Artificial Intelligence-Based Power Plant Load Prediction and Optimal Scheduling Method and System for Gas-Steam Combined Cycle Units

Abstract

The present application discloses an artificial intelligence-based load forecasting and optimal scheduling method and system for a gas-steam combined cycle power plant, which relates to the field of artificial intelligence technology. The method and system record the load data and unit output of the gas-steam combined cycle power plant in real time; extract the load fluctuation characteristics based on the load data, obtain the prediction error feedback, introduce the adaptive time window mechanism, and adjust the time window length of the next prediction according to the current load fluctuation characteristics and the prediction error feedback; extract the multi-time scale fusion characteristics of the load data of each time step in the time window according to the updated time window length; construct a load forecasting model, and predict future load data based on the multi-time scale fusion characteristics of each time step; and construct a reinforcement learning model to optimize the unit output adjustment value of the gas-steam combined cycle power plant based on the current unit output, future load data and the adaptive time window mechanism, thereby achieving the adaptability of load forecasting and optimal scheduling.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and system for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence. Background Art

[0002] The combined cycle gas turbine (CCGT) is a highly efficient and clean power generation technology. It combines a gas turbine with a steam turbine, utilizing the waste heat from the high-temperature flue gas of the gas turbine to generate steam, which drives the steam turbine to generate electricity. This significantly improves power generation efficiency and reduces emissions. CCGT power plants offer excellent flexibility and fast startup, making them a vital source of power in power systems.

[0003] The Chinese patent application publication number CN113107626A discloses a combined cycle power generation unit load prediction method based on multivariable LSTM, including the following steps: S1, data acquisition; S2, data preprocessing; S3, dividing the data of one unit into a training set and a test set, and setting the data of the other unit as a validation set; S4, building a multivariable input LSTM neural network model; S5, substituting the validation set data into the model for prediction, and using the combined cycle power as the prediction target of the unit load; S6, evaluating the model based on two evaluation indicators: the loss value in model training and the root mean square error in prediction; S7, generating a true value-predicted value curve.

[0004] Existing technologies usually use fixed time windows, which are difficult to adapt to the dynamic characteristics of load changes. Summary of the Invention

[0005] The present application aims to address, at least to some extent, one of the technical problems in the related art. To this end, one objective of the present application is to propose an artificial intelligence-based method and system for load forecasting and optimal scheduling of a gas-steam combined cycle power plant, which enables comprehensive modeling and forecasting of short-term load fluctuations and long-term trends.

[0006] One aspect of the present application provides an artificial intelligence-based method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant, including:

[0007] Step S100: Recording load data and unit output of a gas-steam combined cycle power plant in real time;

[0008] Step S200: extracting load fluctuation characteristics based on load data, obtaining prediction error feedback, introducing an adaptive time window mechanism, and adjusting the time window length for the next prediction based on the current load fluctuation characteristics and prediction error feedback;

[0009] Step S300: extracting multi-time-scale fusion features of the load data of each time step in the time window according to the updated time window length;

[0010] Step S400: constructing a load forecasting model to predict future load data based on the multi-time scale fusion features of each time step;

[0011] Step S500: Based on the current unit output, future load data and the adaptive time window mechanism, a reinforcement learning model is constructed to optimize the unit output adjustment value of the gas-steam combined cycle power plant;

[0012] The specific method for real-time recording of load data and unit output of a gas-steam combined cycle power plant is as follows:

[0013] Record the load data and unit output of the gas-steam combined cycle power plant in real time and store them in the database, and obtain the load data of the latest TN time steps from the power plant database , perform data cleaning on load data, remove abnormal data and invalid data, and interpolate missing data; among them, represents the load data of the tth time step, and TN is the total number of time steps;

[0014] The specific method of extracting load fluctuation characteristics based on load data, obtaining prediction error feedback, introducing an adaptive time window mechanism, and adjusting the time window length of the next prediction according to the current load fluctuation characteristics and prediction error feedback is as follows:

[0015] Step S210: Divide the time window layers into NI levels according to different time spans. Assume that the time span of the ni-th time window layer is , calculate the number of time windows at this time window level;

[0016] Step S220: For the ni-th time window level, obtain the load data of the nj-th time window ;

[0017] Step S230: For the nj-th time window in the ni-th time window level, calculate the load fluctuation characteristics of the time window based on the length of the time window and the load data of the adjacent time steps in the corresponding time window. ;

[0018] Step S240: Based on the predicted value of load data and the true value , calculate the prediction error feedback ;

