the current unit state needs to output the corrected loss heat at the current moment; The training steps are: The prediction data set of the previous moment is used as the input of the prediction model, and the corrected heat loss at the current moment is used as the output of the model to train the model; The parameters optimized for the prediction model include learning rate, number of hidden units, time step, and regularization parameter; The constructed optimization model is based on the sparrow search optimization algorithm. The sparrow population consists of discoverers, joiners, and a certain number of scouts. The specific optimization logic is: The discoverer's position update formula is as follows: ； in, is the current iteration number, is the dimension of the parameter to be optimized, , is the total number of parameters to be optimized, For the A sparrow in the Location information in the dimension; is the maximum number of iterations; is a random number, , and are alarm values ​​and safety thresholds; for dimensional matrix; is a random number that conforms to the normal distribution; The position update formula of the joiner is as follows: ； Where: is the optimal position of the current discoverer; The current worst position globally; for A matrix whose elements are randomly assigned 1 or -1, is the sparrow population size; The scout position update formula is as follows: ； in, is the global optimal position, is the iteration step length, is the fitness value of the current sparrow individual, and are the current global optimal and worst fitness values, is a random number, , is a constant to prevent the denominator from being 0; The root mean square error is used to calculate the fitness, and the calculation formula is as follows: ； in, is the number of samples, is the prediction data set of the previous moment, is the corrected heat loss.

6. A thermal power generation coal-fired heat loss prediction system, characterized by: The thermal power generation coal-fired heat loss prediction system is used to execute the thermal power generation coal-fired heat loss prediction method according to any one of claims 1 to 5, comprising: A heat distribution module is used to obtain the type of heat distribution data of the thermal power generation boiler to be predicted, including boiler input heat, boiler effective utilization heat, exhaust heat loss, gas incomplete combustion heat loss, solid incomplete combustion heat loss, heat loss due to heat dissipation, and ash physical heat loss, and to establish a coal heat distribution model; a heat calculation module, which is used to calculate the heat that is not effectively utilized and the heat that is effectively utilized, respectively, based on the coal heat distribution model and the heat distribution data type, by obtaining boiler parameters and ambient temperature at the same time interval, to form a preliminary heat loss data set and a preliminary heat utilization data set; a loss correction module, which is used to calculate a measurement error value based on a preliminary heat loss data set, a preliminary heat utilization data set, and the boiler input heat in each time interval, calculate a loss-use ratio based on the heat loss and the heat utilization, correct the preliminary heat loss, and obtain a corrected heat loss; A model building module is used to obtain the coal particle size in each time interval to calculate the coal average particle value and particle deviation, and obtain the unit load, furnace parameters, steam parameters, air volume parameters and flue gas parameters through normalization processing to form a prediction data set, establish a coal heat loss prediction model through a neural network, and train the model; The model optimization module is used to establish an optimization model for the coal-fired heat loss prediction model, optimize the parameters of the prediction model, accelerate the training of the model, and use the current prediction data set as the input of the prediction model to predict the coal-fired heat loss at future times.

7. A device for predicting heat loss from coal-fired power generation, characterized by: The device includes a processor and a storage medium, wherein the storage medium stores a computer program. When the computer program is executed by the processor, it can implement the method for predicting heat loss of coal-fired power generation according to any one of claims 1 to 5.

Images