Load dynamic regulation method and system for optimizing house service power rate of waste incineration power plant

Abstract

The application discloses a kind of load dynamic regulation method and system of garbage incineration power plant auxiliary power rate optimization, it is related to load regulation technical field, the method includes: obtaining garbage incineration power plant source side and historical operation data of load side, according to this respectively generates garbage calorific value prediction sequence and power grid load demand prediction sequence;Determine the prediction error distribution characteristics of two kinds of prediction sequences, construct the fuzzy set of source load uncertainty;Combining the fuzzy set, power plant safety environmental protection constraint and economic operation target, according to preset rule distribution robust optimization scheduling model is constructed and the optimal scheduling strategy is solved;Finally, according to optimal scheduling strategy generation unit load regulation instruction, control power plant auxiliary machine equipment operation.The application solves the problem that safety, environmental protection and economy cannot be considered in the load regulation mode of prior art.

Description

Technical Field

[0001] This invention relates to the field of load regulation technology, and in particular to a method and system for dynamic load regulation to optimize the power consumption rate of a waste incineration power plant. Background Technology

[0002] Waste-to-energy incineration is an important approach to urban solid waste treatment and energy recovery. Its core process typically includes waste receiving and storage, combustion in an incinerator boiler, heat exchange in a waste heat boiler, turbine operation, and grid connection of the generator. During this process, the power plant's operating conditions are influenced by a combination of factors, including fuel characteristics, equipment status, and grid demand, exhibiting significant nonlinearity, strong coupling, and large time lag characteristics.

[0003] Traditional load regulation strategies for waste-to-energy incineration plants are mostly based on steady-state assumptions or simplified linear models, aiming to maintain stable unit operation and basic power output by keeping key parameters such as main steam pressure and furnace temperature within set ranges. Such methods can meet basic production needs in relatively stable operating environments. However, against the backdrop of current power market reforms and the increasingly urgent need for deep grid peak shaving, waste-to-energy incineration plants face unprecedented challenges from both source and load uncertainties.

[0004] From the perspective of the "source," the fuel for waste-to-energy incineration is municipal solid waste with complex composition and drastically fluctuating calorific value. Influenced by factors such as region, season, climate, and residents' living habits, the physicochemical properties of the waste entering the incinerator, including moisture content, calorific value, and composition, exhibit significant time-varying and randomness. This uncertainty in fuel characteristics directly leads to drastic fluctuations in the combustion efficiency, furnace temperature distribution, and flue gas emission characteristics of the incinerator, making it difficult to accurately predict and control the boiler's heat load, thereby affecting the stability of the entire energy conversion system.

[0005] From the load side, with the large-scale grid connection of renewable energy and the continuous widening of the peak-valley difference in electricity load, the power grid places higher demands on the flexibility of power generation. Waste-to-energy incineration plants, as an important component of peak-shaving power sources, need to frequently respond to grid dispatch commands and participate in load tracking and frequency regulation. This uncertainty in external electricity demand requires units to have rapid and accurate load response capabilities. However, traditional control strategies, due to problems such as model simplification and response lag, often struggle to meet grid dispatch requirements while simultaneously ensuring the internal thermal balance and dynamic stability of the units.

[0006] More importantly, the limitations of existing scheduling models become increasingly apparent under the combined effects of uncertainties in both the "source" and "load" factors. On the one hand, to cope with fluctuations in operating conditions, operators typically need to reserve a large safety margin, leading to long-term deviations of the unit from the optimal economic operating range, resulting in energy waste and increased plant power consumption. On the other hand, crude adjustment methods are prone to risks such as unstable combustion, excessive steam parameters, and excessive pollutant emissions, which can even threaten unit safety in severe cases. Therefore, how to construct a dynamic load adjustment method that can effectively cope with the uncertainties of the "source-load" factor while balancing safety, environmental protection, and economy has become a critical technical bottleneck that urgently needs to be overcome in the current waste-to-energy incineration field. Summary of the Invention

[0007] In view of this, the purpose of the present invention is to provide a method and system for dynamic load adjustment to optimize the power consumption of waste incineration power plants, aiming to solve the problem that existing dynamic load adjustment methods cannot simultaneously take into account safety, environmental protection and economy.

[0008] This invention proposes a dynamic load adjustment method for optimizing the power consumption rate of waste incineration power plants, the method comprising: Obtain historical operating data of the waste incineration power plant from both the source and load sides, and generate waste calorific value prediction sequences and grid load demand prediction sequences based on the historical operating data of the source and load sides, respectively. Determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics; According to the preset rules, based on the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and the economic operation objectives, a sub-Bruker optimal scheduling model is constructed, and the optimal scheduling strategy is obtained by solving the sub-Bruker optimal scheduling model. The corresponding unit load adjustment command is generated according to the optimal scheduling strategy, and the operation of the auxiliary equipment of the waste incineration power plant is controlled according to the unit load adjustment command.

[0009] Furthermore, in the aforementioned method for optimizing the power consumption rate of waste incineration power plants, the step of acquiring historical operating data of the waste incineration power plant's source and load sides, and generating a waste calorific value prediction sequence and a power grid load demand prediction sequence based on the historical operating data of the source and load sides respectively, includes: The source-side data and load-side data of the waste incineration power plant are collected from the historical operation database. The source-side data includes the composition detection data, physical characteristic parameters and combustion response data of the waste entering the furnace. The load-side data includes historical load instructions issued by the power grid dispatch center and actual power feedback data of the units in the plant. Spatiotemporal alignment and feature engineering processing of the source-side data specifically includes: converting the elemental analysis results in the component detection data into feature vectors with fixed carbon content and volatile matter ratios, using the moisture content in the physical property parameters as key input features, and using the furnace temperature and flue gas oxygen content in the combustion response data as auxiliary correction features, to jointly construct a source-side multimodal input feature set; The source-side multimodal input feature set is input into a pre-trained source-side calorific value prediction sub-model based on gated recurrent units, and the source-side calorific value prediction sub-model outputs a waste calorific value prediction sequence within a future preset time window. The deviation analysis and trend decomposition of the load-side data specifically include: calculating the tracking deviation sequence between the historical load command and the actual power feedback data, using the tracking deviation sequence to weight the historical load command with confidence, and constructing a load-side load prediction sub-model by combining the seasonal autoregressive integral moving average model. The load-side load prediction sub-model outputs a predicted sequence of grid load demand within a preset time window.

[0010] Furthermore, in the above-mentioned method for optimizing the power consumption rate of waste incineration power plants, the step of determining the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and constructing a fuzzy set to describe the uncertainty of the source load based on the prediction error distribution characteristics, includes: Obtain the first residual set between the predicted calorific value of the waste and the actual measured value, and the second residual set between the predicted power grid load demand and the actual dispatch value; The first residual set and the second residual set are fitted with probability density functions using the kernel density estimation method to obtain the source-side error probability distribution and the load-side error probability distribution. The corresponding higher-order statistical features are extracted based on the source-side error probability distribution and the load-side error probability distribution. The higher-order statistical features include skewness and kurtosis. Based on the source-side error probability distribution and the load-side error probability distribution, a joint probability distribution model of the source-side and load-side prediction errors is constructed using the Copula function to capture the nonlinear correlation between waste calorific value fluctuations and power grid dispatching requirements. Based on the moment characteristics of the joint probability distribution model, a fuzzy set containing all probability distributions that satisfy the moment characteristics is constructed. The fuzzy set is used to simultaneously consider the coupling effect of the uncertainty on both sides of the source and load and the local morphological characteristics of the distribution function in the split-bar optimization.

