Gas Turbine Unit Scheduling Method Based on Multi-Stage Dual Dynamic Programming

Abstract

The present invention provides a gas turbine unit scheduling method based on multi-stage dual dynamic programming, belonging to the technical field of power scheduling. The specific scheduling method is as follows: Predict the load situation and new energy power generation in the corresponding power supply area, and construct a multi-stage scheduling model for the power supply area on the same day with the goal of minimizing cost; Solve the multi-stage scheduling model based on dual dynamic programming to obtain the scheduling strategy of the gas turbine unit; Set feedback nodes. When reaching the feedback nodes, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction results within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model; Resolve the adjusted multi-stage adjustment model again to update the scheduling strategy until the gas turbine unit scheduling on the same day is completed. While using the introduced pre-expanded multi-stage dual dynamic programming algorithm for scheduling planning, the present invention sets a feedback adjustment mechanism, taking into account both the solution speed and accuracy of the gas turbine unit scheduling planning.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and in particular to a gas turbine unit dispatching method based on multi-stage dual dynamic programming. Background Art

[0002] With the continuous adjustment of the energy structure, the mode of coordinated power supply between gas turbine units and new energy power generation has become increasingly common. Among them, gas turbine units serve as the basic power supply of the power system. Accurate and efficient power dispatching for gas turbine units is crucial for ensuring power supply stability, reducing operating costs, and promoting the rational use of energy.

[0003] Along with the expansion of the scale of the power system and the diversification of the energy structure, the scenario of coordinated power supply between gas turbine units and new energy has become increasingly complex. The computational complexity of traditional solution methods has increased sharply, and the solution efficiency is difficult to meet the requirements. When determining the optimal dispatching strategy for a large-scale system, it is necessary to traverse a vast number of combinations, which takes too long and cannot output decision-making schemes in a timely manner.

[0004] To solve the above problems, related technologies achieve the dispatching planning for gas turbine units by introducing a pre-expanded multi-stage dual dynamic programming algorithm. This algorithm can significantly reduce the computational complexity and remarkably improve the solution efficiency by decomposing sub-problems and applying the dual principle, and quickly achieve the global optimum of the gas turbine unit dispatching planning.

[0005] However, the model construction of introducing the pre-expanded multi-stage dual dynamic programming is carried out based on simplified assumptions, which deviates from the actual operation of the power system, has poor adaptability to extreme or complex change scenarios, and has a high degree of dependence on data. Errors, missing or delays in data will seriously affect the quality of dual variable calculation and dispatching decision-making. For the gas turbine unit dispatching in a power system with a large amount of new energy, the accuracy of the dispatching planning is relatively low. Summary of the Invention

[0006] The object of the present invention is to overcome the disadvantages that when using the pre-expanded multi-stage dual dynamic programming to perform rapid dispatching planning for gas turbine units in the prior art, it has poor adaptability to extreme or complex change scenarios, has a high degree of dependence on data, and the accuracy of the dispatching planning is relatively low. A gas turbine unit dispatching method based on multi-stage dual dynamic programming is provided. On the basis of using the pre-expanded multi-stage dual dynamic programming to perform rapid dispatching planning for gas turbine units, by setting feedback nodes, using the gas turbine unit operation information and predicted fluctuation information, the model parameters and cycle length are dynamically adjusted to re-solve and update the dispatching strategy, which can timely adjust the unit operation to ensure power supply, enhance the adaptability to extreme or complex change scenarios, further reduce the dependence on data, reduce the influence of data errors through data verification and correction and model self-adaptability optimization, and improve the accuracy of the dispatching planning.

[0007] The object of the present invention is achieved by the following technical solutions:

[0008] A gas turbine unit scheduling method based on multi-stage dual dynamic programming, comprising:

[0009] Predicting the load conditions and new energy power generation in the power supply area corresponding to the gas turbine unit;

[0010] Taking cost minimization as the optimization objective, and combining the prediction results, constructing a multi-stage scheduling model for the day-ahead unit commitment planning and intra-day unit output scheduling in the power supply area;

[0011] Solving the multi-stage scheduling model based on dual dynamic programming to obtain the scheduling strategy of the gas turbine unit;

[0012] Setting feedback nodes according to the day-ahead planning period and intra-day scheduling period in the multi-stage scheduling model;

[0013] When reaching the feedback node, obtaining the operation information of the gas turbine unit and the fluctuation information of the prediction results in the corresponding time period, and correspondingly adjusting the model parameters and cycle length of the multi-stage scheduling model;

[0014] Resolving the adjusted multi-stage adjustment model again to update the scheduling strategy of the gas turbine unit until the scheduling of the gas turbine unit for the day is completed.

[0015] Based on the rapid scheduling planning of the gas turbine unit using the multi-stage dual dynamic programming with introduced pre-expansion, by setting feedback nodes, verifying and correcting the load and new energy power generation prediction data using the gas turbine unit operation information and prediction fluctuation information, reducing the influence of data errors. This method of dynamically adjusting the parameters and cycle length of the multi-stage scheduling model according to real-time information, resolving and updating the scheduling strategy again can timely adjust the unit operation to ensure power supply, enhance the adaptability to extreme or complex change scenarios, further reduce the dependence on data, reduce the influence of data errors through data verification and correction and model self-adaptation optimization, and improve the accuracy of the scheduling planning.

