Thermal power generating unit operation optimization method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of thermal power generating unit energy consumption optimization, in particular to a thermal power generating unit operation optimization method and device, electronic equipment and a storage medium. According to the method, a refined multi-state energy consumption prediction model, an uncontrollable parameter prediction model and a unit load prediction model are constructed by deeply mining historical operation data, and are organically combined by using an intelligent optimization algorithm, so that the power grid dispatching efficiency can be improved on the premise of ensuring that the unit output load strictly meets the power grid dispatching requirements (preset conditions). The optimal controllable parameter combination enabling the energy consumption of the unit to be the lowest is automatically and accurately searched; according to the method, the transformation from extensive experience operation to data-driven accurate optimization is realized, the economical efficiency of the thermal power generating unit under peak regulation operation is remarkably improved, and scientific and automatic decision support is provided for energy conservation and consumption reduction of a power plant.

Description

A method, apparatus, electronic device and storage medium for optimizing the operation of thermal power units Technical Field

[0001] This invention relates to the field of energy consumption optimization technology for thermal power units, specifically to a method, apparatus, electronic device, and storage medium for optimizing the operation of thermal power units. Background Technology

[0002] Coal is the primary energy source, and this energy endowment determines the important position of coal-fired power units in energy supply and energy security. Clean, efficient, low-carbon, and safe operation is the inevitable path for the development of traditional thermal power. Deeply exploring the energy-saving potential of coal-fired power units is an important transformation task for current thermal power units. At the same time, in order to adapt to the intermittent and fluctuating nature of new energy power generation, traditional coal-fired power units participate in deep peak shaving at a high proportion. In actual production, the frequency and magnitude of load adjustment have increased.

[0003] With the development of data analysis technology, more and more scholars are applying it to research on thermal power unit diagnosis and operation optimization. There are many studies on optimizing unit operating parameters, but most of them focus on a specific parameter, such as main steam pressure and unit load. There is little research on the performance analysis of main and auxiliary equipment and the overall operation optimization of the unit. Summary of the Invention

[0004] This invention provides a method, apparatus, electronic device, and storage medium for optimizing the operation of thermal power units, in order to address the problem that there is a lack of research on the performance analysis of the main and auxiliary equipment of thermal power units and the overall operation optimization of the units in the prior art.

[0005] In a first aspect, the present invention provides a method for optimizing the operation of a thermal power unit. The method includes: acquiring historical operating state parameters of the thermal power unit and calculating the energy consumption of the thermal power unit based on the historical operating state parameters, wherein the historical operating state parameters and energy consumption constitute a unit state dataset; dividing the unit state dataset into different types of datasets based on the load change direction in the historical operating state parameters, and constructing energy consumption prediction models corresponding to different types of datasets by combining a time-series prediction algorithm or a non-time-series prediction algorithm; constructing an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating state parameters, and constructing a unit load prediction model based on the relationship between the historical operating state parameters and the unit load; using a preset optimization algorithm combined with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption, and using the controllable parameter corresponding to the optimal energy consumption when the unit load meets preset conditions as the unit operation optimization strategy.

[0006] This invention, through in-depth mining of historical operating data, constructs a refined multi-state energy consumption prediction model, an uncontrollable parameter prediction model, and a unit load prediction model. It then organically combines these models using intelligent optimization algorithms. This allows for the automatic and precise search for the optimal combination of controllable parameters that minimizes unit energy consumption, while ensuring that the unit's output load strictly meets the grid dispatch requirements (preset conditions). This method represents a shift from extensive experience-based operation to data-driven precision optimization, significantly improving the economic efficiency of thermal power units under peak-shaving operation and providing scientific and automated decision support for power plant energy conservation and consumption reduction.

[0007] In one optional implementation, the historical operating state parameters include historical operating state parameters at multiple times. Calculating the energy consumption of the thermal power unit based on the historical operating state parameters includes: preprocessing the historical operating state parameters at each time point; calculating the energy loss caused by irreversible losses in the thermal power unit using the preprocessed historical operating state parameters according to the second law of thermodynamics; determining the increase in unit power generation energy consumption corresponding to the energy loss using a single-consumption analysis method, and obtaining the unit energy consumption at each time point by combining it with the unit's theoretical power generation energy consumption; and combining the preprocessed historical operating state parameters at each time point with the corresponding unit energy consumption to obtain a unit state dataset.

[0008] This invention combines loss analysis based on the second law of thermodynamics with practical unit consumption analysis in engineering, enabling the scientific and accurate calculation of the true energy consumption of thermal power units at different operating times. This method not only reveals the specific magnitude and source of irreversible losses during energy conversion, but more importantly, it transforms abstract losses into intuitive increases in coal consumption for power generation, thereby constructing a high-quality unit status dataset. This provides a reliable data foundation for subsequent accurate energy consumption prediction, operational status assessment, and the formulation of optimization strategies.

[0009] In one optional implementation, the unit status dataset is divided into different types of datasets based on the load change direction in the historical operating status parameters, and energy consumption prediction models corresponding to different types of datasets are constructed by combining time-series prediction algorithms or non-time-series prediction algorithms. This includes: selecting parameters from the historical operating status parameters whose impact on energy consumption is higher than a first threshold and whose correlation with each other is lower than a second threshold as energy consumption features; constructing an original training set and an original test set based on the energy consumption features and the corresponding unit energy consumption; dividing the original training set and the original test set into a steady-state training set, a steady-state test set, a load increase training set, a load increase test set, a load decrease training set, and a load decrease test set based on the load change direction in the historical operating status parameters; and constructing corresponding steady-state energy consumption prediction models, load increase energy consumption prediction models, and load decrease energy consumption prediction models based on the steady-state training set, the steady-state test set, the load increase training set, the load increase test set, and the load decrease training set, and the load decrease test set, combined with the time-series prediction algorithm or non-time-series prediction algorithm.

[0010] In this invention, by meticulously considering the direction of unit load changes, historical operating data is accurately divided into different types of datasets such as steady state, load increase, and load decrease. A dedicated energy consumption prediction model is then constructed to effectively overcome the limitations of traditional single models in accurately describing the dynamic operating characteristics of units. This method significantly improves the accuracy and relevance of energy consumption prediction under different operating conditions by screening key energy consumption features with strong and weak correlations and combining the advantages of time-series and non-time-series algorithms. It provides reliable data support and decision-making basis for the refined management and energy conservation of thermal power units under the background of flexible peak shaving.

