Method and device for predicting coal consumption of thermal power supply
Patent Information
- Application Number
- CN202611525248.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-29
- Publication Date
- 2026-10-30
AI Technical Summary
[0003]本申请的目的是提供一种火电供电煤耗量的预测方法及装置，旨在解决现有技术中火电供电煤耗预测模型难以跟随机组运行工况动态变化实现自适应适配的技术问题
本申请的技术方案通过构建火电供电煤耗非线性动态数学模型，以机组负荷和平均热值作为输入变量，以供电煤耗量作为输出变量，能够完整捕获机组运行过程中输入条件与煤耗输出之间的映射关系；通过对所述非线性动态数学模型的静态非线性参数以及动态线性参数进行同步辨识，可一次性完成两类参数的协同求解，保持模型静态特性与动态特性的匹配一致性，减少分步辨识带来的参数耦合偏差；在此基础上根据机组最新运行数据，对所述模型初始参数向量进行迭代更新，获得更新后的供电煤耗预测模型，能够跟随机组实际运行状态持续修正模型内部参数，让模型跟随工况变化完成自适应调整，再利用所述更新后的供电煤耗预测模型输出供电煤耗量预测结果，有利于提升供电煤耗预测结果和机组真实运行工况的贴合程度，得到准确度更高的供电煤耗量预测输出。
Smart Images
Figure CN123014346A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of thermal power prediction technology, and in particular to a method and apparatus for predicting coal consumption for thermal power generation. Background Technology
[0002] In the electricity spot market environment, the predicted coal consumption for power generation directly supports power plant power generation trading decisions, playing a crucial role in helping thermal power companies control fuel costs and improve market operating profits. Existing coal consumption prediction technologies for thermal power plants mostly employ a step-by-step parameter identification approach to solve for model parameters. After parameter identification, these parameters are often fixed and used in coal consumption prediction calculations, resulting in weak dynamic adaptive adjustment capabilities. During long-term operation, thermal power units are susceptible to interference from multiple factors such as seasonal changes, equipment degradation, frequent switching of operating conditions, and fluctuations in the quality of coal fed into the furnace, leading to continuous time-varying drift characteristics in unit operation. Traditional prediction methods struggle to dynamically adapt and update parameters in line with real-time changes in unit operation. Model adaptability gradually decreases over time, resulting in discrepancies between the predicted coal consumption and the actual coal consumption level of the unit. This leads to insufficient prediction accuracy and stability, failing to meet the high accuracy and robustness requirements of refined trading decisions in the electricity spot market. The difficulty in adapting to the real-time operating status of the generating unit results in a limited match between the predicted coal consumption output and the actual operating conditions of the unit. This leads to a tendency for discrepancies between the predicted output and the actual coal consumption, making it difficult to meet the actual usage requirements for coal consumption prediction accuracy in the context of electricity spot trading. Summary of the Invention
[0003] The purpose of this application is to provide a method and apparatus for predicting coal consumption for thermal power generation, aiming to solve the technical problem that existing coal consumption prediction models for thermal power generation are difficult to adapt to dynamic changes in unit operating conditions.
[0004] To achieve the above objectives, this application provides a method for predicting coal consumption for thermal power generation, comprising: A nonlinear dynamic mathematical model of coal consumption for power generation in thermal power plants is constructed, with unit load and average calorific value as input variables and coal consumption for power generation as output variable. The static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model are identified simultaneously to obtain the initial parameter vector of the model. Based on the latest operating data of the unit, the initial parameter vector of the model is iteratively updated to obtain the updated coal consumption prediction model for power supply. The updated coal consumption prediction model for power supply is used to output the predicted coal consumption for power supply.
[0005] In some optional embodiments, the discrete input-output relationship of the nonlinear dynamic mathematical model for coal consumption in thermal power generation satisfies the following expression: , In the formula, The coal consumption for power supply at time k; The coal consumption for power supply at time k-1; The coal consumption for power supply at time k-2; The unit load at time k-1; The average calorific value at time k-1; , For dynamic linear parameters; , , It is a static nonlinear parameter.
[0006] In some optional embodiments, the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model are simultaneously identified to obtain the initial parameter vector of the model, including: Configure the initial values for the covariance matrix and the parameter vector to be identified; Obtain historical operating data of the generator unit and construct a regression vector; The identification operation is performed iteratively to simultaneously solve for the static nonlinear parameters and the dynamic linear parameters, thereby obtaining the initial parameter vector of the model.