[0011] Furthermore, in the aforementioned method for optimizing the power consumption rate of waste incineration power plants, the step of constructing a sub-Browser optimization scheduling model according to preset rules based on the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and economic operation objectives includes: Confidence intervals are obtained for the fuzzy centralized waste calorific value prediction error and the power grid load demand prediction error, respectively. Based on the confidence intervals, the uncertainty constraints of the sub-Bruker optimal scheduling model are determined. Based on the aforementioned uncertainty constraints, and in conjunction with the boundary conditions of boiler combustion temperature, main steam pressure, and flue gas pollutant emission limits, the safety and environmental constraints of the sub-Bluer bar optimization scheduling model are determined. Based on the aforementioned safety and environmental constraints and the function used to characterize fuel cost and efficiency loss, an objective function is constructed with the goal of minimizing the total operating cost under the worst-case scenario. The sub-Blu-ray optimization scheduling model is then determined based on the objective function and the uncertainty constraints.

[0012] Furthermore, in the above-mentioned method for dynamic load adjustment to optimize the power consumption of waste incineration power plants, the step of obtaining the optimal scheduling strategy based on the distributed bar optimization scheduling model includes: The aforementioned split-bar optimization scheduling model is equivalently transformed to include fuzzy set constraints. The bi-level optimization problem is transformed into a solvable linearized convex optimization model; An adaptive hybrid solution algorithm is used to efficiently and accurately solve the transformed linearized convex optimization model and obtain candidate scheduling strategies. The candidate scheduling strategies are verified and robustness checked in multiple scenarios to select the final optimal scheduling strategy that meets safety and environmental protection constraints and achieves the optimal plant power consumption rate and total operating cost.

[0013] Furthermore, in the aforementioned method for optimizing the power consumption rate of waste incineration power plants, the step of performing an equivalent transformation on the sub-Bruker optimization scheduling model, converting the min-max bilevel optimization problem with fuzzy set constraints into a solvable linearized convex optimization model, includes: Based on the joint probability distribution model of source-load prediction error, the moment feature boundary of fuzzy set is extracted, and Lagrange dual variables are introduced to transform the maximization constraint in the worst scenario of the sub-Bruker optimization scheduling model into a dual constraint term, thereby eliminating the nested structure of the two-layer optimization. For nonlinear terms in safety and environmental protection constraints, a piecewise linearization method is used for high-precision fitting. An error correction factor is introduced during the fitting process to ensure that the linearization error is controlled within a preset threshold. In light of the start-up and shutdown characteristics of auxiliary equipment in waste incineration power plants, a relaxation term for equipment state switching constraints is added, and the complementary constraint transformation method is used to transform discrete equipment state variables into continuous variables. By integrating linear constraints, dual constraints, and continuous variable constraints, the objective function of the sub-Brussels bar optimization scheduling model is simultaneously retained and linearized for adaptation, ultimately forming a solvable linearized convex optimization model.

[0014] Furthermore, the above-mentioned method for dynamic load adjustment of waste incineration power plant power utilization optimization, wherein the step of employing an adaptive hybrid solution algorithm to efficiently and accurately solve the transformed linearized convex optimization model to obtain candidate scheduling strategies includes: Initialize the solution parameters, determine the initial search range of the optimization variables based on the source-side multimodal input feature set and the load-side load prediction sequence, and use the particle swarm optimization algorithm to perform a global search on the initial solution; The optimal initial solution obtained from the global search is used as the initial iteration point for the precise solution. The interior point method is used to perform local refinement of the transformed convex optimization model. During the solution process, the obstacle parameters are dynamically adjusted to balance the solution speed and accuracy. Set an iteration termination condition, combining the model convergence accuracy and the number of iterations as dual constraints. When the optimization variables satisfy all constraints, stop solving and output the current optimization solution as a candidate scheduling strategy.

[0015] Another object of the present invention is to provide a load dynamic adjustment system for optimizing the power consumption rate of a waste incineration power plant, the system comprising: The acquisition module is used to acquire historical operating data of the waste incineration power plant on the source side and load side, and generate waste calorific value prediction sequence and power grid load demand prediction sequence based on the historical operating data of the source side and load side, respectively. The determination module is used to determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and to construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics; The solution module is used to construct a sub-Brussels bar optimization scheduling model according to preset rules based on the fuzzy set, the safety and environmental protection constraints of the waste incineration power plant, and the economic operation objectives, and to solve the optimal scheduling strategy based on the sub-Brussels bar optimization scheduling model. The control module is used to generate corresponding unit load adjustment instructions according to the optimal scheduling strategy, and to control the operation of auxiliary equipment of the waste incineration power plant according to the unit load adjustment instructions.

[0016] Another object of the present invention is to provide a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0017] Another object of the present invention is to provide an electronic device including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of the method described above.

[0018] This invention acquires historical operating data from both the source and load sides of a waste-to-energy power plant to generate waste calorific value prediction sequences and grid load demand prediction sequences, respectively. It determines the prediction error distribution characteristics of these two types of sequences and constructs a fuzzy set to describe source-load uncertainties. Based on preset rules, combined with the fuzzy set, safety and environmental constraints of the waste-to-energy power plant, and economic operation objectives, it constructs a bibliometric optimal scheduling model and solves it to obtain the optimal scheduling strategy. Based on the optimal scheduling strategy, it generates unit load adjustment commands to control the operation of auxiliary equipment. Through quantitative modeling of source-load uncertainties, bibliometric optimal scheduling, and precise dynamic load adjustment, it effectively reduces safety margin redundancy, improves the economic operation level of the units, and ensures stable combustion and compliance with steam parameters and pollutant emissions standards. This invention solves the technical problems of traditional waste-to-energy power plant load adjustment methods, which struggle to cope with dual source-load uncertainties, have excessive safety margins leading to high plant power consumption, and whose extensive adjustments easily cause operational and environmental risks. Attached Figure Description

[0019] Figure 1 This is a flowchart of the load dynamic adjustment method for optimizing the power consumption rate of a waste incineration power plant in the first embodiment of the present invention; Figure 2 This is a structural block diagram of the load dynamic adjustment system for optimizing the power consumption rate of a waste incineration power plant, as shown in the third embodiment of the present invention.

[0020] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0021] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0022] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0024] Example 1 Please see Figure 1 The figure shows a method for dynamic load adjustment to optimize the power consumption of a waste incineration power plant in the first embodiment of the present invention, the method including steps S10 to S13.

[0025] Step S10: Obtain historical operating data of the waste incineration power plant on the source side and load side, and generate waste calorific value prediction sequence and power grid load demand prediction sequence based on the historical operating data of the source side and load side, respectively.

[0026] The core of this invention is to accurately predict and quantify the uncertainties of the calorific value of waste on the source side and the load demand of the power grid on the load side, construct a robust optimization scheduling model, and solve it. This ultimately achieves dynamic adjustment of auxiliary equipment, thereby optimizing the plant's power consumption rate while balancing safety, environmental protection, and economy. The specific implementation process is as follows: Among them, the historical operating data on the source side mainly comes from the power plant's real-time monitoring system and historical database, covering various parameters related to waste entering the furnace and thermal parameters during the incineration process; the historical operating data on the load side mainly comes from the historical dispatch records of the power grid dispatch center and the power feedback data of the power plant units.