[0016] Further, the setting of feedback nodes according to the day-ahead planning period and intra-day scheduling period in the multi-stage scheduling model includes:

[0017] Determining the current day-ahead planning information and intra-day scheduling information according to the scheduling strategy;

[0018] Taking the start time of each cycle as an alternative feedback node, and calculating the adjustment operation information of the gas turbine unit at each alternative feedback node according to the corresponding day-ahead planning information and intra-day scheduling information;

[0019] Based on the regulation operation information, calculate the regulation cost and regulation duration at the alternative feedback nodes corresponding to the day-ahead planning and the alternative feedback nodes corresponding to the intra-day scheduling respectively;

[0020] Calculate the time coincidence degree between the alternative feedback nodes corresponding to the day-ahead planning and the intra-day scheduling, and set the feedback nodes in combination with the regulation cost and regulation duration.

[0021] Furthermore, setting the feedback nodes according to the day-ahead planning period and the intra-day scheduling period in the multi-stage scheduling model further includes,

[0022] Every time the scheduling strategy is updated, reset the remaining feedback nodes according to the day-ahead planning period and the intra-day scheduling period in the adjusted multi-stage scheduling model.

[0023] Comprehensively considering various factors such as day-ahead planning information, intra-day scheduling information, regulation duration, time coincidence degree, etc., set the feedback nodes, and dynamically adjust the feedback nodes as the scheduling strategy is updated, providing more real-time data support for the scheduling decision of the gas turbine unit and ensuring the stable and reliable power supply. And the way of dynamically updating the feedback nodes can obtain the real-time changes of the operation state of the gas turbine unit, power load and new energy power generation in a timely manner. In the face of complex and changeable operating environments and emergencies such as extreme weather and large fluctuations in load, it can quickly respond and adjust the scheduling strategy, effectively coping with various uncertainties. And it can also capture the key time points of strategy adjustment in a timely manner, reduce unnecessary waiting and ineffective operations, make the regulation operation of the gas turbine unit more efficient, and improve the overall operation efficiency of the power system.

[0024] Furthermore, when reaching the feedback node, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction result within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model, including:

[0025] Obtain the operation information of the gas turbine unit, the actual load situation and the actual new energy power generation information within the time period from the previous feedback node to the current feedback node under the current scheduling strategy, and determine the fluctuation information of the prediction result;

[0026] Identify the adjustment demand according to the operation information of the gas turbine unit and the fluctuation information of the prediction result;

[0027] According to the identification result of the adjustment demand, combine the operation information of the gas turbine unit and the fluctuation information of the prediction result to adjust the model parameters and cycle length of the multi-stage scheduling model.

[0028] Furthermore, the identification of the adjustment demand according to the operation information of the gas turbine unit and the fluctuation information of the prediction result includes:

[0029] Obtain the gas turbines currently in the operating state and their corresponding outputs, and calculate the output deviation value of the gas turbines in combination with the fluctuation information of the prediction results;

[0030] Calculate the output margin of the gas turbines currently in the operating state according to the gas turbine operation information;

[0031] Based on the output margin and the output deviation value, identify the adjustment requirements for the day-ahead planning and intra-day scheduling.

[0032] Furthermore, according to the recognition result of the adjustment requirements, in combination with the gas turbine operation information and the fluctuation information of the prediction results, adjust the model parameters and cycle length of the multi-stage scheduling model, including:

[0033] When the output margin exceeds the output deviation value, maintain the model parameters and cycle length in the day-ahead planning stage, adjust the output corresponding constraint conditions in the intra-day scheduling stage according to the output deviation value and the gas turbine operation information, and shorten the cycle length of the intra-day scheduling stage;

[0034] When the output margin does not exceed the output deviation value, adjust the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbines in the day-ahead planning stage according to the gas turbine operation information, adjust the output corresponding constraint conditions in the intra-day scheduling stage according to the output deviation value, and shorten the cycle lengths of the day-ahead planning stage and the intra-day scheduling stage.

[0035] Furthermore, the adjustment of the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbines in the day-ahead planning stage according to the gas turbine operation information includes:

[0036] Determine the running duration of the gas turbines currently in the operating state, and set the earliest opening and closing control times of each gas turbine according to the corresponding minimum continuous running time and minimum shutdown time;

[0037] Add the earliest opening and closing control times of each gas turbine to the corresponding constraint conditions of the gas turbines.

[0038] Furthermore, the re-solving of the adjusted multi-stage adjustment model also includes:

[0039] Every time a feedback node is reached, correct the prediction model for load conditions and new energy power generation according to the fluctuation information of the prediction results;

[0040] Re-predict the load conditions and new energy power generation according to the corrected prediction model;

[0041] Based on the newly obtained prediction results, solve the adjusted multi-stage adjustment model.

[0042] Determine the predicted fluctuations by obtaining the actual operation and power generation information between adjacent feedback nodes, so as to accurately identify the adjustment requirements, and then flexibly adjust the parameters and cycle length of the multi-stage scheduling model in combination with relevant information. Further, differentially adjust the day-ahead plan and intra-day scheduling according to the output margin and deviation value. When the margin is sufficient, accurately optimize the intra-day scheduling. When the margin is insufficient, comprehensively adjust the plan and scheduling to ensure the economy and reliability of the scheduling strategy. At the same time, set the earliest start and stop control time of the gas turbine unit and incorporate it into the constraint conditions to ensure the reasonable operation of the unit. Each time reaching the feedback node, use the fluctuation information to correct the prediction model and re-predict, and solve the adjusted model based on the new results, so that the scheduling strategy always keeps up with the actual changes, effectively improving the ability of the power system to cope with uncertainties and achieving multiple goals of cost reduction, efficient resource utilization and stable power supply.