[0007] In some optional embodiments, the identification operation is performed iteratively, simultaneously solving for the static nonlinear parameters and the dynamic linear parameters to obtain the initial parameter vector of the model, including: A one-step identification method based on recursive least squares is adopted to calculate the gain matrix of the current step based on the regression vector of the current step and the covariance matrix of the previous step. The current step covariance matrix is updated based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix. Based on the current step gain matrix, actual coal consumption for power supply, the current step regression vector, and the previous step parameter vector to be identified, the current step parameter vector to be identified is updated and iterated until convergence. The converged current step parameter vector to be identified is then determined as the initial parameter vector of the model.
[0008] In some optional embodiments, the formula used to calculate the current step gain matrix based on the current step regression vector and the previous step covariance matrix is as follows: , In the formula, This is the gain matrix for the nth step; The covariance matrix at step n-1; This is the regression vector for the nth step; for The transpose of ; The formula used to update the current step covariance matrix based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix is as follows: , In the formula, Let I be the covariance matrix at step n; I is the identity matrix. This is the gain matrix for the nth step; This is the regression vector for the nth step; The covariance matrix at step n-1; for The transpose of .
[0009] In some optional embodiments, the current step parameter vector to be identified is updated based on the current step gain matrix, the actual coal consumption for power supply, the current step regression vector, and the parameter vector to be identified in the previous step. This process is iterated until convergence, and the converged current step parameter vector to be identified is determined as the initial parameter vector of the model, including: Calculate the prediction residual between the actual coal consumption for power supply and the coal consumption estimate obtained based on the parameter vector to be identified in the previous step and the regression vector in the current step; Based on the prediction residual, the current step gain matrix, and the previous step parameter vector to be identified, the current step parameter vector to be identified is obtained by using a preset parameter update formula. The steps of calculating the gain matrix, updating the covariance matrix, and updating the current step parameter vector to be identified are performed repeatedly until the preset convergence condition is met. The converged current step parameter vector to be identified is then determined as the initial parameter vector of the model.
[0010] In some optional embodiments, the expression for the preset parameter update formula is: , In the formula, This is the vector of parameters to be identified in the nth step; This is the parameter vector to be identified in the (n-1)th step; This is the gain matrix for the nth step; This represents the actual coal consumption for power supply in step n. This is the regression vector for the nth step; for The transpose of .
[0011] In some optional embodiments, the initial parameter vector of the model is iteratively updated based on the latest operating data of the unit to obtain an updated power supply coal consumption prediction model, including: Obtain the latest operating data of the unit and construct the corresponding regression vector; The pre-configured covariance matrix and the parameter vector to be identified are used as the initial values of the previous time step in the iterative operation; The parameter update operation is performed iteratively to correct the parameter vector to be identified and obtain the updated coal consumption prediction model for power supply.
[0012] In some optional embodiments, the parameter update operation is performed iteratively to correct the parameter vector to be identified, thereby obtaining an updated power supply coal consumption prediction model, including: Calculate the gain matrix for the current step based on the regression vector of the current step and the covariance matrix of the previous step; The current step covariance matrix is updated based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix. The current step parameter vector to be identified is updated based on the current step gain matrix, the actual coal consumption for power supply, the current step regression vector, and the previous step parameter vector to be identified.
[0013] This application also provides a device for predicting coal consumption for thermal power generation, the device being used to execute the above-described method for predicting coal consumption for thermal power generation, the device comprising: The model building module is used to construct a nonlinear dynamic mathematical model of coal consumption for power generation in thermal power plants, with unit load and average calorific value as input variables and coal consumption for power generation as output variables. The parameter synchronization identification module is used to synchronously identify the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model to obtain the initial parameter vector of the model. The parameter iteration update module is used to iteratively update the initial parameter vector of the model based on the latest operating data of the unit, so as to obtain the updated power supply coal consumption prediction model. The prediction output module is used to output the prediction result of power supply coal consumption using the updated power supply coal consumption prediction model.