[0027] When acquiring data, it is necessary to ensure the integrity and timeliness of the data, remove outliers (such as jump data caused by sensor failure, records with missing data exceeding the preset threshold), and use linear interpolation to supplement missing data to ensure data continuity.

[0028] Based on the supplemented and complete historical operating data, the system uses a preset prediction algorithm to predict the calorific value of waste and the grid load demand in the future, and outputs the corresponding prediction sequence. The time interval of the prediction sequence can be set according to the actual operating needs of the power plant, usually 15 minutes, 30 minutes or 1 hour. The prediction duration (preset time window in the future) can be set to 24 hours to meet the daily dispatching needs of the power plant.

[0029] For example, source-side data specifically includes compositional analysis data of the waste fed into the incinerator, physical property parameters, and combustion response data of the incinerator. Compositional analysis data of the waste fed into the incinerator can be obtained through the power plant's waste composition analysis system, including the content of elements such as carbon, hydrogen, oxygen, nitrogen, and sulfur in the waste, as well as the proportions of ash, volatile matter, and fixed carbon. Physical property parameters mainly include the moisture content, density, and particle size distribution of the waste fed into the incinerator. Combustion response data of the incinerator can be collected in real time by sensors within the furnace, including parameters such as furnace temperature, flue gas oxygen content, and furnace pressure.

[0030] Load-side data reflects various types of data on the load demand of the power grid. Specifically, it includes historical load commands issued by the power grid dispatch center and actual power feedback data from generating units within the plant. Historical load commands can be obtained from the historical records of the power grid dispatch system, while actual power feedback data can be collected through the power sensors of the generating units, covering parameters such as active power and reactive power. During the data collection process, it is necessary to ensure that the timestamps of the data are consistent to facilitate subsequent data processing and analysis. The data collection cycle should be consistent with the real-time monitoring cycle of the power plant, typically 1-5 minutes.

[0031] Spatiotemporal alignment and feature engineering were performed on the source-side data. Spatiotemporal alignment refers to calibrating source-side data from different collection times and locations with a unified timestamp to ensure data synchronization and avoid feature distortion caused by time deviations. For furnace temperature, flue gas oxygen content, and other data collected from different spatial locations, a weighted average method was used for fusion to obtain the overall average furnace temperature and average flue gas oxygen content, ensuring spatial consistency of the data. The core of feature engineering is to extract key features that reflect changes in the calorific value of waste. Specifically, this includes converting the elemental analysis results in the component detection data into a feature vector with a fixed carbon content and volatile matter ratio. Assuming the elemental analysis yields carbon content C, hydrogen content H, oxygen content O, nitrogen content N, sulfur content S, ash content A, and moisture content M, the formula for calculating the fixed carbon content FC is as follows: Where V is the volatile matter content and the volatile matter ratio ,Will and As the core component of the feature vector, the moisture content among the physical property parameters is used as the key input feature because the moisture content directly affects the combustion efficiency and calorific value of waste, and is an important influencing factor in the prediction of waste calorific value. The furnace temperature T and flue gas oxygen content in the combustion response data are used as auxiliary correction features because the furnace temperature and flue gas oxygen content can indirectly reflect the combustion state of waste and have a significant correlation with the calorific value of waste, and can be used to correct the deviation of the prediction results.

[0032] Through the above processing, a source-side multimodal input feature set is jointly constructed. Each feature is normalized using the following formula: , These are the original eigenvalues. This is the historical minimum value of this feature. The historical maximum value of this feature is used. After normalization, the feature value is in the range of [0,1], thus avoiding the influence of different units on the prediction model.

[0033] Next, the source-side multimodal input feature set is fed into a pre-trained source-side calorific value prediction sub-model based on gated recurrent units (GRUs). The source-side calorific value prediction sub-model then outputs a predicted sequence of waste calorific values ​​within a preset future time window. The gated recurrent unit (GRU) is an improved recurrent neural network that can effectively capture long-term dependencies in time-series data, making it suitable for time-series prediction of waste calorific values.

[0034] For example, the training process of the source-side calorific value prediction sub-model is as follows: The source-side multimodal input feature set from the past 12 months was selected as training data, and the corresponding actual garbage heat values ​​were used as label data. The model's input layer dimension was 5 (corresponding to the 5 features in the feature set), with 2 hidden layers and 64 neurons per layer, and the output layer dimension was 1 (corresponding to the predicted garbage heat values). The Adam optimizer was used for model training, with a learning rate of 0.001, 1000 iterations, and mean squared error (MSE) as the loss function. The loss function formula is as follows: Where N is the number of training samples, For the first The actual calorific value of each sample For the first The predicted heat value of each sample is obtained. During the training process, an early stopping strategy is adopted. When the loss function of the validation set no longer decreases after 10 consecutive iterations, the training is stopped, and the source-side heat value prediction sub-model is obtained after training.

[0035] The processed source-side multimodal input feature set is input into the model. A preset future time window is set to 24 hours, and the time interval is 30 minutes. The model outputs the predicted calorific value of waste every 30 minutes within the next 24 hours, forming a waste calorific value prediction sequence. ,in For the first Predicted calorific value of waste over a time interval.

[0036] Furthermore, deviation analysis and trend decomposition are performed on the load-side data. The core of deviation analysis is to calculate the tracking deviation between historical load commands and actual power feedback data, quantifying the tracking accuracy of power plant units for grid load commands. The specific calculation process is as follows: Let the historical load command sequence be... The corresponding actual power feedback sequence is ,in The number of historical data samples, the first Tracking bias of individual samples Thus, the tracking deviation sequence is obtained. The historical load commands are weighted by confidence level using the tracking deviation sequence. The calculation formula is: The smaller the deviation, the greater the weight, indicating that the historical load instruction has higher reference value; the weighted historical load instruction sequence is as follows: .

[0037] Trend decomposition refers to decomposing the weighted historical load command sequence to extract the trend, seasonal, and stochastic components. A seasonal autoregressive integral moving average (SARIMA) model is then used to construct a load-side load forecasting sub-model. The general form of the SARIMA model is as follows: ,in Let the order be the autoregressive order. Let be the difference order. The moving average order is... For seasonal autoregression order, For seasonal difference order, The order of the seasonal moving average. It is a seasonal cycle.

[0038] Based on the seasonal characteristics of the power grid load, set =24 (day cycle), the model parameters are determined using the AIC criterion (Akaike Information Criterion). The AIC criterion formula is as follows: ,in The number of model parameters, The likelihood function value of the model is used to select the parameter combination with the smallest AIC value as the final model parameters. Typically, this is set to... =2、 =1、 , 1.

[0039] Finally, the load-side load forecasting sub-model outputs a power grid load demand forecast sequence for the future preset time window. The weighted historical load command sequence is used as training data to train the SARIMA model. After training, the time characteristics (such as hour, date, season, etc.) of the future preset time window are input, and the model outputs the power grid load demand forecast value every 30 minutes for the next 24 hours, forming the power grid load demand forecast sequence. ,in For the first The predicted grid load demand (in MW) for each time interval is consistent with the time interval of the waste calorific value prediction sequence, which facilitates subsequent collaborative optimization.