[0011] In one optional implementation, based on steady-state training sets and steady-state test sets, load increase training sets and load increase test sets, and load decrease training sets and load decrease test sets, and combined with the time-series prediction algorithm or non-time-series prediction algorithm, corresponding steady-state energy consumption prediction models, load increase energy consumption prediction models, and load decrease energy consumption prediction models are constructed, including: constructing an initial steady-state energy consumption prediction model using a non-time-series prediction algorithm based on the steady-state training sets and steady-state test sets; and constructing corresponding time-series original energy consumption prediction models and non-time-series original energy consumption prediction models, time-series load increase energy consumption prediction models and non-time-series load increase energy consumption prediction models, and time-series load decrease energy consumption prediction models based on the original training sets, load increase training sets, and load decrease training sets, respectively, using time-series prediction algorithms and non-time-series prediction algorithms. A time-series and non-time-series load decrease energy consumption prediction model was developed. Based on the original test set, load increase test set, and load decrease test set, the models with higher accuracy among the time-series original energy consumption prediction model and non-time-series original energy consumption prediction model, the time-series load increase energy consumption prediction model and non-time-series load increase energy consumption prediction model, and the time-series load decrease energy consumption prediction model and non-time-series load decrease energy consumption prediction model were selected as the initial original energy consumption prediction model, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model. The accuracy of the initial steady-state energy consumption prediction model, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model was compared with the initial original energy consumption prediction model, and the model with higher accuracy was selected as the steady-state energy consumption prediction model, the load increase energy consumption prediction model, and the load decrease energy consumption prediction model.

[0012] This invention meticulously divides the unit's operating state into three typical conditions: steady state, load increase, and load decrease, and constructs dedicated training and testing sets for each condition. Based on this, it innovatively employs a selection mechanism that combines parallel modeling and cross-validation of time-series and non-time-series prediction algorithms. First, the optimal algorithm model is selected within each condition; then, it is compared a second time with a general original model to ultimately determine the optimal energy consumption prediction model suitable for different operating conditions. This method effectively overcomes the deficiency of insufficient prediction accuracy of a single model under complex dynamic conditions, significantly improving the model's adaptability and prediction accuracy to specific state changes, and providing reliable data support for refined energy consumption management and operation optimization of thermal power units.

[0013] In one optional implementation, an uncontrollable parameter prediction model is constructed based on the mapping relationship between controllable and uncontrollable parameters in the historical operating state parameters, and a unit load prediction model is constructed based on the relationship between the historical operating state parameters and the unit load. This includes: dividing the parameters in the energy consumption characteristics into controllable parameters, uncontrollable parameters, and environmental parameters; constructing an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the original training set and the original test set; and constructing a unit load prediction model based on the relationship between the energy consumption characteristics and the unit load in the original training set and the original test set.

[0014] In this invention, by scientifically classifying the unit's energy consumption characteristic parameters into controllable parameters, uncontrollable parameters, and environmental parameters, and constructing uncontrollable parameter prediction models and unit load prediction models based on historical data, the intrinsic correlation and influence mechanism between operating parameters can be accurately quantified. This method not only achieves accurate prediction of uncontrollable operating results, but also reliably predicts the unit's load capacity under specific parameter combinations, thereby providing key model support for subsequent operation optimization and effectively improving the feasibility of optimization results and the economy of unit operation.

[0015] In one optional implementation, a preset optimization algorithm is used in conjunction with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption. The controllable parameters corresponding to the optimal energy consumption when the unit load meets preset conditions are used as the unit operation optimization strategy. This includes: generating an initial population using the preset optimization algorithm, the initial population including multiple individuals, each individual being a combination of controllable parameters; determining the uncontrollable parameters corresponding to different individuals based on the uncontrollable parameter prediction model; inputting different individuals, their corresponding uncontrollable parameters, and current environmental parameters as different energy consumption features into the unit load prediction model and the energy consumption prediction model respectively to obtain the corresponding unit load and unit energy consumption; iterating the population using the preset optimization algorithm, repeating the process of determining uncontrollable parameters, unit load, and unit energy consumption until the lowest unit energy consumption when the unit load meets preset conditions is obtained, and using the controllable parameter combination corresponding to the lowest unit energy consumption as the unit operation optimization strategy.

[0016] This invention organically combines intelligent optimization algorithms with unit energy consumption, uncontrollable parameters, and load prediction models to construct a complete closed-loop optimization process: First, multiple sets of controllable parameter combinations are generated as an initial population. Then, the prediction model is used to accurately calculate the unit load and energy consumption corresponding to each set of parameters. Through an iterative screening mechanism, the system automatically finds the controllable parameter combination that minimizes the overall system energy consumption under strictly preset load conditions. This method can scientifically and accurately determine the optimal operating strategy, significantly improve the economy and intelligence level of thermal power unit operation, and effectively achieve energy conservation and consumption reduction.

[0017] In one optional implementation, an initial population is generated using a preset optimization algorithm, including: determining the operating condition identification parameters in the historical operating state parameters; based on the operating condition identification parameters, using a clustering algorithm to divide the unit state dataset into operating condition categories to obtain datasets and cluster centers corresponding to different operating conditions; obtaining the operating condition identification parameters of the current unit state, and using a clustering algorithm and the cluster centers to determine the current operating condition of the unit; and generating an initial population using a preset optimization algorithm based on the historical upper and lower limits of controllable parameters under the current operating condition.

[0018] In this invention, by performing cluster analysis on the operating condition identification parameters in the historical operating parameters, different typical operating conditions and their cluster centers are scientifically divided. This method can quickly and accurately identify the specific operating condition to which the unit belongs based on its current state, and generate an initial population based on the upper and lower limits of the controllable parameters under that operating condition, thus avoiding the problem of easily getting trapped in local optima due to excessively large upper and lower limits.

[0019] Secondly, the present invention provides a thermal power unit operation optimization device, the device comprising: a dataset construction module, used to acquire historical operating state parameters of the thermal power unit and calculate the energy consumption of the thermal power unit based on the historical operating state parameters, wherein the historical operating state parameters and energy consumption constitute a unit state dataset; a first model construction module, used to divide the unit state dataset into different types of datasets based on the load change direction in the historical operating state parameters, and to construct energy consumption prediction models corresponding to different types of datasets by combining a time-series prediction algorithm or a non-time-series prediction algorithm; a second model construction module, used to construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating state parameters, and to construct a unit load prediction model based on the relationship between the historical operating state parameters and the unit load; and an optimization module, used to use a preset optimization algorithm combined with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters and the corresponding unit load and energy consumption, and to use the controllable parameters corresponding to the optimal energy consumption when the unit load meets preset conditions as the unit operation optimization strategy.