[0043] Further, the solution of the multi-stage scheduling model based on dual dynamic programming includes:

[0044] For the day-ahead planning stage, define the stage of the corresponding dual dynamic programming according to the day-ahead planning cycle and set the corresponding state variables;

[0045] Set the corresponding dual variables according to the constraint conditions corresponding to the day-ahead planning stage and construct the corresponding dual objective function;

[0046] Starting from the last stage, calculate the optimal scheduling decision for this stage according to the values of the current corresponding state variables and dual variables, and combine the optimal decision result of this stage with the initial state of the next stage, and push forward to the previous stage until the optimal unit combination strategy for the day-ahead planning stage is obtained.

[0047] Further, the solution of the multi-stage scheduling model based on dual dynamic programming also includes:

[0048] For the intra-day scheduling stage, define the stage of the corresponding dual dynamic programming according to the intra-day planning cycle and set the corresponding state variables in combination with the optimal unit combination strategy of the day-ahead planning stage;

[0049] Set the corresponding dual variables according to the constraint conditions corresponding to the intra-day scheduling stage and construct the corresponding dual objective function;

[0050] Starting from the last stage, calculate the optimal scheduling decision for this stage according to the values of the current corresponding state variables and dual variables, and combine the optimal decision result of this stage with the initial state of the next stage, and push forward to the previous stage until the optimal unit output allocation strategy for the intra-day scheduling stage is obtained.

[0051] The beneficial effects of the present invention are:

[0052] (1)Based on the rapid scheduling plan of gas turbines using the multi-stage dual dynamic programming with pre-expansion introduced, by setting feedback nodes, the operation information of gas turbines and the predicted fluctuation information are used to verify and correct the load and new energy power generation prediction data, reducing the impact of data errors. This method of dynamically adjusting the parameters and cycle length of the multi-stage scheduling model according to real-time information and re-solving and updating the scheduling strategy can timely adjust the operation of the units to ensure power supply, enhance the adaptability to extreme or complex change scenarios, further reduce the dependence on data, and reduce the impact of data errors through data verification and correction and model self-adaptation optimization, improving the accuracy of the scheduling plan.

[0053] (2)Comprehensively considering various factors such as the day-ahead planning information, intra-day scheduling information, adjustment duration, and time coincidence, feedback nodes are set, and the feedback nodes are dynamically adjusted as the scheduling strategy is updated, providing more real-time data support for the scheduling decision-making of gas turbines and ensuring the stable and reliable power supply. Moreover, the method of dynamically updating feedback nodes can timely obtain the real-time changes in the operation status of gas turbines, power load, and new energy power generation. In the face of complex and changeable operating environments and emergencies such as extreme weather and large load fluctuations, it can quickly respond and adjust the scheduling strategy, effectively coping with various uncertainties. It can also timely capture the key time points of strategy adjustment, reduce unnecessary waiting and ineffective operations, make the adjustment operations of gas turbines more efficient, and improve the overall operating efficiency of the power system.

[0054] (3)The predicted fluctuation is determined by obtaining the actual operation and power generation information between adjacent feedback nodes, so as to accurately identify the adjustment requirements. Then, combined with relevant information, the parameters and cycle length of the multi-stage scheduling model are flexibly adjusted. Further, the day-ahead planning and intra-day scheduling are differentially adjusted according to the output margin and deviation value. When the margin is sufficient, the intra-day scheduling is accurately optimized; when the margin is insufficient, the planning and scheduling are comprehensively adjusted to ensure the economy and reliability of the scheduling strategy. At the same time, the earliest start and stop control time of gas turbines is set and incorporated into the constraint conditions to ensure the reasonable operation of the units. Each time a feedback node is reached, the prediction model is corrected using the fluctuation information and re-predicted, and the adjusted model is solved based on the new results, so that the scheduling strategy always keeps up with the actual changes, effectively enhancing the power system's ability to cope with uncertainties and achieving multiple goals of cost reduction, efficient resource utilization, and stable power supply. Description of the Drawings

[0055] Figure 1 It is a flow diagram of an embodiment of the present invention. Detailed Embodiments

[0056] The present invention will be further described below with reference to the drawings and embodiments.

[0057] Embodiment: A gas turbine scheduling method based on multi-stage dual dynamic programming, asFigure 1 As shown in

[0058] Predict the load conditions and new energy generation in the power supply area corresponding to the gas turbine unit;

[0059] Taking cost minimization as the optimization goal, combined with the prediction results, construct a multi-stage scheduling model for the day-ahead unit commitment planning and intra-day unit output scheduling in the power supply area;

[0060] Solve the multi-stage scheduling model based on dual dynamic programming to obtain the scheduling strategy of the gas turbine unit;

[0061] Set feedback nodes according to the day-ahead planning period and intra-day scheduling period in the multi-stage scheduling model;

[0062] When reaching the feedback node, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction results within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model;

[0063] Resolve the adjusted multi-stage adjustment model again, update the scheduling strategy of the gas turbine unit until the scheduling of the gas turbine unit for the day is completed.

[0064] Accurate load and new energy generation predictions are the basis for subsequent scheduling decisions. Load prediction is affected by various factors, such as season, time, weather conditions, and economic activities within the power supply area. New energy generation, especially solar and wind power generation, has strong randomness and intermittency, and its power generation is closely related to meteorological conditions such as light intensity and wind speed.

[0065] On this basis, first collect historical load data, new energy generation data, and relevant meteorological data and other influencing factor data in the power supply area, clean the data, remove outliers and missing values, and perform normalization processing for subsequent use in the prediction model.