[0040] Step S11: Determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics.

[0041] Because the calorific value of waste is affected by various factors and the grid load demand fluctuates randomly, there will inevitably be an error between the predicted sequence and the actual value. Therefore, it is necessary to first calculate the prediction error, that is, the difference between the predicted value and the actual measured value (or the actual dispatch value), forming an error set. Statistical analysis methods are used to uncover the distribution pattern of the error set, i.e., the error distribution characteristics, including statistical quantities such as the mean, variance, skewness, and kurtosis of the errors, as well as the probability density distribution of the errors. Based on these error distribution characteristics, a fuzzy set is constructed. This fuzzy set can cover all possible error distributions, thereby accurately describing the uncertainty of fluctuations in the calorific value of waste on the source side and the fluctuations in grid load demand on the load side. This provides a foundation for subsequent robust optimization scheduling and avoids insufficient robustness of the scheduling strategy due to a single error assumption.

[0042] For example, a first residual set between the predicted calorific value of waste and the actual measured value, and a second residual set between the predicted power grid load demand and the actual dispatch value are obtained. The first residual set is a set of errors in the waste calorific value prediction, specifically calculated as follows: Let the waste calorific value prediction sequence be... The corresponding actual measurement value sequence is ,in To predict the length of the sequence, the first... The residuals (prediction error) at each prediction point Thus, the first residual set is obtained. ; The second residual set is the set of errors in the power grid load demand forecast. The specific calculation method is as follows: Let the power grid load demand forecast sequence be... The corresponding actual scheduling value sequence is , No. The residuals at each prediction point Thus, the second residual set is obtained. .

[0043] When obtaining the residual set, it is necessary to ensure that the timestamps of the actual measured values, actual scheduling values ​​and predicted sequences correspond one-to-one to avoid residual calculation errors caused by time deviations.

[0044] Next, the kernel density estimation method is used to fit the probability density function to the first and second residual sets respectively, obtaining the source-side error probability distribution and the load-side error probability distribution. Then, the corresponding higher-order statistical features are extracted based on these distributions. Kernel density estimation (KDE) is a non-parametric probability density estimation method that does not require assumptions about the specific form of the error distribution, and can more accurately fit the actual distribution of the error. Its core formula is: ,in Let be the probability density function. Let be the number of samples in the residual set. For bandwidth (smoothing parameter). For the first in the residual set One sample, This is the kernel function.

[0045] In this embodiment, the kernel function is selected as the Gaussian kernel function; bandwidth The Silverman rule is used to determine the source-side error probability distribution. Kernel density estimation is performed on the first and second residual sets respectively to obtain the source-side error probability distribution. Load side error probability distribution .

[0046] Higher-order statistical features include skewness and kurtosis. Skewness describes the degree of asymmetry in a probability distribution, while kurtosis describes the steepness of the probability distribution. Source-side error skewness... Source-side error kurtosis , load side error skewness kurtosis of load side error The calculation formulas are as follows:

[0047]

[0048]

[0049] , in Let be the mean and standard deviation of the first residual set, respectively. These represent the mean and standard deviation of the second residual set, respectively. Higher-order statistical features can further quantify the morphological characteristics of the error distribution, providing a basis for subsequently constructing a joint probability distribution model.

[0050] Subsequently, based on the source-side error probability distribution and the load-side error probability distribution, a joint probability distribution model of the source-side and load-side prediction errors is constructed using the Copula function to capture the nonlinear correlation between waste calorific value fluctuations and grid dispatching demands. The Copula function can connect the marginal distributions of multiple random variables to construct a joint distribution without assuming that the random variables are independent of each other, making it suitable for describing the nonlinear correlation between source-side and load-side errors.

[0051] In this embodiment, the t-Copula function is selected to construct the joint probability distribution model. The t-Copula function has strong tail correlation, which can better capture the cooperative fluctuation characteristics of source-load errors under extreme scenarios. The joint probability density formula of the t-Copula function is: ,in Let be the density function of the t-Copula function. The marginal distribution function of the source-side error (obtained from kernel density estimation) is given. The marginal distribution function of the load-side error (obtained by kernel density estimation) is given. These are the degrees of freedom parameters. The maximum likelihood estimation method is used to determine which option maximizes the likelihood function value. As the optimal parameter, usually The value range is 2-10. This joint probability distribution model can accurately describe the joint fluctuation law of source-side error and load-side error, capture the nonlinear correlation between the two, and provide support for subsequent uncertainty quantification.

[0052] Finally, based on the moment features of the joint probability distribution model, a fuzzy set containing all probability distributions satisfying these moment features is constructed. This fuzzy set is used to simultaneously consider the coupling effect of uncertainties on both the source and load sides and the local morphological characteristics of the distribution function in the bibru bar optimization. Moment features are the core features describing the probability distribution. In this embodiment, the first four moments (mean, variance, skewness, and kurtosis) are selected as the moment features of the joint probability distribution model. Let the joint probability distribution of the source-side error and the load-side error be P, and its first four moments are: first moment (mean vector). Second moment (covariance matrix) ,in The third moment is the covariance of the source-side error and the load-side error. Fourth-order moment The principle for constructing fuzzy sets is to include all probability distributions that satisfy the above moment feature constraints, i.e., fuzzy sets. The first four moments satisfy the set conditions, which are:

[0053] in The fluctuation thresholds for the mean, covariance matrix, third moment, and fourth moment are used to control the range of the fuzzy set. These thresholds are set based on the degree of uncertainty in the actual operation of the power plant, typically taking 5%-10% of the corresponding moment characteristics. This ensures that the fuzzy set can cover all possible error distributions without making the model overly conservative due to an excessively large range. This fuzzy set can simultaneously consider the coupling effect of uncertainties on both the source and load sides (reflected by the covariance matrix) and the local morphological characteristics of the distribution function (reflected by skewness and kurtosis), providing an uncertainty constraint basis for the subsequent construction of a distributed bar optimization scheduling model.

[0054] Step S12: Construct a sub-Bruker optimal scheduling model according to preset rules based on the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and the economic operation objectives. Solve the sub-Bruker optimal scheduling model to obtain the optimal scheduling strategy.

[0055] Then, based on preset rules, a sub-Bruker optimal scheduling model is constructed according to fuzzy sets, safety and environmental constraints of the waste-to-energy power plant, and economic operation objectives. The optimal scheduling strategy is obtained by solving the sub-Bruker optimal scheduling model. The preset rules mainly include the priority of constraints and the weight allocation of the objective function. Among them, safety and environmental constraints are the primary constraints, which must prioritize meeting requirements such as stable boiler combustion, compliant steam parameters, and pollutant emissions within limits. The economic operation objective mainly focuses on minimizing the plant's power consumption rate and reducing total operating costs. When constructing the model, the uncertainties described by fuzzy sets are incorporated into the constraints to ensure that the model can meet safety and environmental requirements and achieve economic operation objectives under various possible source-load fluctuation scenarios. The model is solved using a suitable solution algorithm to obtain the optimal scheduling strategy. This strategy covers key information such as the operating parameters of each auxiliary equipment of the unit and the load allocation ratio, providing a specific basis for subsequent load regulation.