[0019] Step S250: Preset load fluctuation threshold and prediction error fluctuation threshold , compare the load fluctuation characteristics with the load fluctuation threshold, the prediction error feedback with the prediction error fluctuation threshold, and update the length of the next prediction time window based on the adaptive time window mechanism;

[0020] The specific method for updating the length of the time window for the next prediction based on the adaptive time window mechanism is:

[0021] When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold When calculating the length of the time window and the adjustment step of the time window length The time difference between the two is compared with the minimum value of the time window length. The size between them is updated to the larger value between the two; when the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the length of the time window is calculated and the adjustment step of the time window length is The time between, compare the time and the maximum value with the time window length If the value between them is smaller than the value between the two, the length of the next predicted time window will be updated to the smaller value between the two; otherwise, the length of the next predicted time window will remain unchanged;

[0022] The specific method for extracting the multi-time scale fusion features of the load data of each time step in the time window according to the updated time window length is:

[0023] Step S310: Based on the updated time window length, for each time step, extract the load characteristics corresponding to each time window level and time window, wherein the load characteristics include load characteristics of the load data in the time domain, frequency domain, and time-frequency domain;

[0024] Step S320: For each time step, calculate the fusion weight of the load feature extracted by the nj-th time window at the ni-th time window level at that time step ;

[0025] Step S330: Based on the fusion weight, the load features extracted from different time windows at time step t are fused to obtain the fused load features ;

[0026] Step S340: Integrate the fusion load features of different time window levels to obtain the multi-time scale fusion feature of the t-th time step ;

[0027] The specific method of constructing the load forecasting model and predicting future load data based on the multi-time scale fusion features of each time step is as follows:

[0028] Step S410: Constructing a load forecasting model, using a long short-term memory neural network model as the basic model and introducing an attention mechanism. The load forecasting model includes an input layer, a long short-term memory layer, an attention mechanism layer, an output layer, and a loss function. The loss function includes a mean square error and a regularization term. The load forecasting model is optimized and trained by minimizing the loss function.

[0029] Step S420: Input the multi-time scale fusion features as input data into the input layer, use the long short-term memory layer to learn the temporal features of the input data, and encode them into the hidden state, apply the attention mechanism on the hidden state, and calculate the attention weight of each time step. ;

[0030] Step S430: Concatenate the hidden state output by the long short-term memory layer and the attention context vector output by the attention mechanism, and obtain the predicted future load data through the output layer. ;

[0031] The specific method for constructing a reinforcement learning model to optimize the unit output adjustment value of a gas-steam combined cycle power plant based on the current unit output, future load data, and an adaptive time window mechanism is as follows:

[0032] Step S510: Define the state space of the reinforcement learning model and encode the predicted future load data, current unit output, and unit operation constraints into the state vector of the reinforcement learning model. ;

[0033] The unit operation constraints include unit output upper and lower limit constraints, unit output change rate constraints, unit start and stop constraints, and unit minimum operating time and downtime constraints;

[0034] Step S520: Define the action space of the reinforcement learning model and encode the output adjustment value of each unit into the action vector of the reinforcement learning model , the action space satisfies the unit operation constraints;

[0035] Step S530: defining a reward function of the reinforcement learning model, wherein the reward function includes a unit operation cost, a load following performance, and a unit operation constraint indicator function;

[0036] Step S540: Using a deep reinforcement learning algorithm to train a reinforcement learning model.

[0037] Step S550: Integrate the adaptive time window mechanism into the reinforcement learning model and dynamically adjust the optimization period T of the reinforcement learning model;

[0038] The adaptive time window mechanism is integrated into the reinforcement learning model, and the optimization period T of the reinforcement learning model is dynamically adjusted. When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold Calculate the optimization cycle and optimization adjustment step size The cycle difference between the two is compared with the minimum value of the optimized cycle When the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the optimization cycle and optimization adjustment step size are calculated. The sum of the periods between the two, and compare the sum of the periods with the maximum value of the optimized period If the value between them is smaller, the next optimization cycle will be updated to the smaller value between them; otherwise, the next optimization cycle will remain unchanged;

[0039] Step S560: Use the updated decision time domain to construct the state space, action space and reward function of the reinforcement learning model in the next time step, make a decision in the t+1th time step, and obtain the optimal unit output adjustment value.