[0020] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the thermal power unit operation optimization method of the first aspect or any corresponding embodiment described above.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the thermal power unit operation optimization method of the first aspect or any corresponding embodiment described above.

[0022] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the thermal power unit operation optimization method described in the first aspect or any corresponding embodiment. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 is a schematic diagram of the first type of operation optimization method for thermal power units according to an embodiment of the present invention; Figure 2 is a schematic diagram of the second type of operation optimization method for thermal power units according to an embodiment of the present invention; Figure 3 is a schematic diagram of energy consumption prediction model construction according to an embodiment of the present invention; Figure 4 is a schematic diagram of load constraint optimization model construction according to an embodiment of the present invention; Figure 5 is a structural block diagram of thermal power unit operation optimization device according to an embodiment of the present invention; Figure 6 is a schematic diagram of the hardware structure of electronic equipment according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0027] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0028] According to an embodiment of the present invention, a method for optimizing the operation of thermal power units is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] This embodiment provides a method for optimizing the operation of thermal power units. Figure 1 is a flowchart of the method for optimizing the operation of thermal power units according to an embodiment of the present invention. As shown in Figure 1, the process includes the following steps: Step S101, obtaining historical operating status parameters of the thermal power unit, and calculating the energy consumption of the thermal power unit based on the historical operating status parameters. The historical operating status parameters and energy consumption constitute a unit status dataset.

[0030] Specifically, historical operating status parameters of thermal power units can be collected through information management systems or platforms. These parameters include the working fluid temperature, pressure, and flow rate of the main and auxiliary equipment such as the boiler, turbine, and auxiliary systems, as well as environmental parameters (including temperature and humidity) at multiple historical moments. For example, parameters such as boiler main steam pressure, boiler main steam temperature, third-stage superheater outlet temperature, third-stage superheater outlet pressure, main steam flow rate, and unit load can be collected. Similar data can be sorted and saved chronologically. Furthermore, to improve the accuracy of subsequent energy consumption prediction models, the time interval for collecting historical operating status parameters should be as short as possible.

[0031] Furthermore, since information management platforms typically do not store unit energy consumption parameters, this embodiment calculates the overall energy consumption of the thermal power unit using historical operating status parameters to provide a data foundation for the energy consumption prediction model. The specific calculation method for unit energy consumption can be implemented using relevant technologies and will not be elaborated upon here.

[0032] Step S102: Based on the load change direction in the historical operating status parameters, the unit status dataset is divided into different types of datasets, and energy consumption prediction models corresponding to different types of datasets are constructed by combining time-series prediction algorithms or non-time-series prediction algorithms. Specifically, in constructing the energy consumption prediction model, this embodiment does not directly use historical operating status parameters and energy consumption data as input and output for model training; instead, it first considers the change direction of unit load in the historical operating status parameters, such as whether the unit load is in a steady state, rising, or falling, and divides the unit status dataset based on this, while using the divided datasets to construct the corresponding energy consumption prediction models.

[0033] From the perspective of energy consumption calculation, the energy consumption of a unit at a given moment is only related to the boiler's operating parameters at that moment. Therefore, a non-time-series predictive algorithm can be used to establish an energy consumption prediction model. Considering that when the unit operates under varying loads, the control system will formulate control strategies and adjust operating parameters based on load commands and the current operating state of the unit. Such changes directly affect the energy consumption level of the equipment and also affect the energy consumption at the next moment. Therefore, although unit energy consumption is a state variable and independent of time, the energy consumption level at the current moment still has a certain impact on the energy consumption at the next moment. Therefore, when constructing an energy consumption prediction model, either a time-series predictive algorithm or a non-time-series predictive algorithm can be used. The time-series predictive algorithm can be a Long Short-Term Memory (LSTM) prediction algorithm, and the non-time-series predictive algorithm can be a Least Square Support Vector Machine (LSSVM) algorithm. It is understood that this application is not limited to the mentioned LSTM and LSSVM algorithms, and other algorithms may be used in practice.

[0034] Step S103: Construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating status parameters, and construct a unit load prediction model based on the relationship between the historical operating status parameters and the unit load. Specifically, the historical operating status parameters include controllable and uncontrollable parameters. Both types of parameters have a certain impact on unit energy consumption. However, uncontrollable parameters cannot be adjusted during unit operation. Therefore, this embodiment constructs an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters. That is, given the controllable parameters, this uncontrollable parameter prediction model can be used to directly predict the uncontrollable parameters.

[0035] Furthermore, this embodiment also constructs a unit load prediction model based on the relationship between historical operating status parameters and unit load, which is used to predict the unit load under different parameters, providing a data basis for determining whether the unit load meets the preset conditions.

[0036] Step S104: Using a preset optimization algorithm combined with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model, the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption, are determined. The controllable parameter corresponding to the optimal energy consumption when the unit load meets the preset conditions is used as the unit operation optimization strategy.

[0037] Specifically, in order to determine the unit operation optimization strategy, that is, to determine the controllable parameters under optimal energy consumption, this embodiment first inputs different controllable parameters into the uncontrollable parameter prediction model to obtain the corresponding uncontrollable parameters; then, the controllable parameters and the corresponding uncontrollable parameters are combined and input into the energy consumption prediction model and the unit load prediction model respectively to obtain the unit energy consumption and unit load corresponding to different controllable parameters; in this process, a preset algorithm is used to optimize the controllable parameters to find the optimal energy consumption when the unit load meets the preset conditions, and the controllable parameters corresponding to this energy consumption are used as the unit operation optimization strategy.

[0038] This embodiment provides a method for optimizing the operation of thermal power units. The method includes the following steps: Step S201, obtaining historical operating status parameters of the thermal power unit, and calculating the energy consumption of the thermal power unit based on the historical operating status parameters. The historical operating status parameters and energy consumption constitute a unit status dataset.

[0039] Specifically, step S201 includes: step S2021, preprocessing the historical operating state parameters at each time point.

[0040] Because the historical operating status parameters may be affected by the accuracy of the measuring instruments and the measurement methods, data with large errors or errors may be collected. For example, in actual production, when a coal mill is stopped, the value of its coal feed measurement point is not 0, but still has a small value. For such obviously erroneous data, manual correction can be performed.

[0041] Furthermore, although the operating loads of the units are similar, different operating states and adjustment methods can lead to significant differences in the values ​​of the same parameter. It can be assumed that the unit parameters under the same load condition follow a normal distribution. In this embodiment, the 3-sigma criterion is used for outlier detection. First, the units are sorted in descending order of load and divided into multiple intervals, each containing 50 state data points. Then, the mean e and standard deviation μ of the data within each interval are calculated, and data outside the range [e-3μ, e+3μ] are recorded as outliers. Finally, to ensure the continuity and smoothness of the dataset, Lagrange interpolation is used to calculate a new value to replace the outlier based on the two data points before and after it. After outlier cleanup, the data order is restored.