[0014] The above-mentioned technical solution of this application has at least the following beneficial technical effects: The technical solution of this application constructs a nonlinear dynamic mathematical model for coal consumption in thermal power generation, using unit load and average calorific value as input variables and coal consumption for power generation as output variable. This model can completely capture the mapping relationship between input conditions and coal consumption output during unit operation. By simultaneously identifying the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model, the two types of parameters can be solved collaboratively at once, maintaining the consistency between the static and dynamic characteristics of the model and reducing parameter coupling deviations caused by step-by-step identification. Based on this, the initial parameter vector of the model is iteratively updated according to the latest operating data of the unit to obtain an updated coal consumption prediction model for power generation. This model can continuously correct the internal parameters of the model according to the actual operating state of the unit, allowing the model to adapt to changes in operating conditions. The updated coal consumption prediction model is then used to output the predicted coal consumption for power generation, which helps to improve the fit between the predicted coal consumption for power generation and the actual operating conditions of the unit, resulting in a more accurate predicted coal consumption output for power generation. Attached Figure Description
[0015] Figure 1 This is a flowchart illustrating a method for predicting coal consumption for thermal power generation according to an embodiment of this application. Figure 2 This is a schematic diagram of the synchronous identification process of recursive least squares parameters according to an embodiment of this application. Detailed Implementation
[0016] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to specific embodiments and accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of this application. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.
[0017] The embodiments described in this application are only some, not all, of the embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments described herein without inventive effort are within the scope of protection of this application.
[0018] In the electricity spot market environment, the predicted coal consumption for power generation directly supports power plant power generation trading decisions, playing a crucial role in helping thermal power companies control fuel costs and improve market operating profits. Existing coal consumption prediction technologies for thermal power plants mostly employ a step-by-step parameter identification approach to solve for model parameters. After parameter identification, these parameters are often fixed and used in coal consumption prediction calculations, resulting in weak dynamic adaptive adjustment capabilities. During long-term operation, thermal power units are susceptible to interference from multiple factors such as seasonal changes, equipment degradation, frequent switching of operating conditions, and fluctuations in the quality of coal fed into the furnace, leading to continuous time-varying drift characteristics in unit operation. Traditional prediction methods struggle to dynamically adapt and update parameters in line with real-time changes in unit operation. Model adaptability gradually decreases over time, resulting in discrepancies between the predicted coal consumption and the actual coal consumption level of the unit. This leads to insufficient prediction accuracy and stability, failing to meet the high accuracy and robustness requirements of refined trading decisions in the electricity spot market. The difficulty in adapting to the real-time operating status of the generating unit results in a limited match between the predicted coal consumption output and the actual operating conditions of the unit. This leads to a tendency for discrepancies between the predicted output and the actual coal consumption, making it difficult to meet the actual usage requirements for coal consumption prediction accuracy in the context of electricity spot trading.
[0019] To address the aforementioned technical problems, this application provides a method for predicting coal consumption for thermal power generation.
[0020] The technical solution of this application constructs a nonlinear dynamic mathematical model for coal consumption in thermal power generation, using unit load and average calorific value as input variables and coal consumption for power generation as output variable. This model can completely capture the mapping relationship between input conditions and coal consumption output during unit operation. By simultaneously identifying the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model, the two types of parameters can be solved collaboratively at once, maintaining the consistency between the static and dynamic characteristics of the model and reducing parameter coupling deviations caused by step-by-step identification. Based on this, the initial parameter vector of the model is iteratively updated according to the latest operating data of the unit to obtain an updated coal consumption prediction model for power generation. This model can continuously correct the internal parameters of the model according to the actual operating state of the unit, allowing the model to adapt to changes in operating conditions. The updated coal consumption prediction model is then used to output the predicted coal consumption for power generation, which helps to improve the fit between the predicted coal consumption for power generation and the actual operating conditions of the unit, resulting in a more accurate predicted coal consumption output for power generation.
[0021] In some alternative embodiments, please refer to Figure 1 The method for predicting coal consumption for thermal power generation includes the following steps: Step 1: Construct a nonlinear dynamic mathematical model of coal consumption for power generation, using unit load and average calorific value as input variables and coal consumption for power generation as output variable; Specifically, a nonlinear dynamic mathematical model for coal consumption in thermal power generation is constructed. This model consists of a series of static nonlinear elements and a dynamic linear element, with unit load and average calorific value as input variables and coal consumption for power generation as output variable. The discrete input-output relationship of this nonlinear dynamic mathematical model for coal consumption in thermal power generation satisfies the following expression: , In the formula, The coal consumption for power supply at time k; The coal consumption for power supply at time k-1; The coal consumption for power supply at time k-2; The unit load at time k-1; The average calorific value at time k-1; , For dynamic linear parameters; , , It is a static nonlinear parameter.