[0040] One aspect of the present application provides an artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling system, including:

[0041] Power plant data acquisition module, used to record the load data and unit output of the gas-steam combined cycle power plant in real time;

[0042] The time window adjustment module is used to extract load fluctuation characteristics based on load data, obtain prediction error feedback, introduce an adaptive time window mechanism, and adjust the time window length of the next prediction based on the current load fluctuation characteristics and prediction error feedback;

[0043] The multi-time scale fusion stage is used to extract the multi-time scale fusion features of the load data at each time step in the time window according to the updated time window length;

[0044] The future load forecasting module is used to build a load forecasting model and predict future load data based on the multi-time scale fusion features of each time step;

[0045] The unit output adjustment module is used to build a reinforcement learning model to optimize the unit output adjustment value of the gas-steam combined cycle power plant based on the current unit output, future load data and the adaptive time window mechanism.

[0046] One aspect of the present application provides a readable storage medium storing a computer program suitable for loading by a processor to execute the steps in a method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence.

[0047] The artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling method and system proposed in this application have the following advantages over the existing technology:

[0048] This application introduces an adaptive time window mechanism that dynamically adjusts the time window length based on current load fluctuation characteristics and forecast error feedback, allowing the selection of time windows to adapt to the dynamic characteristics of load changes. When load fluctuations are severe or forecast errors are large, the time window is shortened to achieve a faster response; when the load is stable or the forecast is accurate, the time window is extended to capture long-term trends. This adaptive mechanism improves the flexibility and adaptability of load feature extraction and forecasting.

[0049] This application extracts multi-timescale fusion features, comprehensively considering the characteristics of load data in the time domain, frequency domain, and time-frequency domain at different time window levels. This multi-timescale fusion method can comprehensively characterize the dynamic changes in load, including short-term fluctuations and long-term trends, improving the accuracy and robustness of load forecasting.

[0050] This application introduces an attention mechanism into the load forecasting model, adaptively assigning importance weights to different historical time steps based on the multi-timescale fusion features of the current time step. This attention mechanism enables the model to automatically focus on the historical information most helpful for the current forecast based on the dynamic characteristics of the current load, improving the relevance and accuracy of the forecast.

[0051] This application integrates the adaptive time window mechanism into the reinforcement learning model, dynamically adjusts the optimization cycle, and makes scheduling decisions and load forecasts synchronously to adapt to the dynamic characteristics of load changes.

[0052] This application comprehensively considers unit operating costs, load tracking performance, and unit operating constraints in the reward function, guiding the reinforcement learning model to develop a dispatching strategy that balances economy, reliability, and feasibility. This multi-objective optimization approach can improve operational efficiency and economic benefits while ensuring safe and stable operation of the power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A flow chart of the method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence provided in this application;

[0054] Figure 2 Flowchart of the implementation method of the adaptive time window mechanism provided by this application;

[0055] Figure 3 Flowchart of the method for obtaining multi-time scale fusion features provided in this application;

[0056] Figure 4 Functional module diagram of the artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling system provided in this application. DETAILED DESCRIPTION

[0057] To better understand the present application, various aspects of the present application will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are merely descriptions of exemplary embodiments of the present application and are not intended to limit the scope of the present application in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0058] In the accompanying drawings, the size, dimensions, and shapes of the elements have been slightly adjusted for ease of illustration. The accompanying drawings are for illustration only and are not drawn strictly to scale. As used herein, the terms "substantially," "approximately," and similar terms are used to indicate approximate values, not degrees, and are intended to illustrate inherent deviations in measurements or calculations that would be recognized by a person of ordinary skill in the art. In addition, in this application, the order in which the steps are described does not necessarily represent the order in which these steps would occur in actual operation, unless otherwise specified or inferred from the context.

[0059] It should also be understood that expressions such as "including", "comprising", "having", "containing" and / or "comprising" are open rather than closed expressions in this specification, which indicate the presence of the stated features, elements and / or components, but do not exclude the presence of one or more other features, elements, components and / or combinations thereof. In addition, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features rather than just the individual elements in the list. In addition, when describing embodiments of the present application, "may" is used to mean "one or more embodiments of the present application". And, the term "exemplary" is intended to refer to an example or illustration.

[0060] Unless otherwise defined, all terms used herein (including engineering and scientific terms) have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that, unless otherwise specified in this application, words defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or overly formal sense.