[0056] For example, firstly, the confidence intervals for the fuzzy set of waste calorific value prediction error and power grid load demand prediction error are obtained respectively. Based on these confidence intervals, the uncertainty constraints of the sub-Bruker optimal scheduling model are determined. The confidence intervals are used to quantify the fluctuation range of the prediction error. Based on the probability distribution of the fuzzy set, a 95% confidence level is selected (which can be adjusted to 90% or 99% according to power plant safety requirements), and the confidence intervals for the waste calorific value prediction error and power grid load demand prediction error are calculated. For the waste calorific value prediction error... Its 95% confidence interval is Where 1.96 is the quantile corresponding to the 95% confidence level under the standard normal distribution; for the power grid load demand forecasting error Its 95% confidence interval is The core of the uncertainty constraint is to ensure that the unit operating parameters can meet subsequent safety and environmental protection constraints under all possible error scenarios. The specific expression is: ,in ,in ,in This represents the actual calorific value of the waste. For the actual load demand of the power grid, This is the predicted calorific value of waste. This represents the predicted load demand value for the power grid. By incorporating the uncertainty of the source and load into the model through this uncertainty constraint, the robustness of the model is ensured.

[0057] Secondly, based on the aforementioned uncertainty constraints, and combined with the boundary conditions of boiler combustion temperature, main steam pressure, and flue gas pollutant emission limits, the safety and environmental constraints of the sub-Blu-ray bar optimization scheduling model are determined. These safety and environmental constraints are the bottom line for unit operation and must be prioritized. They specifically include three categories: boiler combustion constraints, steam parameter constraints, and pollutant emission constraints. Boiler combustion constraints mainly target furnace temperature to ensure stable combustion and prevent flameout or coking. The constraint expression is as follows: ,in This is a function for calculating furnace temperature. These are the operating parameters of auxiliary equipment (such as air volume and feed rate). This is the minimum safe temperature for the furnace (usually set to 850°C). The maximum safe temperature of the furnace (usually set to 1100℃); furnace temperature calculation function. An empirical formula can be used: ,in These are empirical coefficients, obtained by fitting historical operating data of the power plant. For air supply volume, This refers to the feed rate. Steam parameter constraints primarily target the main steam pressure to ensure stable steam parameters and meet power generation requirements. The constraint expression is: ,in The main steam pressure calculation function. The minimum safe pressure for main steam (usually set to 9.8 MPa). The maximum safe pressure for main steam (usually set to 13.7 MPa); main steam pressure calculation function. An empirical formula can be used: ,in This is an empirical coefficient. For the power of the water pump, This refers to the opening degree of the steam turbine inlet valve. Pollutant emission constraints mainly target particulate matter, SO2, and NO in the flue gas. x For pollutants such as [list of pollutants], the constraint expression is: ; in, Particulate matter, SO2, and NO, respectively. x emission concentration, These are the corresponding national emission standard limits. All of the above constraints must be combined with uncertainty constraints to ensure that all parameters meet the boundary requirements within the error fluctuation range.

[0058] Subsequently, based on the aforementioned safety and environmental constraints and the function used to characterize fuel cost and efficiency loss, an objective function is constructed to minimize the total operating cost under the worst-case scenario. The sub-Bluerg optimization scheduling model is then determined based on the objective function and uncertainty constraints. The core of sub-Bluerg optimization is to minimize the total operating cost under the worst-case scenario, ensuring that the model can achieve economical operation under various possible source-load fluctuation scenarios.

[0059] Total operating costs mainly consist of three parts: fuel costs, auxiliary equipment energy costs, and efficiency loss costs. The functions used to characterize fuel costs and efficiency losses are: Fuel Cost Function ,in Price of waste fuel (unit: yuan / kJ) For feed rate; efficiency loss cost function ,in Efficiency loss unit price (unit: yuan / kWh) The rated power generation efficiency of the unit. The actual power generation efficiency of the unit is calculated using the following function: ,in , This is an empirical coefficient. Total energy consumption of auxiliary equipment. Energy cost function for auxiliary equipment. ,in For the first Energy consumption of auxiliary equipment (unit: kWh) For the first The electricity price for auxiliary equipment (unit: yuan / kWh). The objective function is expressed as follows: ,in To optimize variables (auxiliary equipment operating parameters), Let be any probability distribution within the fuzzy set. Combining the above objective function, uncertainty constraints, and safety and environmental constraints, the sub-Bruker optimal scheduling model can be determined. The mathematical expression of this model is: ; ; ; ; ; ; ( (Number of auxiliary equipment) in The first Minimum and maximum operating parameter limits for each auxiliary machine are set to ensure that the auxiliary machine operates within a safe range.

[0060] Step S13: Generate corresponding unit load adjustment instructions according to the optimal scheduling strategy, and control the operation of auxiliary equipment of the waste incineration power plant according to the unit load adjustment instructions.

[0061] Specifically, once the optimal scheduling strategy is determined, it is translated into executable load adjustment commands. These commands include the operating frequency and power setpoints of various auxiliary equipment (such as induced draft fans, forced draft fans, feedwater pumps, circulating water pumps, etc.). Through the power plant's control system, these commands are sent to the actuators of each auxiliary equipment, controlling their operating status in real time. This allows the unit load to be dynamically adjusted according to the optimal scheduling strategy, ensuring that the unit operates within its optimal economic range while meeting grid load demands and complying with safety and environmental standards, thereby optimizing the plant's power consumption rate.

[0062] In summary, the load dynamic adjustment method for optimizing the plant power consumption rate of waste incineration power plants in the above embodiments of the present invention obtains historical operating data from both the source and load sides of the waste incineration power plant to generate waste calorific value prediction sequences and grid load demand prediction sequences, respectively. It determines the prediction error distribution characteristics of the two types of prediction sequences and constructs a fuzzy set to describe the uncertainty of the source and load. Based on preset rules combined with the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and economic operation objectives, it constructs a bibliometric optimization scheduling model and solves it to obtain the optimal scheduling strategy. Based on the optimal scheduling strategy, it generates unit load adjustment instructions to control the operation of auxiliary equipment. Through quantitative modeling of source and load uncertainty, bibliometric optimization scheduling, and precise load dynamic adjustment, it effectively reduces safety margin redundancy, improves the economic operation level of the unit, and ensures stable combustion and compliance with steam parameters and pollutant emissions standards. This solves the technical problems of traditional waste incineration power plant load adjustment methods, which struggle to cope with dual uncertainties of source and load, have excessive safety margins leading to high plant power consumption rates, and whose extensive adjustments easily cause operational and environmental risks.

[0063] Example 2 This embodiment also proposes a dynamic load adjustment method for optimizing the power consumption rate of a waste incineration power plant. The difference between the dynamic load adjustment method for optimizing the power consumption rate of a waste incineration power plant in this embodiment and the dynamic load adjustment method for optimizing the power consumption rate of a waste incineration power plant in Embodiment 1 is as follows: The steps for obtaining the optimal scheduling strategy based on the bibliometric optimization scheduling model include: The aforementioned split-bar optimization scheduling model is equivalently transformed to include fuzzy set constraints. The bi-level optimization problem is transformed into a solvable linearized convex optimization model; An adaptive hybrid solution algorithm is used to efficiently and accurately solve the transformed linearized convex optimization model and obtain candidate scheduling strategies. The candidate scheduling strategies are verified and robustness checked in multiple scenarios to select the final optimal scheduling strategy that meets safety and environmental protection constraints and achieves the optimal plant power consumption rate and total operating cost.