[0022] Step 2: Simultaneously identify the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model to obtain the initial parameter vector of the model; In some optional embodiments, step 2 includes the following specific steps: Step 21: Configure the initial values of the covariance matrix and the parameter vector to be identified; Specifically, the initial values of the covariance matrix and the parameter vector to be identified are configured. The initial value of the covariance matrix is set as a diagonal matrix, with all diagonal elements assigned to preset large positive numbers and off-diagonal elements assigned to 0. The initial values of the parameter vector to be identified are all assigned to 0, thus completing the initialization of the recursive least squares one-step identification method. The initial value of the covariance matrix is denoted as... The initial value of the parameter vector to be identified is denoted as .
[0023] Step 22: Obtain historical operating data of the unit and construct a regression vector; Specifically, historical operating data of the generating units is acquired, and historical unit load, historical average calorific value, and historical coal consumption for power supply are extracted at fixed sampling intervals. Outlier sampling points exceeding three standard deviations are eliminated using the 3σ criterion, and first-order exponential filtering is applied to reduce data noise. A regression vector is constructed according to the model structure. The elements of the regression vector are taken in sequence. , , , , , , , .
[0024] Step 23: Iteratively perform the identification operation, simultaneously solve the static nonlinear parameters and dynamic linear parameters, and obtain the initial parameter vector of the model.
[0025] In some optional embodiments, step 23 further includes the following specific steps: Step 231: Using a one-step identification method based on recursive least squares, calculate the gain matrix of the current step based on the regression vector of the current step and the covariance matrix of the previous step. In some optional embodiments, the formula used to calculate the current step gain matrix based on the current step regression vector and the previous step covariance matrix is as follows: , In the formula, This is the gain matrix for the nth step; The covariance matrix at step n-1; This is the regression vector for the nth step; for The transpose of .
[0026] Step 232: Update the current step covariance matrix based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix; The formula used to update the current step covariance matrix based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix is as follows: , In the formula, Let I be the covariance matrix at step n; I is the identity matrix. This is the gain matrix for the nth step; This is the regression vector for the nth step; The covariance matrix at step n-1; for The transpose of .
[0027] Step 233: Based on the current step gain matrix, actual coal consumption for power supply, current step regression vector, and the parameter vector to be identified in the previous step, update the parameter vector to be identified in the current step, iterate until convergence, and determine the converged parameter vector to be identified in the current step as the initial parameter vector of the model.
[0028] In some optional embodiments, step 233 further includes the following specific steps: Calculate the prediction residual between the actual coal consumption for power supply and the coal consumption estimate obtained based on the parameter vector to be identified in the previous step and the regression vector in the current step; Based on the prediction residual, the current step gain matrix and the previous step parameter vector to be identified, the current step parameter vector to be identified is obtained by using a preset parameter update formula. Specifically, the expression for the preset parameter update formula is: , In the formula, This is the vector of parameters to be identified in the nth step; This is the parameter vector to be identified in the (n-1)th step; This is the gain matrix for the nth step; This represents the actual coal consumption for power supply in step n. This is the regression vector for the nth step; for The transpose of .
[0029] The algorithm iteratively performs the steps of calculating the gain matrix, updating the covariance matrix, and updating the parameter vector to be identified in the current step. After each iteration, the covariance matrix and parameter vector to be identified in the current step are assigned the values of the previous step's covariance matrix and parameter vector to be identified in the next iteration, respectively. Convergence is determined by setting two conditions: the first condition is that the number of iterations reaches the preset maximum number of iterations, and the second condition is that the L2 norm of the parameter vectors to be identified in two consecutive iterations is less than a preset threshold. The iteration terminates when either condition is met, and the converged parameter vector to be identified in the current step is determined as the initial parameter vector of the model.
[0030] Step 3: Based on the latest operating data of the unit, iteratively update the initial parameter vector of the model to obtain the updated coal consumption prediction model for power supply; In some optional embodiments, step 3 further includes the following specific steps: Step 31: Obtain the latest operating data of the unit and construct the corresponding regression vector; Specifically, the latest operating data of the unit is obtained, and real-time unit load, real-time average calorific value, and real-time coal consumption for power supply are collected at the same sampling period; the 3σ criterion is used to remove abrupt abnormal data, and the same first-order exponential filtering algorithm is used to reduce noise in the real-time data; according to the vector construction rules in step 22, the regression vector at the current moment is constructed based on the preprocessed real-time data.