[0042] In this embodiment, extreme value standardization is used to process the parameters after outlier processing, in order to eliminate the influence of various data units and numerical values ​​on the accuracy of the prediction model. Specifically, normalization can be used to standardize the data. It is understood that in other embodiments, the method is not limited to the 3-sigma criterion and extreme value standardization mentioned above, and other preprocessing methods can be used depending on the amount of data collected and the complexity of the features.

[0043] Step S2022: According to the second law of thermodynamics, the energy loss caused by irreversible losses in the thermal power unit is calculated using pre-processed historical operating state parameters. Specifically, the second law of thermodynamics characterizes that any actual energy conversion process is irreversible, and will always be accompanied by a decrease or loss of energy quality. Energy (Exergy) is a physical quantity that measures energy quality or the ability to do work. Energy loss indicates that in any actual process, due to irreversibility, a portion of energy will always be lost and cannot be converted into useful work. Energy loss is a direct indicator for measuring the magnitude of process irreversibility and revealing the nature and location of energy loss. Based on the second law of thermodynamics, energy loss = input energy - output energy. For thermal power units, the energy value carried by the working fluid (such as water, steam, flue gas) flowing into and out of the equipment can be calculated using the historical operating state parameters at each moment to determine the input energy and output energy, thereby determining the energy loss. The energy loss determined in this embodiment includes the energy loss caused by irreversible losses in each environment or equipment in the thermal power unit.

[0044] Step S2023: Using the unit consumption analysis method, determine the increase in unit power generation energy consumption corresponding to the loss, and combine it with the theoretical power generation energy consumption of the unit to obtain the unit energy consumption at each time. Combine the preprocessed historical operating state parameters at each time with the unit energy consumption at the corresponding time to obtain the unit state dataset.

[0045] Specifically, by employing unit consumption analysis, fuel loss can be linked to commonly used fuel consumption indicators in the engineering field, thus more intuitively meeting the needs of engineers. Based on unit consumption analysis theory... Where P is the output power of the generator, B is the amount of fuel consumed, and e p e f These represent the ratio of output electrical energy to input fuel, respectively. This refers to the sum of losses incurred in each stage (or device) of the energy transfer process during the conversion of fuel chemical energy into electrical energy due to its irreversibility; furthermore, it includes... ,in, This refers to the energy consumption for power generation by the generating unit; The theoretical energy consumption for power generation of the unit is 0.123 kg / (kW·h); Let X be the increase in unit power generation energy consumption corresponding to the loss of the i-th link (or equipment), also known as the energy consumption of that link or equipment. The unit energy consumption is obtained by summing the theoretical energy consumption with the energy consumption of each link (or equipment). The historical operating state parameters at each moment are combined with the corresponding unit energy consumption as a state data point, denoted by X1, X2...X... n This represents the n state parameters obtained for calculating energy consumption, and the i-th state data point Y. i =[b i , X 1i ,X 2i ,...X ni All status data points are arranged in chronological order, and their set is the unit status dataset.

[0046] Step S202: Based on the load change direction in the historical operating status parameters, the unit status dataset is divided into different types of datasets, and energy consumption prediction models corresponding to different types of datasets are constructed by combining time-series prediction algorithms or non-time-series prediction algorithms.

[0047] Specifically, step S202 includes: step S2021, selecting parameters from the historical operating status parameters whose impact on energy consumption is higher than a first threshold and whose correlation with each other is lower than a second threshold as energy consumption features.

[0048] Specifically, the collected historical operating status parameters can all be used as features of unit energy consumption to characterize the unit's energy consumption level. However, the impact of each feature on unit energy consumption varies, and there may be strong correlations between them (repeated features). In order to reduce the amount of data, improve the modeling speed and save computing resources, this embodiment performs feature extraction.

[0049] In this embodiment, the mRMR (Minimum Redundancy-Maximum Relevance) algorithm is used for initial feature extraction. The specific extraction process can be implemented using relevant technologies. Then, Pearson correlation coefficient analysis is used to remove highly correlated duplicate features. After these two feature selection steps, only features that significantly impact energy consumption and have weak pairwise correlations are retained from the state data points in the unit state dataset. These features are renamed as sample points (i.e., energy consumption features), with m representing the number of features and Z representing the i-th sample point. i =[b i ,X 1i ,X 2i ,...X mi The set is named the sample dataset, and the features contained in the sample points are the inputs to the subsequent energy consumption prediction model, while the unit energy consumption is the output.

[0050] Step S2022: Construct an original training set and an original test set based on the energy consumption characteristics and the corresponding unit energy consumption. Specifically, for the sample dataset composed of energy consumption characteristics and unit energy consumption, it can be divided into an original training set and an original test set according to a certain ratio. For example, the training set and the test set can be divided in a 7:3 ratio.

[0051] Step S2023: Based on the load change direction in the historical operating state parameters, the original training set and the original test set are divided into a steady-state training set, a steady-state test set, a load increase training set, a load increase test set, a load decrease training set, and a load decrease test set.

[0052] Specifically, when determining the load change pattern, a sufficient number of sample points can be selected that simultaneously encompass steady-state, rising, and falling states. A sliding window method can then be used to process this time period (or all time periods can be directly windowed). Finally, based on the load fluctuations within each window, it can be determined whether the window is in steady-state or non-steady-state operation. For example, the standard deviation or range of the load for all units within the window can be calculated, and its magnitude, along with the load fluctuation threshold, can determine whether the window is in steady-state or non-steady-state operation.

[0053] For windows operating in a non-steady-state manner, the difference between the unit load and the load command from AGC (Automatic Generation Control) is further used to determine whether the window is a load increase window or a load decrease window. For example, the average or median value of the unit load within the window can be used as the unit load of the window, and then the difference between this value and the load command from AGC is calculated. Based on the relationship between this difference and a preset threshold, the window is determined to be a load increase window or a load decrease window.

[0054] After determining the steady-state window, load increase window, and load decrease window, select a time window with a wide load variation range from all load increase time windows in the original training set. For example, for a typical deep peak-shaving unit, the load variation should increase from 30% to 100%. Use the sample data within the corresponding time window as the load increase training set. Similarly, select a time window with a wide load variation range from all load decrease time windows in the original training set, and use the sample data within the corresponding time window as the load decrease training set. Randomly arrange the sample points from all steady-state operating time windows in the original training set to create the steady-state training set. Likewise, select a time window with a wide load variation range from all load increase time windows in the original test set, and use the sample data within the corresponding time window as the load increase test set. Similarly, select a time window with a wide load variation range from all load decrease time windows in the original test set, and use the sample data within the corresponding time window as the load decrease test set. Randomly arrange the sample points from all steady-state operating time windows in the original test set to create the steady-state test set.