[0066] For the load conditions and new energy generation in the power supply area, traditional time series analysis methods such as ARIMA (Autoregressive Integrated Moving Average Model) can be used, or machine learning algorithms such as neural networks and support vector machines can also be used as prediction models, combined with the processed historical load data, new energy generation data, and influencing factor data to predict the load conditions and new energy generation.

[0067] For the established multi-stage scheduling model of day-ahead unit commitment planning and intra-day unit output scheduling, it can be specifically divided into two stages, one is day-ahead unit commitment planning, and the other is intra-day unit output scheduling.

[0068] In the day-ahead planning stage, based mainly on the results of load forecasting and new energy power generation forecasting, the start-stop combinations of gas turbines are determined in advance to formulate a preliminary framework for subsequent power supply. Specifically, with cost minimization as the core objective, various cost factors are considered, such as gas cost, equipment maintenance cost, and start-stop cost.

[0069] Among them, the fuel cost is related to the gas consumption of the gas turbine during power generation and can be calculated based on the power generation efficiency and gas price of the gas turbine. The equipment maintenance cost is related to the operating time and start-stop times of the gas turbine. Frequent start-stop will increase equipment wear and require more maintenance costs. The start-stop cost is the additional cost generated each time the gas turbine starts and stops, including costs such as equipment preheating and cooling.

[0070] Therefore, the objective function in the day-ahead planning stage can be expressed as:

[0071] ;

[0072] Among them, is the expected cost of the start-stop combination of gas turbines, is the number of gas turbines, is the number of time periods in the day-ahead planning, is the fuel cost coefficient of the th gas turbine, is the power generation of the th gas turbine at time t, is the maintenance cost coefficient of the th gas turbine, is the operating time of the th gas turbine at is the start-stop cost coefficient, is the start-stop state (start is 1, stop is 0) of the i-th gas turbine at time.

[0073] The constraint conditions in the day-ahead planning stage include power balance constraint, unit output constraint, and start-stop constraint.

[0074] Among them, the power balance constraint is that in each time period, the sum of the power generation of all gas turbines and the new energy power generation must be equal to the load demand in that time period. The unit output constraint is that the output of each gas turbine should be between its minimum output and maximum output.

[0075] The start-stop constraint is to consider the minimum continuous operation time and minimum continuous shutdown time of the unit to avoid frequent start-stop of the gas turbine.

[0076] During the intraday scheduling stage, based on the day-ahead unit commitment planning, the output of the units is dynamically adjusted according to the real-time load changes during the day, the actual situation of new energy power generation, and the real-time operating status of gas units, so as to further optimize the cost and ensure the stable operation of the power system.

[0077] The goal of intraday scheduling is still to minimize costs, but real-time operating conditions need to be considered. In addition to fuel costs, maintenance costs, and start-up and shut-down costs, additional costs incurred due to adjusting the unit output may also need to be considered, such as the loss costs of unit equipment during rapid load increase and decrease.

[0078] Therefore, the objective function for this stage can be expressed as:

[0079] ;

[0080] where, is the number of time periods for intraday scheduling, is the cost coefficient for the output adjustment of the th gas unit, is the output adjustment amount of the i-th gas unit at time.

[0081] The constraint conditions during the intraday scheduling stage are similar to those in the day-ahead planning stage, including power balance constraints and unit output constraints. In addition, they also include the ramping constraints of gas units.

[0082] Among them, the power balance constraint during the intraday scheduling stage is similar to that in the day-ahead planning stage, but the load data and new energy power generation used should be real-time data to ensure the balance of power supply and demand in each time period.

[0083] Based on the above objective function and constraint conditions, the multi-stage scheduling model is solved using dual dynamic programming.

[0084] For the day-ahead planning stage, the stages corresponding to the dual dynamic programming are defined according to the day-ahead planning period, and the corresponding state variables are set;

[0085] According to the corresponding constraint conditions in the day-ahead planning stage, the corresponding dual variables are set, and the corresponding dual objective function is constructed;

[0086] Starting from the last stage, according to the values of the current corresponding state variables and dual variables, the optimal scheduling decision for this stage is calculated, and the optimal decision result of this stage is combined with the initial state of the next stage and advanced to the previous stage until the optimal unit commitment strategy for the day-ahead planning stage is obtained.

[0087] In the definition stage, since the target of the day-ahead plan is generally one day and can be specifically divided into 24 time periods, each time period can be defined as a stage of the dual dynamic programming, and each stage corresponds to a time interval. The unit commitment decision is made within this interval.

[0088] The state variables should include the key information that can describe the system state of the current stage in order to make reasonable scheduling decisions. In this embodiment, the start-stop state of the gas turbine unit, the predicted value of the load demand, and the predicted value of the new energy power generation are specifically set as the state variables.

[0089] According to the corresponding constraints of the day-ahead planning stage, the corresponding dual variables are set. On this basis, according to the objective function of the day-ahead plan, the constraints are introduced through Lagrangian relaxation to construct the corresponding dual objective function.

[0090] Starting from the last stage, through optimization algorithms such as linear programming, the optimal decision of this stage is calculated. The optimal decision includes the start-stop state and power generation of each gas turbine unit during the corresponding time period of this stage under the goal of minimizing cost.

[0091] Then, the optimal decision result of the last stage is combined with the initial state of the previous stage, and the optimization algorithm is used again to obtain the optimal decision of the corresponding stage.

[0092] Continuously repeat the above steps of forward recursion, that is, calculate the optimal decision of each stage and use it as one of the initial conditions of the previous stage until the optimal decision of the first stage is calculated, so as to obtain the optimal start-stop state of each unit during each time period of the entire day-ahead planning stage, that is, the optimal unit commitment strategy.