[0061] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0062] Example 1

[0063] like Figure 1 As shown, the artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling method provided in this application includes:

[0064] Step S100: Recording load data and unit output of a gas-steam combined cycle power plant in real time;

[0065] The specific method for real-time recording of load data and unit output of a gas-steam combined cycle power plant is as follows:

[0066] Record the load data and unit output of the gas-steam combined cycle power plant in real time and store them in the database, and obtain the load data of the latest TN time steps from the power plant database , perform data cleaning on load data, remove abnormal data and invalid data, and interpolate missing data; among them, represents the load data of the tth time step, and TN is the total number of time steps;

[0067] Step S200: extracting load fluctuation characteristics based on load data, obtaining prediction error feedback, introducing an adaptive time window mechanism, and adjusting the time window length for the next prediction based on the current load fluctuation characteristics and prediction error feedback;

[0068] The load fluctuation characteristics are used to measure the degree of fluctuation of load data within a time window, reflecting the rate and amplitude of change of load;

[0069] The prediction error feedback is used to measure the deviation between the load prediction value and the actual value within a time window, reflecting the performance and applicability of the load forecasting model;

[0070] The specific method of extracting load fluctuation characteristics based on load data, obtaining prediction error feedback, introducing an adaptive time window mechanism, and adjusting the time window length of the next prediction according to the current load fluctuation characteristics and prediction error feedback is as follows:

[0071] Step S210: Divide the time window layers into NI levels according to different time spans. Assume that the time span of the ni-th time window layer is , calculate the number of time windows at this time window level;

[0072] Preferably, the time window levels include short-term, medium-term and long-term;

[0073] The calculation formula for the number of time windows is: ,in, represents rounding down, and TN represents the time step of the historical load data used for prediction;

[0074] Step S220: For the ni-th time window level, obtain the load data of the nj-th time window ;

[0075] The load data is expressed as ,in, represents the load data at the tth time step, is the starting time step of the njth time window at the nith time window level, is the length of the nj-th time window at the ni-th time window level;

[0076] The time window length refers to the number of time steps contained in a single time window within each time window level;

[0077] Step S230: For the nj-th time window in the ni-th time window level, calculate the load fluctuation characteristics of the time window based on the length of the time window and the load data of the adjacent time steps in the corresponding time window. ;

[0078] The calculation formula of the load fluctuation characteristics is: ,in, is the load data of the t+1th time step in the time window;

[0079] Step S240: Based on the predicted value of load data and the true value , calculate the prediction error feedback ;

[0080] The calculation formula of the prediction error feedback is: ,in, is the predicted value of the load data at the tth time step;

[0081] Step S250: Preset load fluctuation threshold and prediction error fluctuation threshold , compare the load fluctuation characteristics with the load fluctuation threshold, the prediction error feedback with the prediction error fluctuation threshold, and update the length of the next prediction time window based on the adaptive time window mechanism;

[0082] like Figure 2 As shown, the specific method for updating the length of the time window for the next prediction based on the adaptive time window mechanism is:

[0083] When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold When calculating the length of the time window and the adjustment step of the time window length The time difference between the two is compared with the minimum value of the time window length. The size between them is updated to the larger value between the two; when the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the length of the time window is calculated and the adjustment step of the time window length is The time between, compare the time and the maximum value with the time window length If the value between them is smaller than the value between the two, the length of the next predicted time window will be updated to the smaller value between the two; otherwise, the length of the next predicted time window will remain unchanged;

[0084] The calculation formula for the length of the time window for the next prediction is: , in, is the length of the time window of the last forecast;

[0085] The minimum and maximum values of the time window length are set by those skilled in the art based on the actual operating requirements and limitations of the power plant.

[0086] The load fluctuation threshold and prediction error fluctuation threshold It is set by those skilled in the art based on experience.

[0087] Step S300: extracting multi-time-scale fusion features of the load data of each time step in the time window according to the updated time window length;

[0088] like Figure 3 As shown, the specific method for extracting the multi-time scale fusion features of the load data of each time step in the time window according to the updated time window length is:

[0089] Step S310: Based on the updated time window length, for each time step, extract the load characteristics corresponding to each time window level and time window, wherein the load characteristics include load characteristics of the load data in the time domain, frequency domain, and time-frequency domain;

[0090] The load characteristics of the load data in the time domain include mean, variance and peak-to-valley difference;

[0091] The load characteristics of the load data in the frequency domain are obtained by Fourier transforming the spectrum, and the frequency domain characteristics of the spectrum are extracted as the load characteristics of the load data in the frequency domain; the frequency domain characteristics include frequency distribution entropy and dominant frequency;

[0092] The load characteristics of the load data in the time-frequency domain are extracted by performing wavelet transform on the load data to serve as the load characteristics of the load data in the time-frequency domain; the time-frequency domain characteristics include: energy distribution and energy center of gravity;

[0093] Step S320: For each time step, calculate the fusion weight of the load feature extracted by the nj-th time window at the ni-th time window level at that time step ;

[0094] The calculation formula of the fusion weight is: ,in, is the control parameter;

[0095] The control parameters Used to adjust the impact of prediction error feedback on weights;

[0096] The number of time windows The number of time windows calculated when the time span is equal to the time window step of the current time window level.