[0055] Step S2024: Based on the steady-state training set and steady-state test set, the load increase training set and load increase test set, and the load decrease training set and load decrease test set, and in combination with the time-series prediction algorithm or the non-time-series prediction algorithm, construct the corresponding steady-state energy consumption prediction model, the load increase energy consumption prediction model, and the load decrease energy consumption prediction model.

[0056] When constructing the model, the weighted average algorithm (WAA) can be used to quickly determine the values ​​of hyperparameters in the prediction model.

[0057] For example, the specific steps for determining the hyperparameters of the LSSVM prediction algorithm using the WAA algorithm are as follows: (1) Set the WAA algorithm model parameters, including the population size and the number of iterations. Randomly generate the initial population according to the following formula. In the formula: N is the number of individuals in the population; Dim is the dimension of the search space, i.e., the number of hyperparameters; UB j LB j and are the upper and lower bounds of the j-th dimension solution, respectively; r is a random number between [0,1].

[0058] (2) Set the fitness value of an individual to the root mean square error of the prediction result of the LSSVM algorithm. The root mean square error can characterize the accuracy of the prediction. The smaller the value, the better. Compare the fitness values ​​of all individuals in the population to determine the optimal fitness value and the optimal individual.

[0059] (3) Perform population iteration. For each individual, a new individual is generated. Compare the fitness values ​​of the new and old individuals and retain the individual with the lower fitness value.

[0060] (4) After the last iteration, extract the hyperparameter values ​​from the best individual.

[0061] In an optional implementation, step S2024 includes: step a1, constructing an initial steady-state energy consumption prediction model based on the steady-state training set and the steady-state test set using a non-time-series prediction algorithm; specifically, since the sample points in the steady-state training set and the steady-state test set are randomly arranged, a non-time-series prediction algorithm can be used to construct the energy consumption prediction model.

[0062] Step a2: Based on the original training set, the load increase training set, and the load decrease training set, the corresponding time-series original energy consumption prediction model and non-time-series original energy consumption prediction model, time-series load increase energy consumption prediction model and non-time-series load increase energy consumption prediction model, and time-series load decrease energy consumption prediction model and non-time-series load decrease energy consumption prediction model are constructed by using time-series prediction algorithm and non-time-series prediction algorithm respectively.

[0063] Step a3: Based on the original test set, the load increase test set, and the load decrease test set, select the model with the higher accuracy among the time-series original energy consumption prediction model and the non-time-series original energy consumption prediction model, the time-series load increase energy consumption prediction model and the non-time-series load increase energy consumption prediction model, and the time-series load decrease energy consumption prediction model and the non-time-series load decrease energy consumption prediction model as the initial original energy consumption prediction model, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model.

[0064] Specifically, models can be constructed using time-series prediction algorithms and non-time-series prediction algorithms based on the original training set, the load increase training set, and the load decrease training set, respectively. Then, the corresponding test set is input into the corresponding model, and the model with higher accuracy is selected as the corresponding energy consumption prediction model.

[0065] Based on the original training set, a time-series original energy consumption prediction model was constructed using a time-series prediction algorithm, and then a non-time-series original energy consumption prediction model was constructed using a non-time-series prediction algorithm. The accuracy of both models was determined using the original test set, and the model with the higher accuracy was selected as the initial original energy consumption prediction model.

[0066] Based on the load increase training set, a time-series load increase energy consumption prediction model is constructed using a time-series prediction algorithm. Then, a non-time-series load increase energy consumption prediction model is constructed using a non-time-series prediction algorithm. For both models, the accuracy is determined using a load increase test set, and the model with the higher accuracy is used as the initial load increase energy consumption prediction model.

[0067] Based on the load descent training set, a time-series load descent energy consumption prediction model is constructed using a time-series prediction algorithm. Then, a non-time-series load descent energy consumption prediction model is constructed using a non-time-series prediction algorithm. For both models, the accuracy is determined using a load descent test set, and the model with the higher accuracy is used as the initial load descent energy consumption prediction model.

[0068] Step a4: Compare the accuracy of the initial steady-state energy consumption prediction model, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model with the initial original energy consumption prediction model, and select the model with the higher accuracy as the steady-state energy consumption prediction model, the load increase energy consumption prediction model, and the load decrease energy consumption prediction model.

[0069] Specifically, the steady-state test set can be substituted into the initial steady-state energy consumption prediction model and the initial raw energy consumption prediction model, and the model with higher accuracy between the two models can be selected as the steady-state energy consumption prediction model. Similarly, the load increase test set can be substituted into the initial load increase energy consumption prediction model and the initial raw energy consumption prediction model, and the model with higher accuracy between the two models can be selected as the load increase energy consumption prediction model. Likewise, the load decrease test set can be substituted into the initial load decrease energy consumption prediction model and the initial raw energy consumption prediction model, and the model with higher accuracy between the two models can be selected as the load decrease energy consumption prediction model.

[0070] Step S203: Construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating status parameters, and construct a unit load prediction model based on the relationship between the historical operating status parameters and the unit load.

[0071] Specifically, step S203 includes: step S2031, dividing the parameters in the energy consumption characteristics into controllable parameters, uncontrollable parameters, and environmental parameters; specifically, controllable parameters are unit parameters that can be remotely adjusted by operators, while uncontrollable parameters are unit parameters that are not directly adjustable during operation. In this embodiment, economizer inlet water temperature, intermediate point superheat, low-temperature reheater inlet steam temperature, air preheater inlet flue gas temperature, and reheater desuperheater inlet steam temperature are uncontrollable operating state parameters, while steam-water separator outlet steam pressure, main steam flow rate, final stage superheater inlet steam temperature, and low-temperature superheater outlet steam temperature are directly controllable operating state parameters. In addition, the parameters in the energy consumption characteristics also include environmental parameters. Therefore, the parameters in the energy consumption characteristics can be divided, for example, including m energy consumption characteristics, which can be divided into f controllable parameters, g uncontrollable parameters, and h environmental parameters, where m = f + g + h.

[0072] Step S2032: Construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the original training set and the original test set; specifically, the mapping relationship between controllable and uncontrollable parameters can be represented as [X1, X2, ... X...]. g ]=f([X1,X2,...X f The constructed uncontrollable parameter prediction model characterizes this mapping relationship. Specifically, a prediction algorithm from related technologies can be used to construct the model, taking f controllable parameters as input and the g uncontrollable parameters as output, so that the model can predict the corresponding uncontrollable parameters from the controllable parameters.