[0093] Similarly, for the intra-day scheduling stage, the stages of the corresponding dual dynamic programming are defined according to the intra-day planning cycle, and the corresponding state variables are set in combination with the optimal unit commitment strategy of the day-ahead planning stage;

[0094] According to the corresponding constraints of the intra-day scheduling stage, the corresponding dual variables are set, and the corresponding dual objective function is constructed;

[0095] Starting from the last stage, according to the values of the current corresponding state variables and dual variables, the optimal scheduling decision of this stage is calculated, and the optimal decision result of this stage is combined with the initial state of the next stage, and it is advanced to the previous stage until the optimal unit output allocation strategy of the intra-day scheduling stage is obtained.

[0096] In the definition stage, although the target of the intra-day scheduling is also one day, the time period is short. Specifically, 15 minutes can be used as a time period to implement the corresponding stage setting.

[0097] For intraday scheduling, the start-stop status of gas turbines, the actual values of new energy generation, the actual values of load demand, and the actual output of gas turbines are used as the state variables for intraday scheduling.

[0098] In the corresponding stage and state variables, starting from the last stage, a forward recursion is performed to obtain the optimal output allocation strategy for each unit in each time period during the intraday scheduling stage.

[0099] Considering that the constraint conditions and cycle lengths of the multi-stage scheduling model will be updated according to the actual situation, and both the stage and state variables can be set, the method of solving the optimal strategy can also adapt to the scenarios where the constraint conditions and cycle lengths change.

[0100] Moreover, after updating the multi-stage scheduling model, when solving the multi-stage scheduling model based on dual dynamic programming, the executed time periods are excluded, and the start-stop status and output of gas turbines in the uncompleted time periods are used to re-plan and schedule.

[0101] The setting of feedback nodes according to the day-ahead planning cycle and intraday scheduling cycle in the multi-stage scheduling model includes:

[0102] Determine the current day-ahead planning information and intraday scheduling information according to the scheduling strategy;

[0103] Taking the start time of each cycle as an alternative feedback node, calculate the adjustment operation information of the gas turbine at each alternative feedback node according to the corresponding day-ahead planning information and intraday scheduling information;

[0104] Based on the adjustment operation information, calculate the adjustment cost and adjustment duration at the corresponding alternative feedback nodes of the day-ahead planning and intraday scheduling respectively;

[0105] Calculate the time coincidence degree between the corresponding alternative feedback nodes of the day-ahead planning and intraday scheduling, and set the feedback nodes in combination with the adjustment cost and adjustment duration.

[0106] The day-ahead planning information covers the start-stop plan of gas turbines formulated ahead of time and the set values of power generation at each time period according to load forecasting and new energy generation forecasting. The intraday scheduling information is the relevant information of the output planning of gas turbines after dynamically adjusting the power generation of the units during the day based on real-time load changes, actual new energy generation conditions, and the real-time operating status of gas turbines.

[0107] Considering that the cycle lengths applied in the day-ahead planning and day-ahead scheduling stages are different during planning, therefore, alternative feedback nodes are selected for the day-ahead planning and day-ahead scheduling stages respectively. Here, the cycle can be each hourly time period in the day-ahead planning or a shorter time interval in the intraday scheduling, such as 15 minutes, etc.

[0108] According to the day-ahead planning information and the intra-day scheduling information, calculate the regulation operation information of the gas turbine unit at each alternative feedback node, including the start-stop operation of the unit, the adjustment amount of the power generation, etc. For example, in the intra-day scheduling, one of the alternative feedback nodes is 11:00 am. According to the then load and new energy generation situation, as well as the previously formulated scheduling strategy, it can be calculated that one of the gas turbine units needs to reduce the power generation from 150 MW to 120 MW at this alternative feedback node. Among them, this reduction operation of the power generation is the regulation operation information of this alternative feedback node.

[0109] Based on the regulation operation information, calculate the regulation cost and regulation duration of the day-ahead planning and day-ahead scheduling corresponding to the alternative feedback node.

[0110] Among them, the regulation cost includes the change in fuel cost, equipment wear cost, etc. When the gas turbine unit increases the power generation, the fuel consumption increases, and at the same time, the wear of the equipment will also increase accordingly. And further calculate the regulation cost according to the increased fuel consumption and the wear cost corresponding to the equipment wear.

[0111] The regulation duration refers to the time required to complete the corresponding regulation operation. For example, at one of the alternative feedback nodes, it is necessary to increase the power generation of one of the gas turbine units from 100 MW to 150 MW. Subject to the ramp rate limit of this gas turbine unit, it takes 5 minutes to complete. These 5 minutes are the regulation duration.

[0112] For the time overlap degree between the day-ahead planning and the intra-day scheduling corresponding to the alternative feedback node, it is determined by calculating the proportion of the overlapping time of the two in the total time. And before calculating the time overlap degree, first sort the corresponding alternative feedback nodes according to the regulation cost, regulation duration and execution time point. The higher the regulation cost, the longer the regulation duration, and the earlier the execution time point, the higher the final sorting. Then calculate the time overlap degree between the two alternative feedback nodes in turn according to the sorting.

[0113] For the final feedback node, preferentially select the alternative feedback node with a higher regulation cost, a longer regulation duration and a higher time overlap degree as the actual feedback node. A high regulation cost and a long regulation duration often mean that at this time node, the operation amount is large, and a small prediction deviation may cause a large execution deviation. Therefore, it is necessary to monitor the actual operation situation of this node and adjust the multi-stage scheduling model accordingly to ensure that the multi-stage scheduling model is closer to the actual situation and ensure the accuracy of the subsequent scheduling strategy. And a high time overlap degree can ensure adjustment at the critical time point, timely capture the time node with a large deviation between the plan and the actual situation, so as to more effectively optimize the scheduling strategy and ensure the stable operation of the power system.