[0097] The historical load data is divided into several time windows according to the predicted time window length. The corresponding load features are extracted and fused for each time step of each time window. The fusion weight is the weight of each time window.

[0098] Step S330: Based on the fusion weight, the load features extracted from different time windows at time step t are fused to obtain the fused load features ;

[0099] The calculation formula of the fusion load characteristic is: ,in, is the load characteristic;

[0100] The fused load feature integrates characteristic information of data at different time scales.

[0101] Step S340: Integrate the fusion load features of different time window levels to obtain the multi-time scale fusion feature of the t-th time step ;

[0102] This step aims to extract the characteristics of the load from the perspective of adaptive time scale and different domains, comprehensively characterize the changing pattern of the load, and provide rich information for subsequent load forecasting and optimal scheduling.

[0103] Step S400: constructing a load forecasting model to predict future load data based on the multi-time scale fusion features of each time step;

[0104] The specific method of constructing the load forecasting model and predicting future load data based on the multi-time scale fusion features of each time step is as follows:

[0105] Step S410: Constructing a load forecasting model, using a long short-term memory neural network model as the basic model and introducing an attention mechanism. The load forecasting model includes an input layer, a long short-term memory layer, an attention mechanism layer, an output layer, and a loss function. The loss function includes a mean square error and a regularization term. The load forecasting model is optimized and trained by minimizing the loss function.

[0106] The calculation formula of the loss function is: ,in, represents all learnable parameters of the load forecasting model, is the L2 regularization coefficient, Indicates the predicted value of load data;

[0107] All learnable parameters of the load forecasting model include multiple learnable parameters in the input layer, long short-term memory layer, attention mechanism layer, and output layer;

[0108] The value of the L2 regularization coefficient is set by those skilled in the art based on experience.

[0109] Step S420: Input the multi-time scale fusion features as input data into the input layer, use the long short-term memory layer to learn the temporal features of the input data, and encode them into the hidden state, apply the attention mechanism on the hidden state, and calculate the attention weight of each time step. ;

[0110] The long-term and short-term memory layer is responsible for extracting the temporal characteristics and long-term dependencies of the input data. Its structure includes: a forget gate, an input gate, an output gate, a candidate memory unit, a memory unit and a hidden state. The forget gate is used to control which information in the memory unit of the previous time step should be forgotten. The input gate is used to control which information in the input data of the current time step should be added to the memory unit. The output gate is used to control which information in the memory unit state will eventually be output to the hidden state. The candidate memory unit is used to represent the new memory information brought by the input data of the current time step. The memory unit is used to integrate the memory of the previous time step and the input of the current time step to form a new memory. The hidden state is used to represent the output of the long-term and short-term memory layer of the current time step.

[0111] The role of the attention mechanism is to adaptively assign weights to hidden states of different time steps based on the multi-time-scale fusion features of the current time step, and generate an attention context vector that represents the aggregation of information from previous time steps.

[0112] The calculation process of the attention mechanism is: , in, is the attention score of the t-th time step to the t'th time step, is the learnable attention vector, 、 is the weight matrix, is the hidden state at the t'th time step, The multi-time scale fusion feature input at the t-th time step, is the attention weight of the hidden state at the t'th time step to the prediction at the tth time step, is the attention context vector at the t-th time step;

[0113] The attention score is used to measure the importance of the hidden state at time step t' to the prediction of time step t, where t represents the current time step and t' represents the time step before the current time step;

[0114] The weight matrix Used to map the hidden state at the t'th time step to the same dimension as the attention vector, the weight matrix Used to map the multi-timescale fusion features of the current time step to the same dimension as the attention vector.