[0073] Step S2033: Construct a unit load prediction model based on the relationship between energy consumption characteristics and unit load in the original training set and the original test set. Specifically, a prediction algorithm from related technologies can be used, with m features [X1, X2, ... X...]. m Using [value] as input and unit load as output, a unit load prediction model is constructed so that the model can predict the corresponding unit load through energy consumption characteristics.

[0074] Step S204: Using a preset optimization algorithm combined with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model, the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption, are determined. The controllable parameter corresponding to the optimal energy consumption when the unit load meets the preset conditions is used as the unit operation optimization strategy.

[0075] Specifically, step S204 includes: step S2041, generating an initial population using a preset optimization algorithm. The initial population comprises multiple individuals, each representing a set of controllable parameters. Specifically, when generating the initial population, i.e., determining the multiple sets of controllable parameters, if the upper and lower limits of the parameters are directly determined based on historical operating state parameters, the parameter differences determined based on these limits will be significant, easily leading to local optima during subsequent optimization. Therefore, it is possible to first divide the operating conditions based on historical operating state parameters and determine the upper and lower limits of each operating condition. Then, analysis can be performed based on different operating conditions.

[0076] In an optional implementation, step S2041 includes: step b1, determining the operating condition identification parameter in the historical operating status parameters.

[0077] Specifically, in power plants, operating conditions are typically described using load percentages, such as 100% full load, 60% load, and 40% load. However, in actual operation, the unit's actual state is influenced by numerous factors, including equipment condition and environmental conditions. Relying solely on load percentages makes it difficult to accurately distinguish between different operating conditions. To more accurately classify operating conditions, other operating parameters besides load need to be considered. These parameters may include boiler pressure, temperature, main steam flow rate, condenser vacuum, and circulating water temperature. By comprehensively considering these parameters, operating conditions can be reclassified from multiple dimensions to better reflect the unit's actual operating status. This classification method improves the accuracy and operability of operating conditions, providing more accurate guidance and decision-making basis for power plant operation and management. In this embodiment, "actual unit load," "boiler outlet main steam pressure," and "circulating water inlet temperature" are selected as operating condition identification parameters.

[0078] Step b2: Based on the operating condition identification parameters, a clustering algorithm is used to divide the unit status dataset into operating conditions, thereby obtaining the datasets and cluster centers corresponding to different operating conditions.

[0079] Specifically, this embodiment uses the K-means clustering algorithm for clustering. The principle is to first randomly select K objects as initial cluster centers, then calculate the distance between each object and each cluster center, and assign each object to the nearest cluster center. Each cluster center and the objects assigned to it represent a cluster. Then, the cluster centers are recalculated based on the existing objects in the cluster. This process is repeated until a certain termination condition is met, usually the minimum sum of squared errors.

[0080] When performing clustering, the number of cluster centers can be set first, such as 200 in this embodiment. Then, based on the operating condition identification parameters, a clustering algorithm is used to divide the unit status dataset into operating conditions, resulting in 200 operating condition types. The average value of all data belonging to the same operating condition is regarded as the cluster center.

[0081] Step b3: Obtain the operating condition identification parameters of the current state of the unit, and use a clustering algorithm and the cluster center to determine the current operating condition of the unit; specifically, when determining the unit operation optimization strategy, firstly, through the operating condition identification parameters of the current state of the unit, the K-means clustering algorithm is used to find the closest cluster center to determine the operating condition to which the current state of the unit belongs.

[0082] Step b4: Based on the historical upper and lower limits of controllable parameters under the current operating condition, an initial population is generated using a preset optimization algorithm. Specifically, after determining the current operating condition, within the historical upper and lower limits of controllable parameters under this condition, an initial population is generated according to the initialization logic of the weighted average optimization algorithm. Each individual in the population represents a combination of controllable parameters.

[0083] Step S2042: Determine the uncontrollable parameters corresponding to different individuals based on the uncontrollable parameter prediction model; specifically, the combination of controllable parameters corresponding to each individual can be substituted into the uncontrollable parameter prediction model to obtain the combination of uncontrollable parameters corresponding to each individual.

[0084] Step S2043: Different individuals, their corresponding uncontrollable parameters, and current environmental parameters are input into the unit load prediction model and the energy consumption prediction model as different energy consumption characteristics to obtain the corresponding unit load and unit energy consumption. Specifically, the controllable parameters within each individual, the corresponding uncontrollable parameters of that individual, and the environmental parameters of the unit's current location are all input into the energy consumption prediction model and the unit load prediction model to obtain the unit energy consumption and load of the current individual, and the unit energy consumption of the current individual is used as the individual fitness value.

[0085] It should be noted that before inputting data into the energy consumption prediction model, the operating state of the unit must first be determined based on the unit load, such as steady state, load increase, or load decrease, and then the corresponding energy consumption prediction model should be selected for energy consumption prediction. If the unit is in a steady state, then the steady-state energy consumption prediction model should be selected for energy consumption prediction.

[0086] Step S2044: Use a preset optimization algorithm to perform population iteration, repeatedly determine uncontrollable parameters, unit load and unit energy consumption until the lowest unit energy consumption is obtained when the unit load meets the preset conditions, and use the controllable parameter combination corresponding to the lowest unit energy consumption as the unit operation optimization strategy.

[0087] Specifically, after determining the unit load and energy consumption, it is first determined whether the current individual load is within the given load range. If it is outside the range, the individual is deemed not to meet the load matching condition, its fitness value is recorded as infinity (inf), and the individual is discarded. Simultaneously, a preset optimization algorithm can be used for population iteration (e.g., a preset number of iterations) until the individual with the lowest predicted energy consumption that satisfies the load constraints is selected. The controllable parameter data contained within this individual constitutes the unit energy consumption operation optimization strategy. The preset optimization algorithm can be a weighted average optimization algorithm; in practical applications, other optimization algorithms can also be used.

[0088] This invention collects real-time parameters from the boiler, turbine, and auxiliary systems, calculates the energy consumption of each main and auxiliary equipment and the overall unit using unit energy consumption analysis, and performs outlier cleaning and standardization to construct a unit status dataset. It combines mechanistic analysis and data mining to screen core energy consumption indicators, establishes prediction models based on load change direction (steady-state / rising / falling) and non-time-series / time-series prediction algorithms, and quickly determines hyperparameters in the prediction algorithms using intelligent optimization algorithms to establish a multi-state energy consumption prediction model. It distinguishes the mapping relationship between controllable and uncontrollable parameters in the energy consumption prediction model input, determines the controllable parameter boundaries under different operating conditions through operating condition clustering, and establishes a load-constrained optimization model with unit load as a constraint to find the optimal combination of controllable parameters and obtain the unit operation optimization strategy. This invention aims to help staff understand the energy consumption distribution of coal-fired units under flexible peak shaving from an analytical perspective and provide support for boiler system operation optimization.