[0114] After each update of the scheduling strategy, since both the parameters and the operating conditions of the multi-stage scheduling model have changed, it is necessary to re-evaluate the feedback nodes for the remaining time periods.

[0115] Therefore, after each update of the scheduling strategy, the remaining feedback nodes are reset according to the day-ahead planning period and the intra-day scheduling period in the adjusted multi-stage scheduling model.

[0116] After setting the feedback nodes, when reaching the feedback nodes, obtain the operating information of the gas turbine units and the fluctuation information of the prediction results within the corresponding time period, and correspondingly adjust the model parameters and cycle lengths of the multi-stage scheduling model.

[0117] Specifically, obtain the operating information of the gas turbine units, the actual load conditions, and the actual power generation information of new energy within the time period from the previous feedback node to the current feedback node under the current scheduling strategy, and determine the fluctuation information of the prediction results;

[0118] Identify the adjustment requirements based on the operating information of the gas turbine units and the fluctuation information of the prediction results;

[0119] Based on the identification results of the adjustment requirements, combine the operating information of the gas turbine units and the fluctuation information of the prediction results to adjust the model parameters and cycle lengths of the multi-stage scheduling model.

[0120] Among them, the previous feedback node mentioned refers to the previous time node when the multi-stage scheduling model is executed.

[0121] The operating information of the gas turbine units includes the power generation power, start-stop status, equipment operating parameters, etc. of the gas turbine units.

[0122] After obtaining the actual load conditions and the actual power generation information of new energy, compare them with the prediction results to determine the fluctuation information of the prediction results.

[0123] Since the current scheduling strategy of the gas turbine units is formulated based on the prediction results, the fluctuation information of the prediction results can intuitively reflect the execution deviation caused by the prediction deviation. If there is an execution deviation, there is an adjustment requirement.

[0124] For example, when the actual load conditions are higher than the load conditions in the prediction results and the actual power generation information of new energy is consistent with the prediction results, if the gas turbine units still operate according to the original scheduling strategy, the power generation capacity in the current power supply area will not be able to meet the load demand, and it is necessary to adjust the unit start-stop strategy or adjust the output of the existing operating units to compensate for this part of the load demand deviation.

[0125] Among them, identifying the adjustment requirements based on the operating information of the gas turbine units and the fluctuation information of the prediction results includes:

[0126] Obtain the gas turbines currently in the operating state and their corresponding outputs, and calculate the output deviation value of the gas turbines in combination with the fluctuation information of the prediction results;

[0127] Calculate the output margin of the gas turbines currently in the operating state based on the gas turbine operation information;

[0128] Based on the output margin and the output deviation value, identify the adjustment requirements for the day-ahead planning and intra-day scheduling.

[0129] To achieve accurate adjustment, obtain the gas turbines currently in the operating state and their corresponding outputs, compare the output that should be provided under the fluctuation information of the prediction results with the actual outputs of all operating gas turbines, and determine the output deviation value of the gas turbines that need to be compensated.

[0130] The output margin refers to the range of power generation that the gas turbine can still increase or decrease under the current operating state. For each operating gas turbine, the output margin can be calculated through its technical parameters, such as the maximum output and minimum output limits.

[0131] Compare the output margin and the output deviation value to identify the adjustment requirements.

[0132] Among them, if the output deviation value is large and exceeds the output margin of the gas turbine, it proves that the current dispatching strategy cannot meet the load demand, and it is necessary to adjust the unit start-stop plan or power generation power distribution in the day-ahead planning.

[0133] If the output deviation value is within the output margin of the unit, it is necessary to adjust the output of each gas turbine. If the fluctuation of the prediction results persists or fluctuates greatly, it is necessary to adjust the corresponding multi-stage dispatching model to better match the real-time changes of the load and new energy power generation while making up for this part of the execution deviation.

[0134] Specifically, according to the identification result of the adjustment requirements, combine the gas turbine operation information and the fluctuation information of the prediction results to adjust the model parameters and cycle length of the multi-stage dispatching model, including:

[0135] When the output margin exceeds the output deviation value, maintain the model parameters and cycle length of the day-ahead planning stage, adjust the output corresponding constraint conditions of the intra-day scheduling stage according to the output deviation value and the gas turbine operation information, and shorten the cycle length of the intra-day scheduling stage.

[0136] Since the output margin is sufficient to make up for the output deviation value, it indicates that the overall framework formulated in the current day-ahead planning stage can still meet the changes in power demand. Therefore, the model parameters and cycle length of the day-ahead planning stage are maintained. This means that key parameters such as the unit start-stop plan and power generation power distribution formulated in the day-ahead planning do not need to be adjusted on a large scale, ensuring the stability and continuity of the day-ahead planning.

[0137] Based on the output deviation value, the output compensation target of the gas turbine unit can be determined, and the operation information of the gas turbine unit can provide limitations for the adjustment of the constraint conditions. Specifically, by increasing the lower output limit constraint, etc., to prompt the corresponding gas turbine unit to increase its output to meet the load demand. At the same time, by combining the operation information of the gas turbine unit to determine the real-time operation state of the gas turbine unit, the upper output limit constraint can also be reasonably adjusted through parameters such as the equipment temperature and pressure of the gas turbine unit to ensure that the gas turbine unit increases its power generation under the premise of safe operation.