[0115] Step S430: Concatenate the hidden state output by the long short-term memory layer and the attention context vector output by the attention mechanism, and obtain the predicted future load data through the output layer. ;

[0116] The calculation formula of the output layer is: ,in, 、 are the learnable output layer parameters, is the hidden state of the final output of the long short-term memory layer;

[0117] Step S500: Based on the current unit output, future load data and the adaptive time window mechanism, a reinforcement learning model is constructed to optimize the unit output adjustment value of the gas-steam combined cycle power plant;

[0118] The specific method for constructing a reinforcement learning model to optimize the unit output adjustment value of a gas-steam combined cycle power plant based on the current unit output, future load data, and an adaptive time window mechanism is as follows:

[0119] Step S510: Define the state space of the reinforcement learning model and encode the predicted future load data, current unit output, and unit operation constraints into the state vector of the reinforcement learning model. ;

[0120] The unit operation constraints include unit output upper and lower limit constraints, unit output change rate constraints, unit start and stop constraints, and unit minimum operating time and downtime constraints;

[0121] The upper and lower limits of the unit output are: ,in, Indicates the The unit output of the unit at the tth time step, and Respectively represent The minimum output limit and maximum output limit of each unit;

[0122] The unit output change rate constraint is: ,in, and Respectively represent The maximum descent rate and maximum ascent rate of the unit;

[0123] The unit start and stop constraints are: ,in, is a binary variable, indicating the The start and stop status of the unit at the tth time step, 1 means running, 0 means stopping, Indicates the The maximum number of starts and stops allowed for a unit within the optimization period, T is the optimization period;

[0124] The minimum operating time and downtime constraints of the unit are: , in, and Respectively represent The minimum operating time and minimum downtime of each unit, Indicates the time step when the unit state changes.

[0125] The values of the minimum output limit and the maximum output limit, the maximum decrease rate and the maximum increase rate, the maximum number of starts and stops, the minimum operating time and the minimum downtime are determined by those skilled in the art based on the technical parameters of the unit.

[0126] Step S520: Define the action space of the reinforcement learning model and encode the output adjustment value of each unit into the action vector of the reinforcement learning model , the action space satisfies the unit operation constraints;

[0127] Step S530: defining a reward function of the reinforcement learning model, wherein the reward function includes a unit operation cost, a load following performance, and a unit operation constraint indicator function;

[0128] The calculation formula of the reward function is:

[0129]

[0130] in, 、 、 is the weight coefficient, N represents the number of units, Indicates the The operating cost of each unit, Indicates the Unit No. The unit output of each time step, Indicates the The predicted value of future load data for time steps, It is the indicating function of the unit operation constraint condition;

[0131] The said The predicted value of the load data of a time step is the load demand of the unit at that time step;

[0132] The unit operation constraint condition indicator function is used to determine whether the unit output meets the unit operation constraint condition. , then the unit operation constraint indicator function is equal to 1, otherwise it is equal to 0;

[0133] Step S540: Using a deep reinforcement learning algorithm to train a reinforcement learning model.

[0134] The deep reinforcement learning algorithm adopts the DQN algorithm;

[0135] The process of training the reinforcement learning model is as follows: using predicted future load data and combining it with unit output to generate training data for the reinforcement learning model; during the training process, using the experience replay mechanism to randomly extract previous samples and update the parameters of the reinforcement learning model to improve sample utilization efficiency and training stability; using the ε-greedy strategy for exploration, randomly selecting actions with probability ε at the beginning of training, and randomly exploring with gradually decreasing probability as training progresses to explore a better scheduling strategy.

[0136] The rate of gradual reduction of the probability ε is set by those skilled in the art based on experience.

[0137] Step S550: Integrate the adaptive time window mechanism into the reinforcement learning model and dynamically adjust the optimization period T of the reinforcement learning model;

[0138] The adaptive time window mechanism is integrated into the reinforcement learning model, and the optimization period T of the reinforcement learning model is dynamically adjusted. When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold Calculate the optimization cycle and optimization adjustment step size The cycle difference between the two is compared with the minimum value of the optimized cycle When the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the optimization cycle and optimization adjustment step size are calculated. The sum of the periods between the two, and compare the sum of the periods with the maximum value of the optimized period If the value between them is smaller, the next optimization cycle will be updated to the smaller value between them; otherwise, the next optimization cycle will remain unchanged;

[0139] Step S560: Use the updated decision time domain to construct the state space, action space, and reward function of the reinforcement learning model at the next time step, make a decision at the t+1th time step, and obtain the optimal unit output adjustment value;

[0140] Based on the above steps, an adaptive time window mechanism is integrated into the optimization cycle adjustment of the reinforcement learning model. When load fluctuations are large or prediction errors are large, the optimization cycle is shortened to enable the model to respond promptly to load changes. When load fluctuations are small and prediction errors are small, the optimization cycle is extended to enable the model to develop long-term optimal scheduling strategies. This method of dynamically adjusting the optimization cycle can improve the adaptability and robustness of the reinforcement learning model, enabling it to better adapt to the changing load characteristics of gas-steam combined cycle power plants.