[0089] As one or more specific application embodiments of the present invention, as shown in Figure 2, the thermal power unit operation optimization method is implemented using the following process: S1, constructing a unit energy consumption database, that is, constructing a unit status dataset based on historical operating status parameters and energy consumption.

[0090] S2. Establish a multi-state energy consumption prediction model. This multi-state energy consumption prediction model includes the original energy consumption prediction model, the steady-state energy consumption prediction model, the load increase energy consumption prediction model, and the load decrease energy consumption prediction model, which can be represented as Model 1, Model 2, Model 3, and Model 4, respectively.

[0091] Specifically, as shown in Figure 3, the model building process includes: refining the operating state distinction of the input unit state dataset, dividing it into four types: raw data without state distinction, steady-state operating data, load increase data, and load decrease data, and building models 1 to 4 respectively; for each model, the construction process uses time-series prediction algorithms (such as LSTM) and non-time-series prediction algorithms (such as LSSVM) for parallel modeling, and uses intelligent optimization algorithms (such as WAA) to quickly determine the optimal hyperparameters of each algorithm, thereby generating prediction value 1 and prediction value 2 for the same training set; subsequently, the test set data is input into the models obtained by different algorithms, and the error between the prediction results and the actual values ​​is compared to select the best one, thereby selecting a better prediction model for each data state.

[0092] In addition, the steady-state test sets can be substituted into energy consumption prediction models 1 and 2 respectively, and the prediction results can be compared to measure the merits of the two prediction models. Similarly, the load increase test sets can be substituted into energy consumption prediction models 1 and 3 respectively, and the load decrease test sets can be substituted into energy consumption prediction models 1 and 4 respectively, and the best one can be selected. This constitutes a multi-state energy consumption prediction model that can accurately predict the energy consumption level of thermal power units under different operating conditions.

[0093] S3. Establish a load-constrained optimization model. A weighted average optimization algorithm can be used as the solution algorithm. Combining this algorithm with the uncontrollable parameter prediction model, energy consumption prediction model, and unit load prediction model yields the load-constrained optimization model.

[0094] Specifically, as shown in Figure 4, the model construction and solution process includes: first, optimizing the algorithm parameters; then, determining the current operating condition and randomly generating several individuals to form an initial population; for each individual in the population (i.e., a specific combination of controllable parameters), inputting it into the pre-established uncontrollable parameter prediction model, energy consumption prediction model, and unit load prediction model, thereby comprehensively considering the uncontrollable parameters determined by the controllable parameters and the given environmental parameters, and finally outputting two key prediction values: the unit energy consumption prediction value and the unit load prediction value; next, checking whether the unit load prediction value under this parameter combination meets the load range requirements, if not, then... The fitness value of this individual is denoted as inf (infinity), which means that this scheme is directly eliminated because it does not meet the basic operating requirements. If it is, the fitness value of this individual is set as the predicted energy consumption value of the unit (i.e., the goal is to minimize energy consumption). After calculating the fitness value of all individuals, the algorithm starts to perform population iteration, that is, to generate a new generation of population based on the fitness value, and repeat the previous steps again to determine whether the new generation of population meets the termination condition (such as reaching the maximum number of iterations or the quality of the solution no longer improving). If not, iterative optimization continues; if yes, the calculation is terminated, and the combination of controllable system parameters with the lowest energy consumption obtained by iterative search is output as the final operation optimization strategy.

[0095] This embodiment also provides a thermal power unit operation optimization device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0096] This embodiment provides a thermal power unit operation optimization device, as shown in Figure 5, comprising: a dataset construction module 51, used to acquire historical operating state parameters of the thermal power unit and calculate the energy consumption of the thermal power unit based on the historical operating state parameters, wherein the historical operating state parameters and energy consumption constitute a unit state dataset; a first model construction module 52, used to divide the unit state dataset into different types of datasets based on the load change direction in the historical operating state parameters, and to construct energy consumption prediction models corresponding to different types of datasets by combining a time-series prediction algorithm or a non-time-series prediction algorithm; a second model construction module 53, used to construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating state parameters, and to construct a unit load prediction model based on the relationship between the historical operating state parameters and the unit load; and an optimization module 54, used to use a preset optimization algorithm to combine the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters and the corresponding unit load and energy consumption, and to use the controllable parameters corresponding to the optimal energy consumption when the unit load meets preset conditions as the unit operation optimization strategy.

[0097] The thermal power unit operation optimization device provided in this embodiment of the invention can execute the thermal power unit operation optimization method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0098] Figure 6 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0099] Referring specifically to Figure 6, a schematic diagram of a suitable electronic device for implementing embodiments of the present invention is shown below. The electronic device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 11, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 12 or a program loaded from memory 18 into random access memory (RAM) 13. The RAM 13 also stores various programs and data required for the operation of the electronic device. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0100] Typically, the following devices can be connected to I / O interface 15: input devices 16 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 17 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 18 including, for example, magnetic tapes, hard disks, etc.; and communication devices 19. Communication device 19 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 shows an electronic device with various devices, it should be understood that it is not required to implement or have all the devices shown, and more or fewer devices may be implemented or have alternatively.

[0101] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 19, or installed from a memory 18, or installed from a ROM 12. When the computer program is executed by the processor 11, it performs the functions defined in the thermal power unit operation optimization method of the embodiments of the present invention.

[0102] The electronic device shown in Figure 6 is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0103] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the thermal power unit operation optimization method shown in the above embodiments is implemented.

[0104] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0105] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for optimizing the operation of thermal power units, characterized in that, The method includes: acquiring historical operating status parameters of thermal power units and calculating the energy consumption of thermal power units based on the historical operating status parameters, wherein the historical operating status parameters and energy consumption constitute a unit status dataset; dividing the unit status dataset into different types of datasets based on the load change direction in the historical operating status parameters, and constructing energy consumption prediction models corresponding to different types of datasets by combining time-series prediction algorithms or non-time-series prediction algorithms; constructing an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating status parameters, and constructing a unit load prediction model based on the relationship between the historical operating status parameters and the unit load; using a preset optimization algorithm to combine the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption, and using the controllable parameter corresponding to the optimal energy consumption when the unit load meets preset conditions as the unit operation optimization strategy.