[0138] Furthermore, shorten the cycle length of the intraday scheduling stage to more timely respond to the real-time changes in load and new energy power generation. According to the real-time output deviation value and the operation of the gas turbine unit, dynamically adjust the unit output to achieve a more accurate power supply and demand balance.

[0139] When the output margin does not exceed the output deviation value, adjust the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbine unit in the day-ahead planning stage according to the operation information of the gas turbine unit, adjust the output corresponding constraint conditions in the intraday scheduling stage according to the output deviation value, and shorten the cycle lengths of the day-ahead planning stage and the intraday scheduling stage.

[0140] When the output margin is insufficient, it may be necessary to start more gas turbine units to make up for the output deviation.

[0141] At this time, for the day-ahead planning stage, adjust the opening and closing scheduling cost parameters to optimize the start-stop decision of the gas turbine unit. For example, reduce the start-stop cost coefficient of the newly started gas turbine unit to start more gas turbine units when necessary to meet the power demand. At the same time, considering the operation efficiency and cost differences of different gas turbine units, adjust the fuel cost coefficients of each gas turbine unit accordingly to keep the overall power generation cost within a controllable range.

[0142] At the same time, relax the output limits of some gas turbine units and allow the gas turbine units to operate within a larger power range on the premise of ensuring equipment safety. At the same time, also appropriately adjust the constraint conditions such as the minimum continuous operation time and the minimum continuous shutdown time of the gas turbine unit to adapt to the power supply demand in emergency situations.

[0143] For the intraday scheduling stage, similar to when the output margin exceeds the output deviation value, adjust the output corresponding constraint conditions in the intraday scheduling stage according to the output deviation value.

[0144] At the same time, shortening the cycle lengths of the day-ahead planning stage and the intra-day scheduling stage. For the day-ahead planning stage, the original cycle length in hours can be appropriately shortened to more frequently evaluate and adjust the start-stop plans and power generation power distributions of the units. In the intra-day scheduling stage, the cycle is further shortened, such as from 10 minutes to 5 minutes, to achieve a rapid response to the real-time changes in the power system. By closely monitoring and timely adjusting, strive to meet the power demand and ensure the stable operation of the power system.

[0145] However, in addition to the above situations, there is also a situation where the output deviation value causes the gas turbine unit to generate more electricity due to inaccurate prediction of new energy output. For such a situation, similarly, according to the comparison between the output deviation value and the output margin, the day-ahead planning stage or the intra-day planning stage can be adjusted for processing. Here, the output margin is the amount by which the gas turbine unit can reduce its output.

[0146] Adjusting the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbine unit according to the gas turbine unit operation information in the day-ahead planning stage includes:

[0147] Determining the operation duration of the gas turbine units currently in the operating state, and setting the earliest opening and closing control times of each gas turbine unit according to the corresponding minimum continuous operation time and minimum shutdown time;

[0148] Adding the earliest opening and closing control times of each gas turbine unit to the corresponding constraint conditions of the gas turbine unit.

[0149] The operation duration of the gas turbine units currently in the operating state is directly related to whether the gas turbine units can perform start-stop operations according to the requirements of the minimum continuous operation time and the minimum shutdown time. Suppose the minimum continuous operation time of a certain gas turbine unit is 4 hours. If the unit has been operating for 3 hours, then within the next 1 hour, it cannot be shut down, that is, the earliest shutdown time is at least 1 hour later. Similarly, if a gas turbine unit has just been shut down and its minimum shutdown time is 2 hours, then it cannot be started again within 2 hours, and the earliest start time is 2 hours after shutdown. Through the above calculations, the earliest startable and earliest stoppable time points of each gas turbine unit in the current operating state can be determined.

[0150] Taking the earliest startable and earliest stoppable time points of each gas turbine unit determined in the current operating state as constraint conditions and adding them to the day-ahead planning stage to avoid unreasonable scheduling that violates the requirements of the minimum continuous operation time and the minimum shutdown time, ensure the stable operation of the gas turbine units and the equipment life, and at the same time improve the reliability and economy of the entire power system.

[0151] The re-solving of the adjusted multi-stage adjustment model further includes:

[0152] Each time reaching the feedback node, the prediction models for load conditions and new energy power generation are corrected according to the fluctuation information of the prediction results;

[0153] Predict the load conditions and new energy power generation again according to the corrected prediction models;

[0154] Based on the newly obtained prediction results, solve the adjusted multi-stage adjustment model.

[0155] Considering that the training of the initial prediction model is based on historical data, there will inevitably be deviations in the prediction results. At the feedback node, the fluctuation value of the prediction results can be intuitively obtained. At the feedback node, the latest obtained actual load data, new energy power generation data, and corresponding meteorological data, etc. are used as new training samples to retrain the prediction model, adjust the weights and parameters of the model, improve the prediction accuracy of the model for future load and new energy power generation, so as to further improve the accuracy of the scheduling decision of the multi-stage scheduling model.

[0156] The embodiments described above are only a preferred solution of the present invention, and do not impose any form of limitation on the present invention. There are other variants and modifications without exceeding the technical solutions recorded in the claims.