[0031] Step 32: Use the configured covariance matrix and the parameter vector to be identified as the initial values of the previous time step in the iterative operation; Specifically, using the covariance matrix and the parameter vector to be identified obtained after the iteration convergence in step 2, the converged covariance matrix is directly assigned as the covariance matrix of the previous step in this round of online iteration, and the converged parameter vector to be identified is assigned as the parameter vector to be identified of the previous step in this round of online iteration, thus starting the online model update process.
[0032] Step 33: Iteratively execute parameter update operations, correct the parameter vector to be identified, and continuously update the static nonlinear parameters and dynamic linear parameters in the nonlinear dynamic mathematical model of coal consumption for thermal power supply to obtain the updated coal consumption prediction model for power supply.
[0033] In some optional embodiments, step 33 further includes the following specific steps: Step 331: Calculate the gain matrix for the current step based on the regression vector of the current step and the covariance matrix of the previous step; Step 332: Update the current step covariance matrix based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix; Step 333: Update the current step's parameter vector based on the current step's gain matrix, actual coal consumption for power supply, current step's regression vector, and the previous step's parameter vector to be identified.
[0034] Specifically, based on the current step gain matrix, actual coal consumption for power supply, current step regression vector, and the previous step parameter vector to be identified, the preset parameter update formula is substituted to update the current step parameter vector to be identified. After the update is completed, the current step covariance matrix and the current step parameter vector to be identified are used as the previous step parameters for the next round of online iteration, thus completing one round of online correction of model parameters.
[0035] Step 4: Output the predicted coal consumption for power supply using the updated coal consumption prediction model.
[0036] Specifically, the unit load and average calorific value at the time to be predicted are input into the updated coal consumption prediction model for power supply; the vector operation rules of the nonlinear dynamic mathematical model for coal consumption of thermal power supply are substituted into the discrete input-output expression of the model to complete the calculation and output the prediction result of coal consumption for power supply at the corresponding time.
[0037] The technical solution of this application will be described in detail below with reference to a specific embodiment.
[0038] In this embodiment, operational data collected from a thermal power unit is used to predict the coal consumption for power generation. The sampling interval is set to 1 minute, and three types of time-series data—unit load, average calorific value, and coal consumption for power generation—are continuously collected. The raw sampled data is preprocessed first, and then the model input and output datasets are constructed. Please refer to [link to relevant documentation]. Figure 2 , Figure 2The flowchart of the one-step identification iterative calculation using the recursive least squares method illustrates the complete iterative calculation process of the one-step identification method based on the recursive least squares method used in this embodiment. The process begins with the input matrix A, composed of the processed control quantity and the controlled variable. It sequentially completes initial value assignment, iterative calculation of the gain matrix, covariance matrix, and the parameter vector to be identified, parameter back substitution, convergence determination, and finally outputs the converged model parameters, used to simultaneously identify the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model of coal consumption for thermal power generation. This embodiment first constructs a nonlinear dynamic mathematical model of coal consumption for thermal power generation. This model consists of a series of static nonlinear elements and a dynamic linear element, using unit load and average calorific value as input variables and coal consumption for power generation as the output variable. The discrete input-output relationship of this nonlinear dynamic mathematical model of coal consumption for thermal power generation satisfies the following expression: , In the formula, The coal consumption for power supply at time k; The coal consumption for power supply at time k-1; The coal consumption for power supply at time k-2; The unit load at time k-1; The average calorific value at time k-1; , For dynamic linear parameters; , , It is a static nonlinear parameter.
[0039] Preprocessing is performed on the original time-series data of unit load, average calorific value, and coal consumption for power supply. First, the 3σ criterion is used to remove outlier sampling points exceeding three standard deviations. Then, a first-order exponential filter is used to reduce noise in the remaining data, with a filter coefficient of 0.1. Based on the discrete input-output relationship of the nonlinear dynamic mathematical model of coal consumption for thermal power supply, the preprocessed data is arranged into a matrix A, where each row of the matrix A corresponds to a regression vector at a sampling time. The elements of the regression vector are sequentially... , , , , , , , The total number of rows in matrix A is denoted as That is, the total length of the data samples used for identification.
[0040] For covariance matrix and the parameter vector to be identified Assign initial values, covariance matrix Set as an 8th order diagonal matrix, and assign all diagonal elements a value of 1. Off-diagonal elements are assigned the value 0; the parameter vector to be identified This is an 8-dimensional column vector with all elements set to 0, completing the initialization of the recursive least squares one-step identification method.