[0064] The split-bar optimization scheduling model is essentially a min-max bi-level optimization problem. The outer level minimizes the total operating cost, while the inner level maximizes the cost under the worst-case scenario. Directly solving this problem is difficult, so it needs to be transformed into a linearized convex optimization model through an equivalent transformation to reduce the difficulty while ensuring accuracy. The specific transformation process includes three steps: dual transformation, linearization of nonlinear terms, and continuousization of discrete variables. The detailed process will be further explained in Example 6. This example focuses on the characteristics of the transformed linearized convex optimization model: the transformed model is a convex optimization model, the objective function is a linear function, all constraints are linear constraints, and all optimization variables are continuous variables. It can be efficiently solved using conventional convex optimization algorithms, ensuring the global optimality of the solution.

[0065] Next, an adaptive hybrid solution algorithm is employed to efficiently and accurately solve the transformed linearized convex optimization model, yielding candidate scheduling strategies. This algorithm combines the advantages of global search and local refinement algorithms, avoiding getting trapped in local optima while ensuring both accuracy and efficiency. The specific implementation process will be further explained in Example 7. During the solution process, the operating parameters of auxiliary equipment (air volume, feed rate, feedwater pump power, circulating water pump power, etc.) are used as optimization variables. Through iterative optimization, values ​​that satisfy all constraints are obtained. These optimization variable values ​​are then compiled into candidate scheduling strategies. These strategies must clearly define key information such as the operating parameters of each auxiliary equipment and the unit load allocation ratio within each time interval to ensure direct applicability for subsequent load regulation.

[0066] Finally, the candidate scheduling strategies are validated and robustness verified in multiple scenarios to select the optimal scheduling strategy that meets safety and environmental constraints and achieves the best plant power consumption and total operating cost. Multi-scenario validation involves selecting typical scenarios from the fuzzy set (including baseline, extreme, and random scenarios), applying the candidate scheduling strategy to each scenario, simulating the unit's operation, and verifying the adaptability of the candidate scheduling strategy under different scenarios. The baseline scenario is the scenario where the prediction error is the mean; the extreme scenario is the scenario where the prediction error reaches the confidence interval boundary; and the random scenario is a random error scenario generated through Monte Carlo simulation (generating 1000 random error samples, covering all possible distributions of the fuzzy set). Robustness verification involves calculating whether the unit's operating parameters (furnace temperature, main steam pressure, and pollutant emission concentration) meet safety and environmental constraints in each scenario, and whether the plant power consumption and total operating cost are optimal. The selection criteria are: the candidate scheduling strategy meets safety and environmental constraints in all validation scenarios, has the lowest plant power consumption and the lowest total operating cost in the baseline scenario, and the fluctuation of plant power consumption and total operating cost in the extreme scenario does not exceed 5%. Through multi-scenario verification and robustness check, candidate scheduling strategies that do not meet the constraints or lack robustness are eliminated, and the optimal scheduling strategy is retained as the final optimal scheduling strategy.

[0067] For example, the sub-Bruker optimization scheduling model can be equivalently transformed to include fuzzy set constraints. The specific implementation process of transforming the bi-level optimization problem into a solvable linearized convex optimization model is as follows: First, based on the joint probability distribution model of source-load prediction errors, the moment feature boundary of the fuzzy set is extracted. Lagrange dual variables are introduced to transform the maximization constraint in the worst-case scenario of the distributed bar optimization scheduling model into dual constraint terms, eliminating the nested structure of the two-layer optimization. The moment feature boundary of the fuzzy set, i.e., the fluctuation range of the first four moments, is used to introduce Lagrange dual variables λ1 (corresponding to the mean constraint), λ2 (corresponding to the covariance constraint), λ3 (corresponding to the third moment constraint), and λ4 (corresponding to the fourth moment constraint). According to Lagrange duality theory, the inner-layer maximization problem is transformed into dual constraint terms and integrated into the outer-layer optimization problem. The original objective function... ( The total operating cost can be converted into: ; Simultaneously add dual constraints: This ensures the effectiveness of the dual transformation. This step eliminates the nested structure of the two-level optimization, transforming the original model into a single-level optimization model and reducing the difficulty of solving the problem.

[0068] Next, for the nonlinear terms in the safety and environmental protection constraints, a piecewise linearization method is used for high-precision fitting. An error correction factor is introduced during the fitting process to ensure that the linearization error is controlled within a preset threshold. The furnace temperature calculation function, main steam pressure calculation function, and pollutant emission concentration calculation function in the safety and environmental protection constraints may all have nonlinear terms. For example, there may be a quadratic nonlinear relationship between furnace temperature and feed rate and air volume, which needs to be transformed into linear terms through a piecewise linearization method.

[0069] Subsequently, considering the start-up and shutdown characteristics of auxiliary equipment in waste incineration power plants, relaxation terms for equipment state switching constraints were added. A complementary constraint transformation method was used to convert discrete equipment state variables into continuous variables. The state (start / stop) of auxiliary equipment is a discrete variable (0 represents stop, 1 represents start), which increases the difficulty of solving the model and needs to be converted into continuous variables. The complementary constraint transformation method was used to introduce continuous variables. Indicates the first The status of each auxiliary device. Indicates startup. Indicates stop; simultaneously adds a relaxation term. This is used to alleviate the rigid constraints of state transitions. The expression for the state transition constraint is: ,in The maximum value is set to The 5% relaxation term ensures that the relaxation term will not affect the normal operation of the equipment. This step transforms the discrete equipment state variables into continuous variables, making the model a linear optimization model with fully continuous variables, further reducing the difficulty of solving the problem.

[0070] Finally, by integrating linear constraints, dual constraints, and continuous variable constraints, the objective function of the bilabial optimization scheduling model is simultaneously retained and linearized, ultimately forming a solvable linearized convex optimization model. All constraints are integrated, including linearized safety and environmental constraints, uncertainty constraints, dual constraints, continuous equipment state constraints, and relaxation constraints. The objective function is linearized; if nonlinear terms exist in the objective function (such as quadratic terms in efficiency loss costs), the same piecewise linearization method as in the second step is used to ensure the objective function is linear. The resulting linearized convex optimization model has a linear objective function, all constraints are linear, and all optimization variables are continuous, satisfying all characteristics of a convex optimization model. It can be efficiently solved using conventional convex optimization algorithms (such as the interior-point method and the simplex method), ensuring the global optimality of the solution.

[0071] For example, an adaptive hybrid solution algorithm is used to efficiently and accurately solve the transformed linearized convex optimization model, and the specific implementation process of the candidate scheduling strategy is as follows: First, the solution parameters are initialized. Based on the source-side multimodal input feature set and the load-side load prediction sequence, the initial search range of the optimization variables is determined, and a particle swarm optimization (PSO) algorithm is used to perform a global search on the initial solution. The solution parameters include the parameters of the PSO algorithm and the parameters of the subsequent interior-point method. The parameters of the PSO algorithm are set as follows: number of particles... Maximum number of iterations The initial value of the inertia weight ω is During the iteration process, it decreases linearly to acceleration coefficient The parameter settings for the interior point method are as follows: initial obstacle parameters. Obstacle parameter decrease coefficient Convergence accuracy .