Claims

1. A gas turbine unit scheduling method based on multi-stage dual dynamic programming, characterized in that Including: Predict the load conditions and new energy power generation in the power supply area corresponding to the gas turbine unit; Taking cost minimization as the optimization goal, combined with the prediction results, construct a multi-stage scheduling model for the day-ahead unit commitment planning and the in-day unit output scheduling in the power supply area; Solve the multi-stage scheduling model based on dual dynamic programming to obtain the scheduling strategy of the gas turbine unit; Set feedback nodes according to the day-ahead planning period and the in-day scheduling period in the multi-stage scheduling model; When reaching the feedback node, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction results within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model; Re-solve the adjusted multi-stage adjustment model to update the scheduling strategy of the gas turbine unit until the scheduling of the gas turbine unit for the day is completed; The setting of the feedback node according to the day-ahead planning period and the in-day scheduling period in the multi-stage scheduling model includes: Set the feedback node according to the day-ahead planning information, in-day scheduling information, regulation duration, and time coincidence degree, and dynamically adjust the feedback node as the scheduling strategy is updated; Determine the current day-ahead planning information and in-day scheduling information according to the scheduling strategy; Taking the start time of each cycle as an alternative feedback node, calculate the regulation operation information of the gas turbine unit at each alternative feedback node according to the corresponding day-ahead planning information and in-day scheduling information; Based on the regulation operation information, calculate the regulation cost and regulation duration at the alternative feedback nodes corresponding to the day-ahead planning and the in-day scheduling respectively; Calculate the time coincidence degree between the alternative feedback nodes corresponding to the day-ahead planning and the in-day scheduling, and set the feedback node in combination with the regulation cost and regulation duration; After each update of the scheduling strategy, reset the remaining feedback nodes according to the day-ahead planning period and the in-day scheduling period in the adjusted multi-stage scheduling model; The solving of the multi-stage scheduling model based on dual dynamic programming to obtain the scheduling strategy of the gas turbine unit includes: For the day-ahead planning stage, define the stages of the corresponding dual dynamic programming according to the day-ahead planning period, and set the corresponding state variables; Set the corresponding dual variables according to the constraint conditions corresponding to the day-ahead planning stage, and construct the corresponding dual objective function; Starting from the last stage, calculate the optimal scheduling decision for this stage according to the values of the current corresponding state variables and dual variables, and combine the optimal decision result of this stage with the initial state of the next stage, and move forward to the previous stage until the optimal unit commitment strategy for the day-ahead planning stage is obtained; For the in-day scheduling stage, define the stages of the corresponding dual dynamic programming according to the in-day planning period, and set the corresponding state variables in combination with the optimal unit commitment strategy of the day-ahead planning stage; Set the corresponding dual variables according to the constraint conditions corresponding to the in-day scheduling stage, and construct the corresponding dual objective function; Starting from the last stage, calculate the optimal scheduling decision for this stage according to the values of the current corresponding state variables and dual variables, and combine the optimal decision result of this stage with the initial state of the next stage, and move forward to the previous stage until the optimal unit output allocation strategy for the in-day scheduling stage is obtained; When arriving at the feedback node, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction result within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model, including: Identify the adjustment requirements according to the operation information of the gas turbine unit and the fluctuation information of the prediction result; According to the identification result of the adjustment requirements, combine the operation information of the gas turbine unit and the fluctuation information of the prediction result to adjust the model parameters and cycle length of the multi-stage scheduling model; Based on the output margin and output deviation value, identify the adjustment requirements for the day-ahead planning and intra-day scheduling; When the output margin exceeds the output deviation value, maintain the model parameters and cycle length of the day-ahead planning stage, and adjust the output corresponding constraint conditions of the intra-day scheduling stage according to the output deviation value and the operation information of the gas turbine unit, and shorten the cycle length of the intra-day scheduling stage; When the output margin does not exceed the output deviation value, adjust the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbine unit in the day-ahead planning stage according to the operation information of the gas turbine unit, adjust the output corresponding constraint conditions of the intra-day scheduling stage according to the output deviation value, and shorten the cycle lengths of the day-ahead planning stage and the intra-day scheduling stage.

2. The gas turbine unit scheduling method based on multi-stage dual dynamic programming according to claim 1, characterized in that When arriving at the feedback node, obtain the operation information of the gas turbine unit and the fluctuation information of the prediction result within the corresponding time period, and correspondingly adjust the model parameters and cycle length of the multi-stage scheduling model, including: Obtain the operation information of the gas turbine unit, the actual load situation, and the actual power generation information of new energy within the time period from the previous feedback node to the current feedback node under the current scheduling strategy, and determine the fluctuation information of the prediction result.

3. The gas turbine unit scheduling method based on multi-stage dual dynamic programming according to claim 2, wherein The identification of the adjustment requirements according to the operation information of the gas turbine unit and the fluctuation information of the prediction result includes: Obtain the gas turbine units in the current operating state and the corresponding output, and calculate the output deviation value of the gas turbine units in combination with the fluctuation information of the prediction result; Calculate the output margin of the gas turbine units in the current operating state according to the operation information of the gas turbine units.

4. The gas unit scheduling method based on multi-stage dual dynamic programming according to claim 3, wherein The adjustment of the opening and closing scheduling cost parameters and the corresponding constraint conditions of the gas turbine unit in the day-ahead planning stage according to the operation information of the gas turbine unit includes: Determine the operating duration of the gas turbine units in the current operating state, and set the earliest opening and closing control time of each gas turbine unit according to the corresponding minimum continuous operating time and minimum shutdown time; Add the earliest opening and closing control time of each gas turbine unit to the corresponding constraint conditions of the gas turbine unit.

5. The gas turbine unit scheduling method based on multi-stage dual dynamic programming according to claim 2, characterized in that The re-solving of the adjusted multi-stage adjustment model further includes: Every time arriving at a feedback node, correct the prediction model for predicting the load situation and new energy power generation according to the fluctuation information of the prediction result; Predict the load situation and new energy power generation again according to the corrected prediction model; Based on the newly obtained prediction result, solve the adjusted multi-stage adjustment model.

Images