[0041] Set iteration index Starting from 2, the value is selected and entered into a loop iteration phase. In each iteration, the gain matrix of the current step is calculated sequentially. Current step covariance matrix The parameter vector to be identified in the current step The value of ; the formula used to calculate the gain matrix of the current step is: The formula used to calculate the gain matrix of the current step is: , In the formula, This is the gain matrix for the nth step; The covariance matrix at step n-1; This is the regression vector for the nth step; for The transpose vector. In this embodiment, the regression vector in the nth step. Take the 8-dimensional column vector corresponding to the nth row of matrix A, and the covariance matrix at the (n-1)th step. Take the covariance matrix obtained from the previous iteration, and use the initial value in the first iteration. .
[0042] After obtaining the gain matrix for the current step, the covariance matrix for the current step is updated using the following formula: , In the formula, Let I be the covariance matrix at step n; I is the identity matrix. This is the gain matrix for the nth step; This is the regression vector for the nth step; The covariance matrix at step n-1; for The transpose of .
[0043] After obtaining the covariance matrix of the current step, update the parameter vector to be identified in the current step, and first calculate the actual coal consumption for power supply in the nth step. The prediction residual between the predicted coal consumption and the estimated coal consumption is equal to the transpose of the regression vector of the current step. Left-multiply the parameter vector to be identified in the previous step The predicted residual equals Subtract the estimated coal consumption; then, based on the prediction residual, the current step gain matrix, and the parameter vector to be identified in the previous step, calculate using the preset parameter update formula to obtain the current step parameter vector to be identified. The expression of the preset parameter update formula is: , In the formula, This is the vector of parameters to be identified in the nth step; This is the parameter vector to be identified in the (n-1)th step; This is the gain matrix for the nth step; This represents the actual coal consumption for power supply in step n. This is the regression vector for the nth step; for The transpose of the vector; the parameter vector to be identified in the previous step during the first iteration. Get initial value .
[0044] After each iteration, a parameter back-substitution operation is performed, and the current step's parameter vector to be identified is... Assign the value to the parameter vector to be identified in the previous step. The current step covariance matrix Assign the value to the covariance matrix from the previous step This prepares parameters for the next round of iteration; then, a convergence check is performed to determine the iteration index. Does it meet the requirements? ,in Let A be the total length of the data samples in matrix A; if this condition is not met, then the iterative index will be... Increment by 1, then return to the steps of calculating the current step gain matrix, the current step covariance matrix, and the current step parameter vector to be identified, and continue iterating. If the judgment condition is met, the iteration process terminates. At this point, the converged current step parameter vector to be identified is the model initial parameter vector, which also contains dynamic linear parameters. , With static nonlinear parameters , , The identification results.
[0045] After obtaining the initial parameter vector of the model, the latest operating data of the unit is collected. Real-time unit load, real-time average calorific value, and real-time coal consumption for power supply are obtained at a sampling period of 1 minute. Abnormal data with sudden changes are removed using the 3σ criterion. The real-time data is denoised using a first-order exponential filtering algorithm. According to the aforementioned regression vector construction rules, the regression vector for the current moment is constructed based on the preprocessed real-time data. The covariance matrix and the initial parameter vector of the model obtained by iterative convergence are used as the initial parameters for online iteration. The gain matrix calculation, covariance matrix update, and parameter vector update to be identified are performed in sequence. The parameter vector to be identified is corrected, and the static nonlinear parameters and dynamic linear parameters in the nonlinear dynamic mathematical model of coal consumption for power supply are continuously updated to obtain the updated coal consumption prediction model for power supply.
[0046] The unit load and average calorific value at the time to be predicted are input into the updated coal consumption prediction model for power supply. The discrete input-output expression of the nonlinear dynamic mathematical model for coal consumption of thermal power is then substituted into the model to complete vector operations, and the predicted coal consumption for power supply at the corresponding time is output.
[0047] This application also provides a device for predicting coal consumption for thermal power generation. The device is used to execute the above-described method for predicting coal consumption for thermal power generation. The device includes: The model building module is used to construct a nonlinear dynamic mathematical model of coal consumption for power generation in thermal power plants, with unit load and average calorific value as input variables and coal consumption for power generation as output variables. The parameter synchronization identification module is used to synchronously identify the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model to obtain the initial parameter vector of the model. The parameter iteration update module is used to iteratively update the initial parameter vector of the model based on the latest operating data of the unit, so as to obtain the updated power supply coal consumption prediction model. The prediction output module is used to output the prediction results of coal consumption for power supply using the updated coal consumption prediction model.