[0072] The initial search range of the optimization variables is determined based on the source-side multimodal input feature set and the load-side load prediction sequence, such as the air supply volume. The initial search range is ,in Determined based on the minimum value of the waste calorific value prediction sequence, The feed rate is determined based on the maximum value in the waste calorific value prediction sequence. The initial search range is The algorithm adjusts its performance based on the fluctuation range of the power grid load demand forecast sequence. The global search process of the particle swarm optimization algorithm is as follows: Initialize 50 particles, each corresponding to a set of optimization variables; calculate the fitness value (i.e., the objective function value) of each particle; update the individual optimal solution and the global optimal solution for each particle. The individual optimal solution is the solution with the smallest fitness value during the iteration process of that particle, and the global optimal solution is the solution with the smallest fitness value during the iteration process of all particles; iteratively update the velocity and position of the particles; when the number of iterations reaches the maximum number of iterations, stop the global search and output the global optimal solution as the initial iteration point for accurate solution.

[0073] Next, the optimal initial solution obtained from the global search is used as the initial iteration point for the precise solution. The interior-point method is then employed to perform local refinement of the transformed convex optimization model. During the solution process, the obstacle parameters are dynamically adjusted to balance the solution speed and accuracy. The interior-point method is an efficient algorithm for solving convex optimization problems, suitable for solving linearized convex optimization models. Its core is to transform the constrained optimization problem into an unconstrained optimization problem by introducing an obstacle function, and then obtain the optimal solution through iterative solving.

[0074] The specific process is as follows: The global optimal solution output by the particle swarm optimization algorithm is used as the initial iteration point. Introducing a barrier function ,in For the first Slack variables for each constraint, ensuring >0; Construct the augmented objective function ,in For obstacle parameters, The original objective function is used; Newton's method is employed to perform unconstrained optimization of the augmented objective function, and the obstacle parameters are dynamically adjusted during the solution process. μ Initial obstacle parameters After each iteration As the obstacle parameters decrease, the augmented objective function gradually approaches the original objective function; while dynamically adjusting the obstacle parameters, the solution accuracy is monitored. Less than the convergence accuracy ε=10 -6 At the same time, the decreasing rate of the obstacle parameters should be appropriately increased to balance the solution speed and accuracy, and to avoid solution oscillations caused by excessively small obstacle parameters.

[0075] Finally, an iteration termination condition is set, combining the dual constraints of model convergence accuracy and the number of iterations. When the optimization variables satisfy all constraints, the solution is stopped, and the current optimal solution is output as a candidate scheduling strategy. The iteration termination condition includes two aspects: first, the convergence accuracy constraint, which is reached when the difference between the objective function values ​​of two iterations reaches a certain threshold. And the slack variables of all constraints The model is considered to have converged when the number of iterations reaches 50. The second constraint is the number of iterations. The maximum number of iterations for the interior point method is set to 50. If the model has not converged after 50 iterations, the obstacle parameters and the initial iteration point are adjusted, and the solution is solved again.

[0076] When any of the above termination conditions are met, and the optimization variables satisfy all constraints (safety and environmental constraints, uncertainty constraints, etc.), the solution is stopped, the current values ​​of the optimization variables are output, and these are compiled into candidate scheduling strategies. The candidate scheduling strategies must clearly define the operating parameters (air volume, feed rate, feedwater pump power, etc.), unit load allocation ratio, power generation, and other key information for each auxiliary equipment in each time interval, providing a foundation for subsequent multi-scenario verification and robustness testing.

[0077] In summary, the load dynamic adjustment method for optimizing the plant power consumption rate of waste incineration power plants in the above embodiments of the present invention obtains historical operating data from both the source and load sides of the waste incineration power plant to generate waste calorific value prediction sequences and grid load demand prediction sequences, respectively. It determines the prediction error distribution characteristics of the two types of prediction sequences and constructs a fuzzy set to describe the uncertainty of the source and load. Based on preset rules combined with the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and economic operation objectives, it constructs a bibliometric optimization scheduling model and solves it to obtain the optimal scheduling strategy. Based on the optimal scheduling strategy, it generates unit load adjustment instructions to control the operation of auxiliary equipment. Through quantitative modeling of source and load uncertainty, bibliometric optimization scheduling, and precise load dynamic adjustment, it effectively reduces safety margin redundancy, improves the economic operation level of the unit, and ensures stable combustion and compliance with steam parameters and pollutant emissions standards. This solves the technical problems of traditional waste incineration power plant load adjustment methods, which struggle to cope with dual uncertainties of source and load, have excessive safety margins leading to high plant power consumption rates, and whose extensive adjustments easily cause operational and environmental risks.

[0078] Example 3 Please see Figure 2 The figure shows a dynamic load regulation system for optimizing the power consumption of a waste incineration power plant, as proposed in the third embodiment of the present invention. The system includes: The acquisition module 100 is used to acquire historical operating data of the waste incineration power plant on the source side and load side, and generate waste calorific value prediction sequence and power grid load demand prediction sequence based on the historical operating data of the source side and load side, respectively. The determination module 200 is used to determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and to construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics. The solution module 300 is used to construct a sub-Brussels bar optimization scheduling model according to preset rules based on the fuzzy set, the safety and environmental protection constraints of the waste incineration power plant, and the economic operation objectives, and to solve the optimal scheduling strategy based on the sub-Brussels bar optimization scheduling model. The control module 400 is used to generate corresponding unit load adjustment instructions according to the optimal scheduling strategy, and control the operation of auxiliary equipment of the waste incineration power plant according to the unit load adjustment instructions.

[0079] The functions or operation steps implemented by the above modules are largely the same as those in the above method embodiments, and will not be repeated here.

[0080] Example 4 In another aspect, the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the method described in any one of Embodiments 1 to 2 above.

[0081] Example 5 In another aspect, the present invention provides an electronic device, the electronic device including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of any one of the methods described in Embodiments 1 to 2 above.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0084] More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable storage media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0085] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0086] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0087] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for dynamic load adjustment to optimize the power consumption rate of a waste incineration power plant, characterized in that, The method includes: Obtain historical operating data of the waste incineration power plant from both the source and load sides, and generate waste calorific value prediction sequences and grid load demand prediction sequences based on the historical operating data of the source and load sides, respectively. Determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics; According to the preset rules, based on the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and the economic operation objectives, a sub-Bruker optimal scheduling model is constructed, and the optimal scheduling strategy is obtained by solving the sub-Bruker optimal scheduling model. The corresponding unit load adjustment command is generated according to the optimal scheduling strategy, and the operation of the auxiliary equipment of the waste incineration power plant is controlled according to the unit load adjustment command. The steps of acquiring historical operating data of the waste incineration power plant on both the source and load sides, and generating waste calorific value prediction sequences and grid load demand prediction sequences based on the historical operating data of the source and load sides, respectively, include: The source-side data and load-side data of the waste incineration power plant are collected from the historical operation database. The source-side data includes the composition detection data, physical characteristic parameters and combustion response data of the waste entering the furnace. The load-side data includes historical load instructions issued by the power grid dispatch center and actual power feedback data of the units in the plant. Spatiotemporal alignment and feature engineering processing of the source-side data specifically includes: converting the elemental analysis results in the component detection data into feature vectors with fixed carbon content and volatile matter ratios, using the moisture content in the physical property parameters as key input features, and using the furnace temperature and flue gas oxygen content in the combustion response data as auxiliary correction features, to jointly construct a source-side multimodal input feature set; The source-side multimodal input feature set is input into a pre-trained source-side calorific value prediction sub-model based on gated recurrent units, and the source-side calorific value prediction sub-model outputs a waste calorific value prediction sequence within a future preset time window. The deviation analysis and trend decomposition of the load-side data specifically include: calculating the tracking deviation sequence between the historical load command and the actual power feedback data, using the tracking deviation sequence to weight the historical load command with confidence, and constructing a load-side load prediction sub-model by combining the seasonal autoregressive integral moving average model. The load-side load prediction sub-model outputs a predicted sequence of grid load demand within a preset time window.