[0141] Example 2

[0142] like Figure 4 As shown, the artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling system provided in this application includes:

[0143] Power plant data acquisition module, used to record the load data and unit output of the gas-steam combined cycle power plant in real time;

[0144] The time window adjustment module is used to extract load fluctuation characteristics based on load data, obtain prediction error feedback, introduce an adaptive time window mechanism, and adjust the time window length of the next prediction based on the current load fluctuation characteristics and prediction error feedback;

[0145] The multi-time scale fusion stage is used to extract the multi-time scale fusion features of the load data at each time step in the time window according to the updated time window length;

[0146] The future load forecasting module is used to build a load forecasting model and predict future load data based on the multi-time scale fusion features of each time step;

[0147] The unit output adjustment module is used to build a reinforcement learning model to optimize the unit output adjustment value of the gas-steam combined cycle power plant based on the current unit output, future load data and the adaptive time window mechanism.

[0148] Example 3

[0149] According to one embodiment of the present application, a computer-readable storage medium is also provided. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by a processor, the artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling method according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc.

[0150] In addition, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions capable of being executed by a processor to execute instructions corresponding to the method steps provided herein, such as: recording load data and unit output of a gas-steam combined cycle power plant in real time; extracting load fluctuation characteristics based on the load data, obtaining prediction error feedback, introducing an adaptive time window mechanism, and adjusting the time window length for the next prediction based on the current load fluctuation characteristics and prediction error feedback; extracting multi-time-scale fusion features of the load data at each time step within the time window based on the updated time window length; constructing a load forecasting model to predict future load data based on the multi-time-scale fusion features at each time step; and constructing a reinforcement learning model to optimize the unit output adjustment value of the gas-steam combined cycle power plant based on the current unit output, future load data, and the adaptive time window mechanism. When executed by a central processing unit (CPU), this computer program performs the above-described functions defined in the method of the present application.

[0151] The methods, apparatus, and devices of the present application may be implemented in many ways. For example, the methods, apparatus, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the methods according to the present application. Therefore, the present application also covers recording media that store programs for executing the methods according to the present application.

[0152] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0153] The above-described specific embodiments further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for load forecasting and optimal scheduling of gas-steam combined cycle power plants based on artificial intelligence, characterized in that: include: Real-time recording of load data and unit output of gas-steam combined cycle power plants; Based on the load data, the load fluctuation characteristics are extracted, the prediction error feedback is obtained, and the adaptive time window mechanism is introduced. The time window length of the next prediction is adjusted according to the current load fluctuation characteristics and prediction error feedback. The specific method is as follows: according to different time spans, NI time window levels are divided. Assume that the time span of the ni-th time window level is , calculate the number of time windows at the time window level, the calculation formula for the number of time windows is: ,in, Indicates rounding down, TN indicates the time step of the historical load data used for prediction; for the ni-th time window level, obtain the load data of the nj-th time window ,in, represents the load data at the tth time step, is the starting time step of the njth time window at the nith time window level, is the length of the njth time window in the nith time window level; for the njth time window in the nith time window level, the load fluctuation characteristics of the time window are calculated based on the length of the time window and the load data of the adjacent time steps in the corresponding time window ; Based on the predicted value of load data and the true value , calculate the prediction error feedback ;Preset load fluctuation threshold and prediction error fluctuation threshold ; When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold When calculating the length of the time window and the adjustment step of the time window length The time difference between the two is compared with the minimum value of the time window length. The size between them is updated to the larger value between the two; when the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the length of the time window is calculated and the adjustment step of the time window length is The time between, compare the time and the maximum value with the time window length If the value between them is smaller than the value between the two, the length of the next predicted time window will be updated to the smaller value between the two; otherwise, the length of the next predicted time window will remain unchanged; Extract the multi-time scale fusion features of the load data of each time step in the time window according to the updated time window length; Construct a load forecasting model to predict future load data based on the multi-time scale fusion features of each time step; Based on the current unit output, future load data and adaptive time window mechanism, a reinforcement learning model is constructed to optimize the unit output adjustment value of the gas-steam combined cycle power plant.

2. The method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence according to claim 1, characterized in that: The specific method for extracting the multi-time scale fusion features of the load data of each time step in the time window according to the updated time window length is: Based on the updated time window length, for each time step, extract the load characteristics of each time window level and the time window to which it belongs, wherein the load characteristics include the load characteristics of the load data in the time domain, the frequency domain, and the time-frequency domain; For each time step, calculate the fusion weight of the load feature extracted by the nj-th time window at the ni-th time window level at that time step ; Based on the fusion weight, the load features extracted at time step t in different time windows are fused to obtain the fused load features ; Combining the fusion load characteristics of different time window levels, the multi-time scale fusion characteristics of the t-th time step are obtained .