[0048] The thermal power coal consumption prediction device provided in this application, employing the thermal power coal consumption prediction method described in the above embodiments, can solve the technical problem that the thermal power coal consumption prediction model is difficult to adapt to the dynamic changes in unit operating conditions. The beneficial effects of the thermal power coal consumption prediction device provided in this application are the same as those of the thermal power coal consumption prediction method provided in the above embodiments, and other technical features in the thermal power coal consumption prediction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.
[0049] Furthermore, this application also provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the method for predicting coal consumption for thermal power generation in Embodiment 1 above. The electronic device in this application embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. The electronic device provided in this application, employing the method for predicting coal consumption for thermal power generation in the above embodiments, can solve the technical problem that the coal consumption prediction model for thermal power generation is difficult to adaptively adapt to the dynamic changes in the unit's operating conditions. The beneficial effects of the electronic device provided in this application are the same as those of the method for predicting coal consumption for thermal power generation provided in the above embodiments, and other technical features of this electronic device are the same as those disclosed in the previous embodiment method, and will not be repeated here.
[0050] Furthermore, this application also provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the method for predicting coal consumption for thermal power generation in the above embodiments. The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. The readable storage medium provided in this application can solve the technical problem that the coal consumption prediction model for thermal power generation is difficult to adaptively adapt to the dynamic changes in the unit's operating conditions. The beneficial effects of the computer-readable storage medium provided in this application are the same as those of the method for predicting coal consumption for thermal power generation provided in the above embodiments, and will not be repeated here.
[0051] This application aims to protect a method and apparatus for predicting coal consumption for thermal power generation. The technical solution of this application constructs a nonlinear dynamic mathematical model for coal consumption in thermal power generation, using unit load and average calorific value as input variables and coal consumption for power generation as the output variable. This model can completely capture the mapping relationship between input conditions and coal consumption output during unit operation. By simultaneously identifying the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model, the two types of parameters can be solved collaboratively in one step, maintaining the consistency between the static and dynamic characteristics of the model and reducing parameter coupling deviations caused by step-by-step identification. Based on this, the initial parameter vector of the model is iteratively updated according to the latest operating data of the unit to obtain an updated coal consumption prediction model for power generation. This model can continuously correct its internal parameters according to the actual operating state of the unit, allowing the model to adaptively adjust to changes in operating conditions. The updated coal consumption prediction model is then used to output the predicted coal consumption for power generation, which helps improve the fit between the predicted coal consumption and the actual operating conditions of the unit, resulting in a more accurate predicted coal consumption output.
[0052] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of this application and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of this application should be included within the protection scope of this application. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.
Claims
1. A method for predicting coal consumption for thermal power generation, characterized in that, include: A nonlinear dynamic mathematical model of coal consumption for power generation in thermal power plants is constructed, with unit load and average calorific value as input variables and coal consumption for power generation as output variable. The static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model are identified simultaneously to obtain the initial parameter vector of the model. Based on the latest operating data of the unit, the initial parameter vector of the model is iteratively updated to obtain the updated coal consumption prediction model for power supply. The updated coal consumption prediction model for power supply is used to output the predicted coal consumption for power supply.
2. The method for predicting coal consumption for thermal power generation according to claim 1, characterized in that, The discrete input-output relationship of the nonlinear dynamic mathematical model for coal consumption in thermal power generation satisfies the following expression: ， In the formula, The coal consumption for power supply at time k; The coal consumption for power supply at time k-1; The coal consumption for power supply at time k-2; The unit load at time k-1; The average calorific value at time k-1; , For dynamic linear parameters; , , It is a static nonlinear parameter.
3. The method for predicting coal consumption for thermal power generation according to claim 1, characterized in that, The static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model are simultaneously identified to obtain the initial parameter vector of the model, including: Configure the initial values for the covariance matrix and the parameter vector to be identified; Obtain historical operating data of the generator unit and construct a regression vector; The identification operation is performed iteratively to simultaneously solve for the static nonlinear parameters and the dynamic linear parameters, thereby obtaining the initial parameter vector of the model.