2. The method for dynamic load adjustment to optimize the power consumption rate of waste incineration power plants according to claim 1, characterized in that, The step of determining the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and constructing a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics, includes: Obtain the first residual set between the predicted calorific value of the waste and the actual measured value, and the second residual set between the predicted power grid load demand and the actual dispatch value; The first residual set and the second residual set are fitted with probability density functions using the kernel density estimation method to obtain the source-side error probability distribution and the load-side error probability distribution. The corresponding higher-order statistical features are extracted based on the source-side error probability distribution and the load-side error probability distribution. The higher-order statistical features include skewness and kurtosis. Based on the source-side error probability distribution and the load-side error probability distribution, a joint probability distribution model of the source-side and load-side prediction errors is constructed using the Copula function to capture the nonlinear correlation between waste calorific value fluctuations and power grid dispatching requirements. Based on the moment characteristics of the joint probability distribution model, a fuzzy set containing all probability distributions that satisfy the moment characteristics is constructed. The fuzzy set is used to simultaneously consider the coupling effect of the uncertainty on both sides of the source and load and the local morphological characteristics of the distribution function in the split-bar optimization.

3. The method for dynamic load adjustment to optimize the power consumption rate of waste incineration power plants according to claim 1, characterized in that, The step of constructing a sub-Browser optimal scheduling model according to preset rules based on the fuzzy set, the safety and environmental constraints of the waste incineration power plant, and the economic operation objectives includes: Confidence intervals are obtained for the fuzzy centralized waste calorific value prediction error and the power grid load demand prediction error, respectively. Based on the confidence intervals, the uncertainty constraints of the sub-Bruker optimal scheduling model are determined. Based on the aforementioned uncertainty constraints, and in conjunction with the boundary conditions of boiler combustion temperature, main steam pressure, and flue gas pollutant emission limits, the safety and environmental constraints of the sub-Bluer bar optimization scheduling model are determined. Based on the aforementioned safety and environmental constraints and the function used to characterize fuel cost and efficiency loss, an objective function is constructed with the goal of minimizing the total operating cost under the worst-case scenario. The sub-Blu-ray optimization scheduling model is then determined based on the objective function and the uncertainty constraints.

4. The method for dynamic load adjustment to optimize the power consumption rate of waste incineration power plants according to claim 3, characterized in that, The step of obtaining the optimal scheduling strategy based on the bibliometric optimization scheduling model includes: The aforementioned split-bar optimization scheduling model is equivalently transformed to include fuzzy set constraints. The bi-level optimization problem is transformed into a solvable linearized convex optimization model; An adaptive hybrid solution algorithm is used to efficiently and accurately solve the transformed linearized convex optimization model and obtain candidate scheduling strategies. The candidate scheduling strategies are verified and robustness checked in multiple scenarios to select the final optimal scheduling strategy that meets safety and environmental protection constraints and achieves the optimal plant power consumption rate and total operating cost.

5. The method for dynamic load adjustment to optimize the power consumption rate of a waste incineration power plant according to claim 4, characterized in that, The step of performing an equivalent transformation on the sub-Bruker optimal scheduling model, converting the min-max bilevel optimization problem with fuzzy set constraints into a solvable linearized convex optimization model, includes: Based on the joint probability distribution model of source-load prediction error, the moment feature boundary of fuzzy set is extracted, and Lagrange dual variables are introduced to transform the maximization constraint in the worst scenario of the sub-Bruker optimization scheduling model into a dual constraint term, thereby eliminating the nested structure of the two-layer optimization. For nonlinear terms in safety and environmental protection constraints, a piecewise linearization method is used for high-precision fitting. An error correction factor is introduced during the fitting process to ensure that the linearization error is controlled within a preset threshold. In light of the start-up and shutdown characteristics of auxiliary equipment in waste incineration power plants, a relaxation term for equipment state switching constraints is added, and the complementary constraint transformation method is used to transform discrete equipment state variables into continuous variables. By integrating linear constraints, dual constraints, and continuous variable constraints, the objective function of the sub-Brussels bar optimization scheduling model is simultaneously retained and linearized for adaptation, ultimately forming a solvable linearized convex optimization model.

6. The method for dynamic load adjustment to optimize the power consumption rate of a waste incineration power plant according to claim 5, characterized in that, The steps of employing an adaptive hybrid solution algorithm to efficiently and accurately solve the transformed linearized convex optimization model and obtain candidate scheduling strategies include: Initialize the solution parameters, determine the initial search range of the optimization variables based on the source-side multimodal input feature set and the load-side load prediction sequence, and use the particle swarm optimization algorithm to perform a global search on the initial solution; The optimal initial solution obtained from the global search is used as the initial iteration point for the precise solution. The interior point method is used to perform local refinement of the transformed convex optimization model. During the solution process, the obstacle parameters are dynamically adjusted to balance the solution speed and accuracy. Set an iteration termination condition, combining the model convergence accuracy and the number of iterations as dual constraints. When the optimization variables satisfy all constraints, stop solving and output the current optimization solution as a candidate scheduling strategy.

7. A load dynamic adjustment system for optimizing the power consumption rate of a waste incineration power plant, characterized in that, The load dynamic adjustment method for optimizing the power consumption rate of a waste incineration power plant as described in any one of claims 1 to 6, the system comprising: The acquisition module is used to acquire historical operating data of the waste incineration power plant on the source side and load side, and generate waste calorific value prediction sequence and power grid load demand prediction sequence based on the historical operating data of the source side and load side, respectively. The determination module is used to determine the prediction error distribution characteristics of the waste calorific value prediction sequence and the power grid load demand prediction sequence, and to construct a fuzzy set to describe the source-load uncertainty based on the prediction error distribution characteristics; The solution module is used to construct a sub-Brussels bar optimization scheduling model according to preset rules based on the fuzzy set, the safety and environmental protection constraints of the waste incineration power plant, and the economic operation objectives, and to solve the optimal scheduling strategy based on the sub-Brussels bar optimization scheduling model. The control module is used to generate corresponding unit load adjustment instructions according to the optimal scheduling strategy, and to control the operation of auxiliary equipment of the waste incineration power plant according to the unit load adjustment instructions.

8. A readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1 to 6.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the program, implements the steps of the method as described in any one of claims 1 to 6.

Images