3. The method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence according to claim 2, characterized in that: The specific method of constructing the load forecasting model and predicting future load data based on the multi-time scale fusion features of each time step is as follows: Construct a load forecasting model using a long short-term memory neural network model as the base model and introducing an attention mechanism. The load forecasting model includes an input layer, a long short-term memory layer, an attention mechanism layer, an output layer, and a loss function. The loss function includes a mean square error and a regularization term. The load forecasting model is optimized and trained by minimizing the loss function. The multi-time scale fusion features are input as input data to the input layer, the long short-term memory layer is used to learn the temporal features of the input data and encode them into the hidden state, the attention mechanism is applied to the hidden state, and the attention weight of each time step is calculated. ; The hidden state output by the long short-term memory layer and the attention context vector output by the attention mechanism are spliced together to obtain the predicted future load data through the output layer. .

4. The method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence according to claim 3, characterized in that: The specific method for constructing a reinforcement learning model to optimize the unit output adjustment value of a gas-steam combined cycle power plant based on the current unit output, future load data, and an adaptive time window mechanism is as follows: Step S510: Define the state space of the reinforcement learning model and encode the predicted future load data, current unit output, and unit operation constraints into the state vector of the reinforcement learning model. ; Step S520: Define the action space of the reinforcement learning model and encode the output adjustment value of each unit into the action vector of the reinforcement learning model , the action space satisfies the unit operation constraints; Step S530: defining a reward function of the reinforcement learning model, wherein the reward function includes a unit operation cost, a load following performance, and a unit operation constraint indicator function; Step S540: using a deep reinforcement learning algorithm to train a reinforcement learning model; Step S550: Integrate the adaptive time window mechanism into the reinforcement learning model and dynamically adjust the optimization period T of the reinforcement learning model; Step S560: Use the updated decision time domain to construct the state space, action space and reward function of the reinforcement learning model in the next time step, make a decision in the t+1th time step, and obtain the optimal unit output adjustment value.

5. The method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence according to claim 4, characterized in that: The unit operation constraints include unit output upper and lower limit constraints, unit output change rate constraints, unit start and stop constraints, and unit minimum operating time and shutdown time constraints.

6. The method for load forecasting and optimal scheduling of a gas-steam combined cycle power plant based on artificial intelligence according to claim 5, characterized in that: The adaptive time window mechanism is integrated into the reinforcement learning model, and the optimization period T of the reinforcement learning model is dynamically adjusted. When the load fluctuation characteristic is greater than the load fluctuation threshold When the prediction error feedback is greater than the prediction error fluctuation threshold Calculate the optimization cycle and optimization adjustment step size The cycle difference between the two is compared with the minimum value of the optimized cycle When the load fluctuation characteristic is less than the load fluctuation threshold and the prediction error feedback is less than the prediction error fluctuation threshold, the optimization cycle and optimization adjustment step size are calculated. The sum of the periods between the two, and compare the sum of the periods with the maximum value of the optimized period If the value between them is smaller than , the next optimization cycle will be updated to the smaller value between them; otherwise, the next optimization cycle will remain unchanged.

7. An artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling system, which is implemented based on the artificial intelligence-based gas-steam combined cycle power plant load forecasting and optimization scheduling method according to any one of claims 1 to 6, and is characterized in that: include: Power plant data acquisition module, used to record the load data and unit output of the gas-steam combined cycle power plant in real time; The time window adjustment module is used to extract load fluctuation characteristics based on load data, obtain prediction error feedback, introduce an adaptive time window mechanism, and adjust the time window length of the next prediction based on the current load fluctuation characteristics and prediction error feedback; The multi-time scale fusion stage is used to extract the multi-time scale fusion features of the load data at each time step in the time window according to the updated time window length; The future load forecasting module is used to build a load forecasting model and predict future load data based on the multi-time scale fusion features of each time step; The unit output adjustment module is used to build a reinforcement learning model to optimize the unit output adjustment value of the gas-steam combined cycle power plant based on the current unit output, future load data and the adaptive time window mechanism.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which is suitable for being loaded by a processor to execute the steps in the artificial intelligence-based gas-steam combined cycle unit power plant load forecasting and optimization scheduling method as described in any one of claims 1-6.

Images