4. The method for predicting coal consumption for thermal power generation according to claim 3, characterized in that, The identification operation is performed iteratively, simultaneously solving for the static nonlinear parameters and the dynamic linear parameters to obtain the initial parameter vector of the model, including: A one-step identification method based on recursive least squares is adopted to calculate the gain matrix of the current step based on the regression vector of the current step and the covariance matrix of the previous step. The current step covariance matrix is updated based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix. Based on the current step gain matrix, actual coal consumption for power supply, the current step regression vector, and the previous step parameter vector to be identified, the current step parameter vector to be identified is updated and iterated until convergence. The converged current step parameter vector to be identified is then determined as the initial parameter vector of the model.
5. The method for predicting coal consumption for thermal power generation according to claim 4, characterized in that, The formula used to calculate the gain matrix of the current step, based on the regression vector of the current step and the covariance matrix of the previous step, is as follows: ， In the formula, This is the gain matrix for the nth step; The covariance matrix at step n-1; This is the regression vector for the nth step; for The transpose of ; The formula used to update the current step covariance matrix based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix is as follows: ， In the formula, Let I be the covariance matrix at step n; I is the identity matrix. This is the gain matrix for the nth step; This is the regression vector for the nth step; The covariance matrix at step n-1; for The transpose of .
6. The method for predicting coal consumption for thermal power generation according to claim 4, characterized in that, Based on the current step gain matrix, actual coal consumption for power supply, the current step regression vector, and the previous step parameter vector to be identified, the current step parameter vector to be identified is updated and iterated until convergence. The converged current step parameter vector to be identified is then determined as the initial parameter vector of the model, including: Calculate the prediction residual between the actual coal consumption for power supply and the coal consumption estimate obtained based on the parameter vector to be identified in the previous step and the regression vector in the current step; Based on the prediction residual, the current step gain matrix, and the previous step parameter vector to be identified, the current step parameter vector to be identified is obtained by using a preset parameter update formula. The steps of calculating the gain matrix, updating the covariance matrix, and updating the current step parameter vector to be identified are performed repeatedly until the preset convergence condition is met. The converged current step parameter vector to be identified is then determined as the initial parameter vector of the model.
7. The method for predicting coal consumption for thermal power generation according to claim 6, characterized in that, The expression for the preset parameter update formula is: ， In the formula, This is the vector of parameters to be identified in the nth step; This is the parameter vector to be identified in the (n-1)th step; This is the gain matrix for the nth step; This represents the actual coal consumption for power supply in step n. This is the regression vector for the nth step; for The transpose of .
8. The method for predicting coal consumption for thermal power generation according to claim 1, characterized in that, Based on the latest operating data of the generating unit, the initial parameter vector of the model is iteratively updated to obtain the updated coal consumption prediction model for power supply, including: Obtain the latest operating data of the unit and construct the corresponding regression vector; The pre-configured covariance matrix and the parameter vector to be identified are used as the initial values of the previous time step in the iterative operation; The parameter update operation is performed iteratively to correct the parameter vector to be identified and obtain the updated coal consumption prediction model for power supply.
9. The method for predicting coal consumption for thermal power generation according to claim 8, characterized in that, Iteratively perform parameter update operations, correct the parameter vector to be identified, and obtain the updated power supply coal consumption prediction model, including: Calculate the gain matrix for the current step based on the regression vector of the current step and the covariance matrix of the previous step; The current step covariance matrix is updated based on the current step gain matrix, the current step regression vector, and the previous step covariance matrix. The current step parameter vector to be identified is updated based on the current step gain matrix, the actual coal consumption for power supply, the current step regression vector, and the previous step parameter vector to be identified.
10. A device for predicting coal consumption for thermal power generation, characterized in that, The apparatus is used to perform the method for predicting coal consumption for thermal power generation as described in any one of claims 1 to 9, the apparatus comprising: The model building module is used to construct a nonlinear dynamic mathematical model of coal consumption for power generation in thermal power plants, with unit load and average calorific value as input variables and coal consumption for power generation as output variables. The parameter synchronization identification module is used to synchronously identify the static nonlinear parameters and dynamic linear parameters of the nonlinear dynamic mathematical model to obtain the initial parameter vector of the model. The parameter iteration update module is used to iteratively update the initial parameter vector of the model based on the latest operating data of the unit, so as to obtain the updated power supply coal consumption prediction model. The prediction output module is used to output the prediction result of power supply coal consumption using the updated power supply coal consumption prediction model.