2. The method according to claim 1, characterized in that, The historical operating status parameters include historical operating status parameters at multiple times. Calculating the energy consumption of the thermal power unit based on these historical operating status parameters includes: preprocessing the historical operating status parameters at each time point; calculating the energy loss caused by irreversible losses in the thermal power unit using the preprocessed historical operating status parameters according to the second law of thermodynamics; determining the increase in unit power generation energy consumption corresponding to the energy loss using a single-consumption analysis method, and obtaining the unit energy consumption at each time point by combining it with the unit's theoretical power generation energy consumption; and combining the preprocessed historical operating status parameters at each time point with the corresponding unit energy consumption to obtain a unit status dataset.

3. The method according to claim 1, characterized in that, Based on the load change direction in the historical operating status parameters, the unit status dataset is divided into different types of datasets. Energy consumption prediction models corresponding to these different types of datasets are constructed using either a time-series prediction algorithm or a non-time-series prediction algorithm. This includes: selecting parameters from the historical operating status parameters whose impact on energy consumption is higher than a first threshold and whose correlation with each other is lower than a second threshold as energy consumption features; constructing an original training set and an original test set based on the energy consumption features and the corresponding unit energy consumption; dividing the original training set and the original test set into a steady-state training set, a steady-state test set, a load increase training set, a load increase test set, a load decrease training set, and a load decrease test set based on the load change direction in the historical operating status parameters; and constructing corresponding steady-state energy consumption prediction models, load increase energy consumption prediction models, and load decrease energy consumption prediction models based on the steady-state training set, the steady-state test set, the load increase training set, the load increase test set, and the load decrease training set, and the load decrease test set, combined with the time-series prediction algorithm or a non-time-series prediction algorithm.

4. The method according to claim 3, characterized in that, Based on steady-state training and test sets, load increase training and test sets, and load decrease training and test sets, and in conjunction with the time-series prediction algorithm or non-time-series prediction algorithm, corresponding steady-state energy consumption prediction models, load increase energy consumption prediction models, and load decrease energy consumption prediction models are constructed. This includes: constructing an initial steady-state energy consumption prediction model using a non-time-series prediction algorithm based on the steady-state training and test sets; and constructing corresponding time-series and non-time-series energy consumption prediction models, time-series and non-time-series load increase energy consumption prediction models, and time-series and non-time-series load decrease energy consumption prediction models based on the original training sets, load increase training sets, and load decrease training sets, respectively, using time-series and non-time-series prediction algorithms. Energy consumption prediction models for load decrease: Based on the original test set, load increase test set, and load decrease test set, the models with higher accuracy among the time-series original energy consumption prediction model and non-time-series original energy consumption prediction model, the time-series load increase energy consumption prediction model and non-time-series load increase energy consumption prediction model, and the time-series load decrease energy consumption prediction model and non-time-series load decrease energy consumption prediction model are selected as the initial original energy consumption prediction models, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model. The accuracy of the initial steady-state energy consumption prediction model, the initial load increase energy consumption prediction model, and the initial load decrease energy consumption prediction model is compared with the initial original energy consumption prediction model, and the model with higher accuracy is selected as the steady-state energy consumption prediction model, the load increase energy consumption prediction model, and the load decrease energy consumption prediction model.

5. The method according to claim 3, characterized in that, An uncontrollable parameter prediction model is constructed based on the mapping relationship between controllable and uncontrollable parameters in the historical operating status parameters, and a unit load prediction model is constructed based on the relationship between the historical operating status parameters and the unit load. This includes: dividing the parameters in the energy consumption characteristics into controllable parameters, uncontrollable parameters, and environmental parameters; constructing an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the original training set and the original test set; and constructing a unit load prediction model based on the relationship between the energy consumption characteristics and the unit load in the original training set and the original test set.

6. The method according to claim 2, characterized in that, A preset optimization algorithm is used in conjunction with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters, as well as the corresponding unit load and energy consumption. The controllable parameters corresponding to the optimal energy consumption when the unit load meets preset conditions are used as the unit operation optimization strategy. This includes: generating an initial population using the preset optimization algorithm, wherein the initial population includes multiple individuals, each individual being a combination of controllable parameters; determining the uncontrollable parameters corresponding to different individuals based on the uncontrollable parameter prediction model; inputting different individuals, their corresponding uncontrollable parameters, and current environmental parameters as different energy consumption features into the unit load prediction model and the energy consumption prediction model respectively to obtain the corresponding unit load and unit energy consumption; iterating the population using the preset optimization algorithm, repeating the process of determining uncontrollable parameters, unit load, and unit energy consumption until the lowest unit energy consumption when the unit load meets preset conditions is obtained, and using the controllable parameter combination corresponding to the lowest unit energy consumption as the unit operation optimization strategy.

7. The method according to claim 6, characterized in that, The initial population is generated using a preset optimization algorithm, including: determining the operating condition identification parameters in the historical operating status parameters; based on the operating condition identification parameters, using a clustering algorithm to divide the unit status dataset into operating condition categories to obtain datasets and cluster centers corresponding to different operating conditions; obtaining the operating condition identification parameters of the current unit status, and using a clustering algorithm and the cluster centers to determine the current operating condition of the unit; and generating the initial population using a preset optimization algorithm based on the historical upper and lower limits of controllable parameters under the current operating condition.

8. A thermal power unit operation optimization device, characterized in that, The device includes: a dataset construction module, used to acquire historical operating status parameters of thermal power units and calculate the energy consumption of thermal power units based on the historical operating status parameters, wherein the historical operating status parameters and energy consumption constitute a unit status dataset; a first model construction module, used to divide the unit status dataset into different types of datasets based on the load change direction in the historical operating status parameters, and construct energy consumption prediction models corresponding to different types of datasets by combining a time-series prediction algorithm or a non-time-series prediction algorithm; a second model construction module, used to construct an uncontrollable parameter prediction model based on the mapping relationship between controllable and uncontrollable parameters in the historical operating status parameters, and construct a unit load prediction model based on the relationship between the historical operating status parameters and the unit load; and an optimization module, used to use a preset optimization algorithm combined with the energy consumption prediction model, the uncontrollable parameter prediction model, and the unit load prediction model to determine the uncontrollable parameters corresponding to different controllable parameters and the corresponding unit load and energy consumption, and to use the controllable parameters corresponding to the optimal energy consumption when the unit load meets preset conditions as the unit operation optimization strategy.

9. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the thermal power unit operation optimization method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the thermal power unit operation optimization method according to any one of claims 1 to 7.

Images